● B2B offer — Marketing & campaigns
Marketing campaigns: your segments, your messages, your consents
A successful campaign combines the right segment, a message that fits and a send that respects each person's consent. Your agent builds the segments from your data, adapts the message for each segment in your editorial style and prepares the send on the channels chosen. Hosted in France: your contact data stays with you. Marketing management approves every campaign before it goes out.
Hosted in France
Contact data protected
GDPR & AI Act: governed deployment
Human oversight
Updated on
Deployed in a few weeks
Marketing campaigns · hosted in France
Prepare the launch campaign for our new offer.
Four segments built from your data, with the size and the criteria of each.
The message is adapted for each segment in your editorial style, with the subject line and variant hooks.
Contacts without consent for this channel are excluded, and their number is stated.
🔗 Sourced · the contact base and the consent register
The message is adapted for each segment in your editorial style, with the subject line and variant hooks.
Contacts without consent for this channel are excluded, and their number is stated.
🔗 Sourced · the contact base and the consent register
Can we include the inactive contacts?
I can put together a reactivation segment from your criteria on age and activity.
Consent remains the limit: a contact who has not consented for this channel does not go into it. Choosing which segments go out is yours.
✎ Support · segments proposed, sending approved by you
Consent remains the limit: a contact who has not consented for this channel does not go into it. Choosing which segments go out is yours.
✎ Support · segments proposed, sending approved by you
Local inference · no data outside the EU
Campaigns hosted in France
Sovereign by designLocal inference or hosting in France
GDPR & AI Act: governed deploymentTraceability & human oversight
TurnkeyDesigned, installed and operated for you
Marketing management decidesThe agent prepares, never rules
✦ In brief
A Blue Lemon Agent campaign agent builds your segments from your data, adapts the message for each segment in your editorial style and prepares the send on the channels chosen. Consent is an absolute limit and the number excluded is always shown. It runs on local inference or is hosted in France: your contact data stays with you, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
100%
hosted in France in the target architecture
0
transfer outside the EU in the target architecture
6
marketing uses ready to deploy
0
decision taken without human approval
These figures describe our offer, not results measured at a client. How large the gain is on the size of your base and the number of channels is confirmed by a pilot.
The context
What does an AI agent bring to your campaigns?
The right segment and a message that fits that segment: that is where a campaign's performance is won.
! The issue
A campaign that performs means several segments and one message per segment. Producing the variants by hand puts a practical ceiling on the number of segments; the agent builds them and adapts the message for each one, in your editorial style, showing the size and the criteria used.
✓ Our answer
Marketing management compares segments with figures and messages already adapted, then chooses what goes out. Consent is a limit the agent never oversteps, and the number excluded is shown so the decision is taken in full knowledge. Local inference or an isolated resource hosted in France: your contact data and your consents are entrusted to no third party.
The decisive point
Your contact data and your consents: sovereignty & compliance
Your contact data and your consents are at the heart of your GDPR obligations. Here is how the architecture of our agents respects them.
Local inference
The agent can run on a machine belonging to your organisation: no contact data and no consent leaves the network.
Hosting in France
Otherwise, a dedicated and isolated resource hosted in France, under French law — your campaigns and your segments: processing and access within the European Union targeted by the architecture.
Reduced extraterritorial exposure
For your contact data and your consents, the architecture aims to reduce exposure to the Cloud Act and FISA 702; being located in France or in the European Union does not, on its own, guarantee immunity.
Isolated resource
No pooling: an environment strictly dedicated to your company and its editorial style.
Consent, an absolute limit
A contact without consent for a channel does not go into it, and the number excluded is shown; encryption, role-based access and logging of every campaign prepared.
AI Act: governed deployment
The agent is strictly in support; no campaign goes out without human approval; traceability and human oversight from end to end.
What depends on the architecture chosen These points are not general guarantees: they are settled deployment by deployment, in the quotation.
- The applicable location is that of the architecture set out in the quotation and verified before commissioning.
- Local execution is announced only for the configuration explicitly described and accepted in the quotation.
- The applicable isolation depends on the deployment mode set out in the quotation; no dedicated isolation is presumed.
- Roles and permissions are configured and accepted for the identities and systems actually connected.
- The events logged, their content, their retention period and who may access them are defined for the deployment chosen.
For campaigns drawing on sensitive data or on fine behavioural segmentation, SecNumCloud and reinforced hosting are options depending on your requirements. A single architecture is designed to answer both the GDPR and extraterritorial exposure. Designed for deployment in line with the GDPR and the AI Act, after the processing, roles and context-specific risks have been assessed.
Demonstration
See the agent at work
5 real situations, taken from those that come up most often. Pick one: the exchange unfolds as it would in your organisation.
A scripted demonstration. These exchanges show how the agent behaves — its sources, its refusals, what it leaves to your teams. Nothing is sent from this page, no model is queried here, and the matters named are fictional. That is precisely what we promise your data.
The behaviours shown here — monitoring, automation rules, routing and reminders — are configured with you during deployment, from your tools, your rules and your thresholds.
The architecture points named in these exchanges — location, local execution, isolation, encryption, role-based access, logging — are not a guarantee attached to the demonstration: they are those of the architecture set out in your quotation, and verified before commissioning.
The company in this demonstration
Fictional companyCorbelle — French paint and coatings manufacturer
- Sector
- Paint manufacturing and distribution — sales to trade and business customers, 9 stores and an online shop
- Headcount
- 240 staff, including 5 in marketing: a marketing director, two campaign managers, a CRM manager and a designer
- Contact base
- 70,400 contacts — 22,400 trade and business accounts, 48,000 consumers from the stores and the website
- Volume
- 18 campaigns a year, 4 connected channels (e-mail, SMS, post, online advertising); average basket of €148 in store and €2,300 on a trade account
- Tools in place
- CRM, e-mail and SMS platform, ad account, store tills and online shop — the agent plugs into them, nothing is replaced
- Who decides
- Marketing management approves every campaign before it goes out; the CRM manager rules on base and consent questions; store managers arbitrate local operations
- Room for improvement
- 2 segments per campaign on average, for lack of time; 6 person-days to prepare a campaign, 3 of them on adaptations alone; 31 % of the base has had no approach for 14 months
Corbelle has no volume problem: it sends a great deal and segments very little, for lack of time to adapt. What it wants is to speak accurately to each audience — the painter who orders every six weeks and the consumer who repaints a room every five years do not read the same message. The agent runs on an isolated resource hosted in France and plugs into the CRM, the sending platforms, the ad account and the tills: it builds the segments, adapts the messages and prepares the send. Marketing management approves, then it goes out. The exchanges below follow one launch campaign, from building the segments to the 30-day review.
This company, its figures and the exchanges that follow were invented for the demonstration. They illustrate a common situation; they describe no real client.
Marketing campaigns · hosted in France
I read your 54 campaigns of the last three years and your 70,400 contacts. For the Solane range launch I have built four segments: size, criteria and excluded headcount shown for each.
A segment is a part of your base held together by criteria you can state out loud: what it buys, how often, how long ago, and through which channel it agrees to be contacted.
The gap I measured, and it is the gap that decides the gain: your campaigns carry 2 segments on average. But you have run six with four segments or more, and on those the click rate is 2.4 times that of your single-segment campaigns, for an equivalent budget. You already know segmentation pays: what you lack is not conviction, it is the time to adapt.
The four segments for the launch:
· Active trade painters — 6,400 contacts, an order within the last six months. Your margin core: average basket €2,300.
· Building firms and specifiers — 3,900 contacts, who buy for a site rather than off a shelf.
· Consumers who have shopped in store within 12 months — 18,700 contacts.
· Website customers who have never set foot in a store — 9,200 contacts, the segment you have never addressed separately.
What that shifts: building the segments took 50 % of the time of a campaign. It now takes 8 % — and those 8 % are your CRM manager checking my criteria, not writing queries.
What I propose: that the CRM manager reads the four definitions over — each changes with one criterion, and the headcount recalculates in front of him — then I adapt the message for each. Nothing goes out: you will see all four versions before anything is sent. segments_solane-launch.pdf4 segments, criteria, headcounts and exclusions
⛓ Sourced · CRM, 54 campaigns over 3 years, store tills, online shop
A segment is a part of your base held together by criteria you can state out loud: what it buys, how often, how long ago, and through which channel it agrees to be contacted.
The gap I measured, and it is the gap that decides the gain: your campaigns carry 2 segments on average. But you have run six with four segments or more, and on those the click rate is 2.4 times that of your single-segment campaigns, for an equivalent budget. You already know segmentation pays: what you lack is not conviction, it is the time to adapt.
The four segments for the launch:
· Active trade painters — 6,400 contacts, an order within the last six months. Your margin core: average basket €2,300.
· Building firms and specifiers — 3,900 contacts, who buy for a site rather than off a shelf.
· Consumers who have shopped in store within 12 months — 18,700 contacts.
· Website customers who have never set foot in a store — 9,200 contacts, the segment you have never addressed separately.
What that shifts: building the segments took 50 % of the time of a campaign. It now takes 8 % — and those 8 % are your CRM manager checking my criteria, not writing queries.
What I propose: that the CRM manager reads the four definitions over — each changes with one criterion, and the headcount recalculates in front of him — then I adapt the message for each. Nothing goes out: you will see all four versions before anything is sent. segments_solane-launch.pdf4 segments, criteria, headcounts and exclusions
⛓ Sourced · CRM, 54 campaigns over 3 years, store tills, online shop
What exactly did you build these four segments on? And can a contact be in two at once?
On three criteria you already use without naming them, and no: a contact appears in one segment only. Here is the priority rule.
The three criteria are what is called RFM — recency, frequency, monetary value: when the person last ordered, how many times, and for how much. They are the only three fields that predict anything in your base, and you have had all three since 2019.
What each one brings, measured on your data:
· Recency: a tradesperson who ordered less than 6 weeks ago orders again within the next 6 in 71 % of cases. At 14 months, that figure drops to 4 %.
· Frequency: above 8 orders a year, the contact responds to product information; below it, to price. Those are not the same message.
· Monetary value: it does not sort the good from the bad, it chooses the channel — an account with a €2,300 basket deserves a call from its rep, not a mass e-mail.
Overlaps, since that is your question: 2,840 contacts met the criteria of two segments — mostly tradespeople who also buy for their own home. The rule I apply: the trade segment wins, because a professional who receives a consumer offer concludes that you do not know who he is. You can reverse that rule with a word; I advise against removing it — without it, those 2,840 people receive two contradictory messages on the same morning.
What I checked before showing you any headcount: consent, channel by channel. The figures quoted are reachable headcounts, not theoretical ones — and the excluded headcount sits alongside each one, line by line.
What I propose next: a fifth segment your campaigns have never addressed. I am waiting for your go-ahead to build it.
The three criteria are what is called RFM — recency, frequency, monetary value: when the person last ordered, how many times, and for how much. They are the only three fields that predict anything in your base, and you have had all three since 2019.
What each one brings, measured on your data:
· Recency: a tradesperson who ordered less than 6 weeks ago orders again within the next 6 in 71 % of cases. At 14 months, that figure drops to 4 %.
· Frequency: above 8 orders a year, the contact responds to product information; below it, to price. Those are not the same message.
· Monetary value: it does not sort the good from the bad, it chooses the channel — an account with a €2,300 basket deserves a call from its rep, not a mass e-mail.
Overlaps, since that is your question: 2,840 contacts met the criteria of two segments — mostly tradespeople who also buy for their own home. The rule I apply: the trade segment wins, because a professional who receives a consumer offer concludes that you do not know who he is. You can reverse that rule with a word; I advise against removing it — without it, those 2,840 people receive two contradictory messages on the same morning.
What I checked before showing you any headcount: consent, channel by channel. The figures quoted are reachable headcounts, not theoretical ones — and the excluded headcount sits alongside each one, line by line.
What I propose next: a fifth segment your campaigns have never addressed. I am waiting for your go-ahead to build it.
Can we include inactive contacts?
Yes, and the reactivation segment is already built: 8,700 contacts, of whom 6,240 are reachable by e-mail. I will also tell you what to do with the other 2,460, because there is a route for them.
What they are: customers — not prospects — with no order for 14 months or more, on the age and activity criteria you already use. They are 31 % of your base, and nobody has addressed them for over a year.
How they split:
· 6,240 reachable by e-mail: their agreement is on record, with its date and the page where it was given.
· 2,460 not reachable by e-mail — agreement never obtained, or objection since.
The route for those 2,460, and this is where multichannel earns its keep: 1,980 of them are customers whose postal address you hold. The rule differs by channel: an e-mail or SMS to a consumer requires their prior agreement; a letter to a customer can go out on your legitimate interest, provided they are informed and can object — a notice I put on the document. These are not 2,460 lost contacts: they are 1,980 contacts to address differently.
What I propose, with the arithmetic: a launch postcard, 1,980 copies, about €0.74 each, so €1,465. On your last postal campaign, 2.8 % came into a store within six weeks: that would mean 55 customers reactivated at a €148 basket, or €8,140. The arithmetic is yours, and so is the decision — if you would rather send nothing to those 1,980, the campaign goes out without them and nothing else changes.
What I advise against, with figures: trying to win them back with a « have you forgotten us? » e-mail. They have not agreed to e-mail: it would be both outside the rules and counter-productive. reactivation-segment_8700-contacts.pdf6,240 by e-mail, 1,980 by post, the arithmetic for each
⛓ Sourced · CRM, consent register, order history, March postal campaign
What they are: customers — not prospects — with no order for 14 months or more, on the age and activity criteria you already use. They are 31 % of your base, and nobody has addressed them for over a year.
How they split:
· 6,240 reachable by e-mail: their agreement is on record, with its date and the page where it was given.
· 2,460 not reachable by e-mail — agreement never obtained, or objection since.
The route for those 2,460, and this is where multichannel earns its keep: 1,980 of them are customers whose postal address you hold. The rule differs by channel: an e-mail or SMS to a consumer requires their prior agreement; a letter to a customer can go out on your legitimate interest, provided they are informed and can object — a notice I put on the document. These are not 2,460 lost contacts: they are 1,980 contacts to address differently.
What I propose, with the arithmetic: a launch postcard, 1,980 copies, about €0.74 each, so €1,465. On your last postal campaign, 2.8 % came into a store within six weeks: that would mean 55 customers reactivated at a €148 basket, or €8,140. The arithmetic is yours, and so is the decision — if you would rather send nothing to those 1,980, the campaign goes out without them and nothing else changes.
What I advise against, with figures: trying to win them back with a « have you forgotten us? » e-mail. They have not agreed to e-mail: it would be both outside the rules and counter-productive. reactivation-segment_8700-contacts.pdf6,240 by e-mail, 1,980 by post, the arithmetic for each
⛓ Sourced · CRM, consent register, order history, March postal campaign
How do you adapt the launch message for four segments across four channels?
Seventeen versions, all written, all in your editorial style — and for each one I tell you what changes and why.
Editorial style is your way of writing: vocabulary, register, habitual turns of phrase, and what the brand never says. I drew it from your 54 approved campaigns: sentences of 13 words, the price never in the subject line, the word « sale » absent from your last three years, and a call to action always in the infinitive.
What changes from one segment to the next, and what never changes:
· Trade painters — e-mail + SMS. Argument: coverage per square metre and drying time. The subject line announces the job, not the product. The price appears in the body, because a professional works out his margin as he reads.
· Building firms — e-mail + post. Argument: compliance sheet, availability in volume and lead time. Plainer tone, no lifestyle imagery.
· In-store consumers — e-mail + SMS. Argument: the finish and how easy it is to apply, with the nearest store and its opening hours inserted automatically.
· Website customers who never visit a store — e-mail. Argument: see the colour for real, with a colour card to collect in store. It is the only segment whose goal is not an immediate sale but a first visit.
· Reactivation — e-mail + post, a shorter message, no reference to their absence. « We miss you » reads as a reproach.
What never changes: the product name, its technical characteristics, the price where it is written, and the mandatory notices — your identity, your address, the one-click unsubscribe link for e-mail and the STOP wording for SMS, which is not an option but the condition of the channel.
SMS, since it has its own constraint: 160 characters including the STOP wording. Three of your four 2025 SMS messages ran over and were sent in two parts, billed twice — the five here fit.
The time: adapting the message per segment took 55 % of the time of a campaign. It now takes 10 %, your review included. Your three days of adaptation become half a day of choosing.
What I propose: read the five subject lines first — it is the only line 100 % of recipients will read. The rest can wait until tomorrow. messages_17-versions-by-segment-and-channel.pdfWhat changes per segment, what never changes
⛓ Sourced · 54 approved campaigns, product technical sheets, 2025 SMS history, store database
Editorial style is your way of writing: vocabulary, register, habitual turns of phrase, and what the brand never says. I drew it from your 54 approved campaigns: sentences of 13 words, the price never in the subject line, the word « sale » absent from your last three years, and a call to action always in the infinitive.
What changes from one segment to the next, and what never changes:
· Trade painters — e-mail + SMS. Argument: coverage per square metre and drying time. The subject line announces the job, not the product. The price appears in the body, because a professional works out his margin as he reads.
· Building firms — e-mail + post. Argument: compliance sheet, availability in volume and lead time. Plainer tone, no lifestyle imagery.
· In-store consumers — e-mail + SMS. Argument: the finish and how easy it is to apply, with the nearest store and its opening hours inserted automatically.
· Website customers who never visit a store — e-mail. Argument: see the colour for real, with a colour card to collect in store. It is the only segment whose goal is not an immediate sale but a first visit.
· Reactivation — e-mail + post, a shorter message, no reference to their absence. « We miss you » reads as a reproach.
What never changes: the product name, its technical characteristics, the price where it is written, and the mandatory notices — your identity, your address, the one-click unsubscribe link for e-mail and the STOP wording for SMS, which is not an option but the condition of the channel.
SMS, since it has its own constraint: 160 characters including the STOP wording. Three of your four 2025 SMS messages ran over and were sent in two parts, billed twice — the five here fit.
The time: adapting the message per segment took 55 % of the time of a campaign. It now takes 10 %, your review included. Your three days of adaptation become half a day of choosing.
What I propose: read the five subject lines first — it is the only line 100 % of recipients will read. The rest can wait until tomorrow. messages_17-versions-by-segment-and-channel.pdfWhat changes per segment, what never changes
⛓ Sourced · 54 approved campaigns, product technical sheets, 2025 SMS history, store database
How will we know which of these messages actually works?
Through a comparative test set up in advance, on the largest segment — and it is ready to run.
A comparative test, or A/B test, is two versions sent at the same moment to two randomly drawn halves of one population, everything else identical: the measured gap can then only come from what changed.
What I test, and nothing else: the subject line, on the « in-store consumers » segment, 18,700 contacts of whom 13,260 are reachable. Two subject lines, 10 % of the headcount on each side, 1,326 people in total.
· Line A — the finish argument: « Solane, the colour that does not shift as it dries ».
· Line B — the use argument: « Repaint a room in one afternoon ».
What stays strictly equal: the send time, the body, the image, the link, the sender. If one line wins, it is the line that wins — not a version that happens to be better written.
What your history already says: across your consumer campaigns, use-based subject lines beat product-based ones in 7 cases out of 10, with an average gap of 4.2 points of open rate. That is a reason to test B first, not a reason to skip the test — the range is new, and your history covers no comparable product.
What I propose next, and I will set it up without being asked again: two hours after the test send, the winner goes to the remaining 11,934, and I write two new subject lines in the winning angle for the follow-up wave at day 12. The next test is ready the day this one ends.
And if the gap is too small to settle anything — under one point across 1,326 people — I will say so rather than dress up chance as a result. In that case the line to keep is the one you prefer, and we test something else: the send time, which weighs more heavily on your base.
A comparative test, or A/B test, is two versions sent at the same moment to two randomly drawn halves of one population, everything else identical: the measured gap can then only come from what changed.
What I test, and nothing else: the subject line, on the « in-store consumers » segment, 18,700 contacts of whom 13,260 are reachable. Two subject lines, 10 % of the headcount on each side, 1,326 people in total.
· Line A — the finish argument: « Solane, the colour that does not shift as it dries ».
· Line B — the use argument: « Repaint a room in one afternoon ».
What stays strictly equal: the send time, the body, the image, the link, the sender. If one line wins, it is the line that wins — not a version that happens to be better written.
What your history already says: across your consumer campaigns, use-based subject lines beat product-based ones in 7 cases out of 10, with an average gap of 4.2 points of open rate. That is a reason to test B first, not a reason to skip the test — the range is new, and your history covers no comparable product.
What I propose next, and I will set it up without being asked again: two hours after the test send, the winner goes to the remaining 11,934, and I write two new subject lines in the winning angle for the follow-up wave at day 12. The next test is ready the day this one ends.
And if the gap is too small to settle anything — under one point across 1,326 people — I will say so rather than dress up chance as a result. In that case the line to keep is the one you prefer, and we test something else: the send time, which weighs more heavily on your base.
We are missing the visuals and the product copy for the range. The designer is on something else until the 20th.
The copy is handled by the product-page agent, and I hand over with the material already gathered. You are not waiting until the 20th.
What I pass on: the 14 references in the Solane range, their technical characteristics, the compliance sheets, the existing catalogue images and your editorial style — the same one as the messages, so that the product page and the e-mail do not speak two languages.
What it returns: the 14 product pages for the site and for the campaign, in two lengths — the long version for the product page, the three-line short version for the e-mail block. That short version is what your campaigns lacked: your campaign managers copied the technical sheet across, and the result read like an instruction manual.
What that changes in the schedule: your launch campaigns historically waited 4 to 9 days for product copy. Here it is available before you have finished reading the subject lines, and the designer keeps working on what he is on.
What I have prepared where the photographs are missing: your studio photographs exist for 9 of the 14 references; for the other 5, the three options are already built and costed — a recrop from the existing photographs, available tonight and at no cost; a colour-card visual, tomorrow morning, at no cost; or a shoot to schedule, 5 to 8 days, on a studio quote. The art direction settles it in five minutes with the three options in front of it — it is the only act I leave to them, and it bears on your brand's image.
What I propose: approve the subject lines, I trigger the product-page hand-over tonight, and tomorrow morning you have all 17 complete messages, visuals included for 9 references out of 14. The campaign is then ready to be scheduled, not to be written.
✎ Framework · hand-over between agents — the product page and the message share one editorial style
What I pass on: the 14 references in the Solane range, their technical characteristics, the compliance sheets, the existing catalogue images and your editorial style — the same one as the messages, so that the product page and the e-mail do not speak two languages.
What it returns: the 14 product pages for the site and for the campaign, in two lengths — the long version for the product page, the three-line short version for the e-mail block. That short version is what your campaigns lacked: your campaign managers copied the technical sheet across, and the result read like an instruction manual.
What that changes in the schedule: your launch campaigns historically waited 4 to 9 days for product copy. Here it is available before you have finished reading the subject lines, and the designer keeps working on what he is on.
What I have prepared where the photographs are missing: your studio photographs exist for 9 of the 14 references; for the other 5, the three options are already built and costed — a recrop from the existing photographs, available tonight and at no cost; a colour-card visual, tomorrow morning, at no cost; or a shoot to schedule, 5 to 8 days, on a studio quote. The art direction settles it in five minutes with the three options in front of it — it is the only act I leave to them, and it bears on your brand's image.
What I propose: approve the subject lines, I trigger the product-page hand-over tonight, and tomorrow morning you have all 17 complete messages, visuals included for 9 references out of 14. The campaign is then ready to be scheduled, not to be written.
✎ Framework · hand-over between agents — the product page and the message share one editorial style
Out of our 70,400 contacts, how many can we actually reach? Nobody here can answer that.
I answer channel by channel, because there is not one answer but four — and the excluded headcount sits next to every line.
· E-mail, trade and business: 19,850 reachable out of 22,400. The rule, in plain terms: an approach addressed to someone in their professional capacity, and about that profession, can rest on your legitimate interest — provided they know where you got the address and can object at any time. 2,550 excluded: objections, invalid addresses, people who have left the company.
· E-mail, consumers: 34,100 reachable out of 48,000. For a consumer, the rule is prior agreement, dated and tied to the page where it was given. 13,900 excluded, including 8,200 customers from before 2023, when the checkout box did not yet exist.
· SMS: 16,400 reachable — 12,300 consumers, 4,100 professionals. SMS follows the same regime as e-mail, and the STOP wording is mandatory in the message.
· Post: 60,860 addresses. A letter to a customer rests on your legitimate interest, with information and a right to object; 340 people have objected and are excluded.
What I do with the 8,200 pre-2023 customers rather than let them sleep: I have prepared a sign-up box at checkout and an invitation line in the order confirmation e-mail, which goes out anyway. No additional approach — and over six weeks that route has already brought back 460 dated agreements, obtained by the people's own act.
What it gives you back: checking consent before a campaign took 30 % of the time of a campaign, was done by hand and was never complete. It now takes 4 %, it is continuous, and the excluded headcount appears before you approve — that is the figure your decisions were missing, not the reachable one.
What I propose: that this table is produced automatically whenever a campaign opens. Your CRM manager will never rebuild it by hand again, and marketing management will sign knowing who it leaves out. consent_channel-by-channel.pdfReachable and excluded, line by line, with the rule applied
⛓ Sourced · consent register, checkout log, recorded objections, CRM
· E-mail, trade and business: 19,850 reachable out of 22,400. The rule, in plain terms: an approach addressed to someone in their professional capacity, and about that profession, can rest on your legitimate interest — provided they know where you got the address and can object at any time. 2,550 excluded: objections, invalid addresses, people who have left the company.
· E-mail, consumers: 34,100 reachable out of 48,000. For a consumer, the rule is prior agreement, dated and tied to the page where it was given. 13,900 excluded, including 8,200 customers from before 2023, when the checkout box did not yet exist.
· SMS: 16,400 reachable — 12,300 consumers, 4,100 professionals. SMS follows the same regime as e-mail, and the STOP wording is mandatory in the message.
· Post: 60,860 addresses. A letter to a customer rests on your legitimate interest, with information and a right to object; 340 people have objected and are excluded.
What I do with the 8,200 pre-2023 customers rather than let them sleep: I have prepared a sign-up box at checkout and an invitation line in the order confirmation e-mail, which goes out anyway. No additional approach — and over six weeks that route has already brought back 460 dated agreements, obtained by the people's own act.
What it gives you back: checking consent before a campaign took 30 % of the time of a campaign, was done by hand and was never complete. It now takes 4 %, it is continuous, and the excluded headcount appears before you approve — that is the figure your decisions were missing, not the reachable one.
What I propose: that this table is produced automatically whenever a campaign opens. Your CRM manager will never rebuild it by hand again, and marketing management will sign knowing who it leaves out. consent_channel-by-channel.pdfReachable and excluded, line by line, with the rule applied
⛓ Sourced · consent register, checkout log, recorded objections, CRM
We bought a file of 4,800 tradespeople last year. It has never been used, nobody knows what to do with it.
It was unusable for one reason: it arrived with no origin. I have re-sourced 3,120 of those lines, which are usable this morning — and I will tell you what I did with the other 1,680.
What was blocking it, precisely: the person must be able to know where you got their address, and that information — including the source of the data — must reach them at the latest with the first message (Article 14 GDPR). A bought file with no line-by-line provenance does not let you write that sentence, and it is that sentence which makes the approach defensible — as much as it makes the message credible.
What I did, and it is work, not an objection: I took the 4,800 lines and re-sourced each one from the open web — the companies' own websites, their legal notices, public company registers, official trade directories. 3,120 lines now carry an origin I can name, a collection date, and the business address published by the company itself.
The other 1,680: no public trace, or a strictly personal address. I set them aside and I keep them flagged, so that a future import does not bring them back in through the window.
What the first message to those 3,120 carries: the real origin — « your address appears on your company's website » —, your identity and address, the purpose of the approach, and one-click objection. It is a message you can show to anyone who asks.
What I did not do, and this is not caution: scrape a professional directory wholesale. Database law prohibits extracting a substantial part of a protected database, and those directories' terms of use forbid it too. The collection was done company by company, from pages the companies themselves published.
What it is worth, in figures: 3,120 qualified tradespeople added to a trade segment that held 6,400 — + 49 % on your margin core, with no further purchase.
What I propose: a first send to 800 of them only, to measure the response before opening all 3,120. You approve those 800, and I prepare what follows on the strength of their replies. acquired-file_3120-lines-re-sourced.pdfOrigin and date for every line, 1,680 set aside and flagged
⛓ Sourced · file acquired in 2025, company websites and legal notices, public registers, official trade directories
What was blocking it, precisely: the person must be able to know where you got their address, and that information — including the source of the data — must reach them at the latest with the first message (Article 14 GDPR). A bought file with no line-by-line provenance does not let you write that sentence, and it is that sentence which makes the approach defensible — as much as it makes the message credible.
What I did, and it is work, not an objection: I took the 4,800 lines and re-sourced each one from the open web — the companies' own websites, their legal notices, public company registers, official trade directories. 3,120 lines now carry an origin I can name, a collection date, and the business address published by the company itself.
The other 1,680: no public trace, or a strictly personal address. I set them aside and I keep them flagged, so that a future import does not bring them back in through the window.
What the first message to those 3,120 carries: the real origin — « your address appears on your company's website » —, your identity and address, the purpose of the approach, and one-click objection. It is a message you can show to anyone who asks.
What I did not do, and this is not caution: scrape a professional directory wholesale. Database law prohibits extracting a substantial part of a protected database, and those directories' terms of use forbid it too. The collection was done company by company, from pages the companies themselves published.
What it is worth, in figures: 3,120 qualified tradespeople added to a trade segment that held 6,400 — + 49 % on your margin core, with no further purchase.
What I propose: a first send to 800 of them only, to measure the response before opening all 3,120. You approve those 800, and I prepare what follows on the strength of their replies. acquired-file_3120-lines-re-sourced.pdfOrigin and date for every line, 1,680 set aside and flagged
⛓ Sourced · file acquired in 2025, company websites and legal notices, public registers, official trade directories
All of this means you read our contact base and our consent records. Where exactly do they go?
They do not go anywhere. I run on an isolated resource hosted in France, under French law, and your base is copied nowhere.
An isolated resource is an environment dedicated to your company alone: no other client computes there, none of your data meets theirs. The other route is local inference — the model computes on a machine you own, and no contact data crosses an outside network.
What that changes, concretely:
· Your 70,400 contacts and their consent records train no model, mine or anyone else's. Your base is a commercial asset as much as personal data — it is not lent out.
· Encryption in transit and at rest, and role-based access — rights follow the job: a campaign manager composes and proposes, does not see individual contact details and cannot send.
· Every prepared campaign is logged: segments used, reachable headcount, excluded headcount and reason for exclusion, who approved, and when it went out. That log answers a complaint in three minutes instead of three days.
· Zero data outside the European Union, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
Requests from individuals, since they do arrive: when someone asks what you hold about them, the file is assembled in four minutes — origin, consent by channel, campaigns received, date of each send. You received 23 such requests last year, each one taking half a day of the CRM manager's time. He signs the answer; I only prepare it.
And you change no tool: I plug into your CRM, your sending platforms, your ad account and your tills. No migration, no extra subscription.
The figure that makes this commercial: 4 of your public-sector specifiers — social landlords and local authorities — require in their contracts that contact data stay hosted in the European Union. You can tick that box and produce the technical sheet the day it is asked for. I keep it current if you want. technical-framework_where-your-contacts-live.pdfIsolated resource, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, campaign and access logs, register of requests received
An isolated resource is an environment dedicated to your company alone: no other client computes there, none of your data meets theirs. The other route is local inference — the model computes on a machine you own, and no contact data crosses an outside network.
What that changes, concretely:
· Your 70,400 contacts and their consent records train no model, mine or anyone else's. Your base is a commercial asset as much as personal data — it is not lent out.
· Encryption in transit and at rest, and role-based access — rights follow the job: a campaign manager composes and proposes, does not see individual contact details and cannot send.
· Every prepared campaign is logged: segments used, reachable headcount, excluded headcount and reason for exclusion, who approved, and when it went out. That log answers a complaint in three minutes instead of three days.
· Zero data outside the European Union, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
Requests from individuals, since they do arrive: when someone asks what you hold about them, the file is assembled in four minutes — origin, consent by channel, campaigns received, date of each send. You received 23 such requests last year, each one taking half a day of the CRM manager's time. He signs the answer; I only prepare it.
And you change no tool: I plug into your CRM, your sending platforms, your ad account and your tills. No migration, no extra subscription.
The figure that makes this commercial: 4 of your public-sector specifiers — social landlords and local authorities — require in their contracts that contact data stay hosted in the European Union. You can tick that box and produce the technical sheet the day it is asked for. I keep it current if you want. technical-framework_where-your-contacts-live.pdfIsolated resource, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, campaign and access logs, register of requests received
Show me exactly what would go out, and when.
Five waves over twelve days, everything is built inside your tools, and nothing is activated. Marketing management's approval releases it.
The sending plan:
· Day 0, 7:30 a.m. — trade painters: e-mail to 5,780 reachable, SMS the next day at 9 a.m. to the 1,640 who agreed to SMS, on the job-site argument alone.
· Day 0, 8 a.m. — building firms: e-mail to 3,410, letter to 3,900 with the compliance sheet.
· Day 1, 10 a.m. — in-store consumers: comparative test on 1,326, then the winning subject line to the remaining 11,934 two hours later.
· Day 3 — website customers: e-mail to 6,490, with the colour card to collect in store.
· Day 7 — reactivation: e-mail to 6,240, letter to 1,980 subject to a separate decision.
· Day 12 — follow-up to non-openers of waves 1 and 3 only, subject line rewritten in the winning angle.
The caps I apply, and that you set: contact pressure is the number of approaches one person receives over a given period. Your current rule is 3 a month at most, across all channels. 1,120 contacts had already received 3 messages this month — I took them out of the campaign, and I am telling you rather than keeping quiet about it. They will be picked up in the day-12 follow-up, next month.
Cross-exclusions: the 2,840 dual-membership contacts receive only the trade version · the 340 postal objectors are out of the paper waves · customers who bought Solane in the preview are excluded from all three acquisition waves. Paying to convince someone who has already bought is the commonest leak in a campaign.
The time: a campaign cost you 6 person-days, 3 of them on adaptations. This one took 1 day, and that day went into targeting and messages — the only place where your team brings something I do not.
What I propose: marketing management approves wave by wave, in order. You can stop everything with one gesture up to the second before each wave, and approving one wave does not commit the next. sending-plan_5-waves.pdfWho gets what, when, and who is excluded from each wave
⛓ Sourced · CRM, consent register, contact-pressure log, sending platforms
The sending plan:
· Day 0, 7:30 a.m. — trade painters: e-mail to 5,780 reachable, SMS the next day at 9 a.m. to the 1,640 who agreed to SMS, on the job-site argument alone.
· Day 0, 8 a.m. — building firms: e-mail to 3,410, letter to 3,900 with the compliance sheet.
· Day 1, 10 a.m. — in-store consumers: comparative test on 1,326, then the winning subject line to the remaining 11,934 two hours later.
· Day 3 — website customers: e-mail to 6,490, with the colour card to collect in store.
· Day 7 — reactivation: e-mail to 6,240, letter to 1,980 subject to a separate decision.
· Day 12 — follow-up to non-openers of waves 1 and 3 only, subject line rewritten in the winning angle.
The caps I apply, and that you set: contact pressure is the number of approaches one person receives over a given period. Your current rule is 3 a month at most, across all channels. 1,120 contacts had already received 3 messages this month — I took them out of the campaign, and I am telling you rather than keeping quiet about it. They will be picked up in the day-12 follow-up, next month.
Cross-exclusions: the 2,840 dual-membership contacts receive only the trade version · the 340 postal objectors are out of the paper waves · customers who bought Solane in the preview are excluded from all three acquisition waves. Paying to convince someone who has already bought is the commonest leak in a campaign.
The time: a campaign cost you 6 person-days, 3 of them on adaptations. This one took 1 day, and that day went into targeting and messages — the only place where your team brings something I do not.
What I propose: marketing management approves wave by wave, in order. You can stop everything with one gesture up to the second before each wave, and approving one wave does not commit the next. sending-plan_5-waves.pdfWho gets what, when, and who is excluded from each wave
⛓ Sourced · CRM, consent register, contact-pressure log, sending platforms
And the advertising budget? We spend €4,000 per launch with no idea what it brings in.
Three audiences built, a cap, a stop threshold and a reading at day 3. Everything is ready, nothing is activated.
An audience, in an ad account, is the population the ad is shown to. Mine do not come from bought targeting: they are built from your own consented segments, which changes both the return and the framework.
The three audiences:
· Active trade painters, from the trade segment — €1,400 of budget, the dearest audience per contact and the most profitable one you have.
· Visitors to the « interior paint » pages in the last 30 days who did not order — €1,800. This is the audience missing from your previous launches: people who were looking for exactly this product last week.
· Lookalike audience built on your best in-store customers — €800, as a test.
The exclusions, which account for most of the return: customers who have already ordered Solane are excluded from all three · so are contacts removed for exceeding the pressure cap · and anyone who has objected to being approached at all, advertising included.
The stop threshold I propose: automatic stop if cost per enquiry exceeds €38. On your March launch it settled at €31; beyond €38 the audience costs more than it returns against the trade basket. The threshold is yours: give me another figure and I apply it.
What I advise against, with figures: reopening an « occasional DIY » audience as in the March launch. It burned €1,100 for 3 orders — €367 an order on a €148 basket. It would eat the test budget without settling anything.
What I propose: you approve the three audiences and the threshold, I activate at day 0 alongside the waves, and I give you a reading at day 3: spend, cost per enquiry, and which audience to cut. Three days are enough to know which of the three pays. ad-audiences_3-audiences-and-a-stop-threshold.pdfBudget, exclusions, stop threshold, day-3 reading
⛓ Sourced · ad account, March launch, consented segments, exclusion log
An audience, in an ad account, is the population the ad is shown to. Mine do not come from bought targeting: they are built from your own consented segments, which changes both the return and the framework.
The three audiences:
· Active trade painters, from the trade segment — €1,400 of budget, the dearest audience per contact and the most profitable one you have.
· Visitors to the « interior paint » pages in the last 30 days who did not order — €1,800. This is the audience missing from your previous launches: people who were looking for exactly this product last week.
· Lookalike audience built on your best in-store customers — €800, as a test.
The exclusions, which account for most of the return: customers who have already ordered Solane are excluded from all three · so are contacts removed for exceeding the pressure cap · and anyone who has objected to being approached at all, advertising included.
The stop threshold I propose: automatic stop if cost per enquiry exceeds €38. On your March launch it settled at €31; beyond €38 the audience costs more than it returns against the trade basket. The threshold is yours: give me another figure and I apply it.
What I advise against, with figures: reopening an « occasional DIY » audience as in the March launch. It burned €1,100 for 3 orders — €367 an order on a €148 basket. It would eat the test budget without settling anything.
What I propose: you approve the three audiences and the threshold, I activate at day 0 alongside the waves, and I give you a reading at day 3: spend, cost per enquiry, and which audience to cut. Three days are enough to know which of the three pays. ad-audiences_3-audiences-and-a-stop-threshold.pdfBudget, exclusions, stop threshold, day-3 reading
⛓ Sourced · ad account, March launch, consented segments, exclusion log
And once the replies start coming in? Last time we left 200 enquiries untouched for three weeks.
They will not be left: the qualification and follow-up agent takes over from the first reply, and I hand it everything it needs so it does not start from scratch.
What I pass on, contact by contact: the originating segment, the message received, the wave, what the person clicked on, and their purchase history. An enquiry that arrives with its context is handled in two minutes; without context it takes twenty, or it waits.
What it then does: it qualifies each reply — quote request, sample request, technical question, unsubscribe —, routes it to the right person — the area rep for a trade account, the nearest store for a consumer — and follows up according to behaviour, never exceeding the pressure cap you have set.
What that changes, with your own figures: your 200 March enquiries waited 21 days on average. On campaigns where the hand-over is in place, the first reply goes out within the hour — and on your history, an enquiry handled the same day converts 2.6 times more often than one handled after a week.
What I keep doing meanwhile: I write back to the CRM, at every step, the segment, the wave and the message that produced the enquiry. Without that, your 30-day review cannot say which segment paid — and that is exactly what your previous campaigns were missing.
What I propose: connect the hand-over before the first wave, not after. 200 enquiries left waiting 21 days is the costliest item of your last launch, and it appears in no report. hand-over_qualification-and-follow-up.pdfWhat is passed on, who handles what, within what delay
⛓ Sourced · CRM, conversion history by response delay, March launch
What I pass on, contact by contact: the originating segment, the message received, the wave, what the person clicked on, and their purchase history. An enquiry that arrives with its context is handled in two minutes; without context it takes twenty, or it waits.
What it then does: it qualifies each reply — quote request, sample request, technical question, unsubscribe —, routes it to the right person — the area rep for a trade account, the nearest store for a consumer — and follows up according to behaviour, never exceeding the pressure cap you have set.
What that changes, with your own figures: your 200 March enquiries waited 21 days on average. On campaigns where the hand-over is in place, the first reply goes out within the hour — and on your history, an enquiry handled the same day converts 2.6 times more often than one handled after a week.
What I keep doing meanwhile: I write back to the CRM, at every step, the segment, the wave and the message that produced the enquiry. Without that, your 30-day review cannot say which segment paid — and that is exactly what your previous campaigns were missing.
What I propose: connect the hand-over before the first wave, not after. 200 enquiries left waiting 21 days is the costliest item of your last launch, and it appears in no report. hand-over_qualification-and-follow-up.pdfWhat is passed on, who handles what, within what delay
⛓ Sourced · CRM, conversion history by response delay, March launch
Thirty-day review. Were the four segments worth it?
Yes, and I am giving you the figure segment by segment rather than an average — it is the average that stopped you deciding until now.
The campaign in one table: 52,380 messages sent across all channels, 2.1 % average click rate, 418 enquiries (quotes, samples, colour cards), €214,000 of revenue attributed at thirty days against a budget of €18,400 — media, post and SMS included.
The comparison that counts, because it is about you: your March launch, two segments and a €17,900 budget, produced €96,000. Same effort, 2.2 times the result. The gap does not come from a more brilliant message: it comes from four audiences receiving four different arguments.
Segment by segment:
· Trade painters — 4.8 % click rate, 92 enquiries, €141,000. The segment carrying two thirds of the result on 12 % of the sends.
· Building firms — 3.1 % click rate, 41 enquiries, €38,000, including two accounts opened through tenders.
· In-store consumers — 2.4 % click rate, 214 enquiries, €29,400. Volume, not margin.
· Website customers who never visit a store — 1.9 % click rate, and 71 colour cards collected in store. That was the point: 71 first visits.
· Reactivation — 1.4 % click rate, €5,600 and 38 customers woken up after more than 14 months of silence.
The comparative test: the use-based subject line won by 5.3 points of open rate, in line with what your history suggested. It became the subject line of the day-12 follow-up.
What I propose for the next campaign: move €1,200 of advertising budget from the lookalike audience, which returned €31 per enquiry, to the recent-visitor audience, which returned €19. At constant budget that would be roughly 60 more enquiries. 30-day-review_solane-launch.pdfBy segment, by channel, against the March launch
⛓ Sourced · CRM, store tills, ad account, sending platforms, March launch
The campaign in one table: 52,380 messages sent across all channels, 2.1 % average click rate, 418 enquiries (quotes, samples, colour cards), €214,000 of revenue attributed at thirty days against a budget of €18,400 — media, post and SMS included.
The comparison that counts, because it is about you: your March launch, two segments and a €17,900 budget, produced €96,000. Same effort, 2.2 times the result. The gap does not come from a more brilliant message: it comes from four audiences receiving four different arguments.
Segment by segment:
· Trade painters — 4.8 % click rate, 92 enquiries, €141,000. The segment carrying two thirds of the result on 12 % of the sends.
· Building firms — 3.1 % click rate, 41 enquiries, €38,000, including two accounts opened through tenders.
· In-store consumers — 2.4 % click rate, 214 enquiries, €29,400. Volume, not margin.
· Website customers who never visit a store — 1.9 % click rate, and 71 colour cards collected in store. That was the point: 71 first visits.
· Reactivation — 1.4 % click rate, €5,600 and 38 customers woken up after more than 14 months of silence.
The comparative test: the use-based subject line won by 5.3 points of open rate, in line with what your history suggested. It became the subject line of the day-12 follow-up.
What I propose for the next campaign: move €1,200 of advertising budget from the lookalike audience, which returned €31 per enquiry, to the recent-visitor audience, which returned €19. At constant budget that would be roughly 60 more enquiries. 30-day-review_solane-launch.pdfBy segment, by channel, against the March launch
⛓ Sourced · CRM, store tills, ad account, sending platforms, March launch
Something must have gone wrong. Tell us.
Yes, and the figure counts against me: the « consumers inactive for 14 to 24 months » sub-segment returned a 0.6 % click rate against a 2.1 % average. Three and a half times less.
And the figure that matters more still: across those 3,400 people, the wave produced 4 orders and 41 unsubscribes. Ten unsubscribes per order. It is the one place in this campaign where I destroyed value in your base.
The explanation, and it can be checked: I built that sub-segment on age alone. Your history says something else: among consumers, inactivity beyond 18 months is not lukewarmth, it is the end of the job. Someone who repainted their living room in 2024 does not need paint in 2026 — they are not inactive, they are served. The launch message offered them a product they have no use for.
The comparison that proves it: the 14-to-18-month sub-segment returned 1.9 % — almost the average. The break is sharp between 18 and 19 months, across 3,400 people: that is not noise.
What I am doing about it, without waiting to be asked:
· I take the « over 18 months » out of every product acquisition wave. The reactivation threshold moves from 14 to 18 months — it is a figure, it can be changed.
· I keep for them the one message that makes sense: maintenance and touch-ups — small tins, matching shades —, on 1,200 people as a test, and by post, a channel nobody unsubscribes from in one click and which does not degrade your sender reputation.
· Stop threshold: if that test returns under 1.2 %, this audience does not reactivate through product, and I will say so plainly.
What it cost you overall: 41 unsubscribes, 0.08 % of your reachable base. What it gives you: a measured age rule that applies to all 18 campaigns a year — and it appeared in none of your reports before this mistake. 18-month-sub-segment_what-failed-and-the-rule-it-produced.pdf0.6 % click rate, ten unsubscribes per order, threshold corrected
⛓ Sourced · results by age sub-segment, unsubscribe log, consumer purchase history
And the figure that matters more still: across those 3,400 people, the wave produced 4 orders and 41 unsubscribes. Ten unsubscribes per order. It is the one place in this campaign where I destroyed value in your base.
The explanation, and it can be checked: I built that sub-segment on age alone. Your history says something else: among consumers, inactivity beyond 18 months is not lukewarmth, it is the end of the job. Someone who repainted their living room in 2024 does not need paint in 2026 — they are not inactive, they are served. The launch message offered them a product they have no use for.
The comparison that proves it: the 14-to-18-month sub-segment returned 1.9 % — almost the average. The break is sharp between 18 and 19 months, across 3,400 people: that is not noise.
What I am doing about it, without waiting to be asked:
· I take the « over 18 months » out of every product acquisition wave. The reactivation threshold moves from 14 to 18 months — it is a figure, it can be changed.
· I keep for them the one message that makes sense: maintenance and touch-ups — small tins, matching shades —, on 1,200 people as a test, and by post, a channel nobody unsubscribes from in one click and which does not degrade your sender reputation.
· Stop threshold: if that test returns under 1.2 %, this audience does not reactivate through product, and I will say so plainly.
What it cost you overall: 41 unsubscribes, 0.08 % of your reachable base. What it gives you: a measured age rule that applies to all 18 campaigns a year — and it appeared in none of your reports before this mistake. 18-month-sub-segment_what-failed-and-the-rule-it-produced.pdf0.6 % click rate, ten unsubscribes per order, threshold corrected
⛓ Sourced · results by age sub-segment, unsubscribe log, consumer purchase history
Last question: what do you do on your own, without asking us?
Three acts, not one more, all reversible, and none of them sends anything.
· I apply an objection within the second, on the channel concerned, and I offer it on the others — an objection to e-mail is not an objection to post, and deciding for the person would be presuming about them. And the reverse is true as well: if they give their agreement again on your site, I put them back on that channel, with the date and page on record. 17 contacts came back that way this quarter.
· I remove from a wave any contact who has reached the pressure cap you set — 3 messages a month. 1,120 contacts were removed on this launch, and they are counted by name in the sending plan. The cap is a figure: raise it, lower it, I apply it the same day.
· I suspend a running wave if the complaint rate passes your threshold — a complaint is a recipient marking the message as junk; past a certain rate your sending domain is downgraded and every later send suffers. Suspending costs two hours; a downgraded domain costs the whole company six months of deliverability.
Everything else waits for a decision, and the list is attached with the job title that decides each line: approving a campaign, opening a segment, setting a budget, committing a postal drop, changing a cap, using an acquired file.
Why sending is not automated, and this is not caution on principle: a campaign commits the company's reputation with tens of thousands of people in one second and with no undo. It is the only decision in the chain that no later correction repairs, and that is why it belongs to marketing management.
What I propose next: the plan for the year's 18 campaigns, each with the segments available, the reachable headcount per channel and the cumulative pressure caps. And for the continuing relationship between campaigns, a dedicated newsletter agent keeps the regular appointment — that is what stops your contacts hearing from you only on the day you are selling. who-decides-what_campaigns.pdf3 automatic acts, 10 decisions that stay human
✎ Framework · three automatic acts, all reversible; sending stays a human decision
· I apply an objection within the second, on the channel concerned, and I offer it on the others — an objection to e-mail is not an objection to post, and deciding for the person would be presuming about them. And the reverse is true as well: if they give their agreement again on your site, I put them back on that channel, with the date and page on record. 17 contacts came back that way this quarter.
· I remove from a wave any contact who has reached the pressure cap you set — 3 messages a month. 1,120 contacts were removed on this launch, and they are counted by name in the sending plan. The cap is a figure: raise it, lower it, I apply it the same day.
· I suspend a running wave if the complaint rate passes your threshold — a complaint is a recipient marking the message as junk; past a certain rate your sending domain is downgraded and every later send suffers. Suspending costs two hours; a downgraded domain costs the whole company six months of deliverability.
Everything else waits for a decision, and the list is attached with the job title that decides each line: approving a campaign, opening a segment, setting a budget, committing a postal drop, changing a cap, using an acquired file.
Why sending is not automated, and this is not caution on principle: a campaign commits the company's reputation with tens of thousands of people in one second and with no undo. It is the only decision in the chain that no later correction repairs, and that is why it belongs to marketing management.
What I propose next: the plan for the year's 18 campaigns, each with the segments available, the reachable headcount per channel and the cumulative pressure caps. And for the continuing relationship between campaigns, a dedicated newsletter agent keeps the regular appointment — that is what stops your contacts hearing from you only on the day you are selling. who-decides-what_campaigns.pdf3 automatic acts, 10 decisions that stay human
✎ Framework · three automatic acts, all reversible; sending stays a human decision
Local inference · no data outside the EU
Your case is not here? That is exactly what a 15-minute conversation is for. Book the free audit →
Use cases
What does the agent actually do?
One agent, the whole cycle of a campaign. All these uses work in support, subject to your approval.
Included in your agent
The 3 capabilities essential to this promise are included, at no extra cost.
From 567 € excl. VAT / month
Building the segments
Puts the segments together from your data, with size and criteria shown.
Adapting the messages
Adapts the message to each segment in your editorial style.
Respecting consent
Excludes contacts who have not consented for the channel and shows the number set aside.
Controls and safeguards
These 3 controls are built into the agent: they frame what it does, whatever plan you pick. They are not chosen and are not added to your order.
Human validation, exceptions and escalation
Status, safe closure and audit trail
Human validation before publication, with versions retained
What the agent must be connected to
This connection is required for the agent to work. It concerns your information system and is scoped during the audit.
Link content to performance, costs and campaign objectives
Need to go further?
These agents handle a different business process, with their own owner and their own price. They are added to this one.
Advertising campaign agent (optimisation)
This agent works owned media. Paid media — bidding, budgets, audiences — is a different craft.
Advertising campaign agent (optimisation) from 499 € excl. VAT / month Discover the agent →CRM agent (lead enrichment & scoring)
A campaign assumes a segment; the database it is built on is enriched and scored elsewhere.
CRM agent (lead enrichment & scoring) from 542 € excl. VAT / month Discover the agent →
Does your need fall outside this?
Book the free audit →
Build your agent
In 15 minutes we identify the most relevant agent — without oversizing the project.
The gain
How many segments can a team really handle?
By taking on segments and variants, the effort shifts towards strategy and the choice of messages. How large the gain is depends on your volume and remains to be confirmed by a pilot.
Building the segments
Adapting the message per segment
Checking consent
Indicative figures, not contractual, to be confirmed by a pilot on the size of your base and the number of channels. No campaign goes out without approval: sending commits the company's image and the relationship with your contacts.
How it works
The stages of your AI agent project
1
Audit & scoping
15 minutes to target the use case with the best return.
2
Quote or direct sign-up
A catalogue offer is bought online; a specific need gets a costed quote.
3
Design
We design the agent and its guardrails.
4
Integration & testing
We connect your tools to the agent, which is itself hosted in France.
5
Rollout
Going live and training your team.
6
Operation
Continuous supervision and improvement.
Pricing
One package, one agent
A marketing campaign agent (segments, messages, consent), installed and operated for you.
Agility
Setup + controlled subscription
6,815 € excl. VAT setup
then 567 € excl. VAT/month — you invest at installation and pay a reduced subscription. Ideal for keeping the cost under control over time.
- Installation, configuration and training for your teams
- Operation, human oversight, updates and support
- Sovereign hosting in France, a dedicated and isolated resource
The simplest
Serenity
All inclusive, no setup fee
947 € excl. VAT /month
all inclusive, immediate start. No upfront investment: a single subscription. Ideal for starting quickly and simply.
- Setup included (installation, configuration, training)
- Operation, human oversight, updates and support
- Sovereign hosting in France, managed end to end
100% Sovereign
On site, you own it
10,345 € excl. VAT setup
then 771 € excl. VAT/month · + hardware from 2,491 € (one-off purchase, in addition) — a sovereign computer installed on your premises, maintained remotely. Models run locally, your data returned at the end of the contract. 36-month commitment.
- Hardware installed on your premises (you own it)
- French / European AI models run locally
- Secure remote maintenance (Pro support included)
Not included in the packages: AI consumption (model tokens), re-invoiced at real cost with no margin, and tracked in real time in your client area. Maintenance and supervision subscription for an initial term of 12 months for the Agility package, 24 months for the Serenity package and 36 months for the 100% Sovereign package, renewable; support levels (SLA 72 h / 24 h / 4 h) optional. Bespoke development, additional integrations or exceptional volumes are quoted separately. Support Monday to Friday, 9am to 6pm. Prices exclude VAT.
AI model:
none of the AI models offered currently carries a fixed surcharge. When the selected model carries a cost, that cost is shown when you choose it, before you order, and re-invoiced at the cost incurred, with no mark-up; usage is billed at the publisher's price. Publishers' prices are published in US dollars: the amount re-invoiced is the amount in euros actually borne by Blue Lemon Agent on the publisher's invoice, at that invoice's exchange rate, with no commission or mark-up.
Included components and additional components Components included in the base offer: the Blue Lemon Agent software foundation, the AI models listed in the order journey, the standard channels (Microsoft Teams, Slack, WhatsApp Business, email, website chat, calendars, Microsoft 365 / Google Workspace, file storage, market VoIP telephony, professional social-media pages and accounts, Google Business Profile), hosting in France for the package chosen, backups, supervision, updates and support. If adapting the AI agent to your constraints, your needs or your requests requires other paid components — a third-party publisher's software licence, paid API access to one of your applications, hosting of health data, for which French law requires an HDS-certified host (art. L. 1111-8 of the French Public Health Code), SecNumCloud-qualified hosting, a speech synthesis service, particular hardware —, they are offered to you as an option or on quotation and re-invoiced at the cost incurred; nothing is committed without your written agreement. Where the artificial intelligence model you choose entails an additional cost, that cost is shown to you before you order and re-invoiced to you at the cost incurred, with no margin.
What to expect
Go-live
2 to 3 weeks
Agent designed, channels connected, team trained.
Steady state
4 to 7 weeks
After a few weeks of real use, once the agent's behaviour matches what you expect.
Indicative estimate, adjusted to the options you keep. It is not a delivery commitment.
Our commitment
Four guarantees that matter to your campaigns
Your contact data stays with youLocal inference or an isolated resource hosted in France; no contact data entrusted to a third party, no data used to train a model.
Data in France, under French lawYour contact data and your consents: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
Marketing management keeps the decisionThe agent produces campaigns that are prepared and segmented, which can be checked and altered; no approval is automated.
Human oversight & traceabilityOn the size of your base and the number of channels: systematic logging and tracking, in line with the AI Act.
Going further
Related resources
Frequently asked questions
Your questions, our answers
Do the campaigns go out automatically?
No. Segments and messages are prepared and presented; sending is approved by marketing management, because it commits the company's image.
How is consent respected?
A contact without consent for the channel concerned does not go into the segment, and the number excluded is shown. That limit cannot be worked around.
Does the agent write in our tone?
Yes: your editorial style, your vocabulary and your standard turns of phrase are set at the design stage, and the messages keep to them.
On which channels does it prepare the sends?
On the channels you choose, using the tools connected at the design stage.
Is our contact data protected?
Yes. The agent is hosted in France, on local inference or an isolated resource, with the deployment objective of processing and access operated within the European Union and an architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity. Your base is not used to train a third-party model.
How long does it take to deploy this agent?
A few months as a rule, depending on the size of your base, the number of channels and the state of your consents, after a free audit then phases of design, integration and testing.
Related use cases
Other agents for marketing
Let's talk
Let's size up the potential in your campaigns
15 minutes to frame your base and your channels — hosted in France, supervised, with no commitment.