Rajesh Jain

BRTN and the Transactions-Attention Table

Parts 2 and 3 mapped the Tax lever. Part 4 maps the Time lever.

The seven transaction buckets reveal where Tax is leaking. They show whether a brand is buying revenue, owning revenue, reacquiring revenue, or surrendering revenue to intermediaries. But Tax is only one half of Alpha. The other half is Time — the gap between one transaction and the next, the speed at which a customer climbs from Next to Test to Best, the rate at which Best customers drift silently into Rest. To see Time, the brand needs a different framework altogether — one that classifies customers, not transactions.

A customer relationship does not collapse in a single moment. It decays in stages. The customer does not wake up one morning and decide to leave the brand. They stop opening. Then they stop clicking. Then they stop visiting. Then the time since last transaction stretches. Only later does the brand notice the absence — usually when the customer has already drifted far enough that the only reliable way back is paid media at 20–25%, or worse, an Intermediated route at 30–40%+.

This is why Time between Transactions cannot be measured only by transactions. By the time the transaction gap becomes visible, the attention gap has already done the damage.

For four decades the working customer-state framework in retail and direct marketing has been RFM — Recency, Frequency, Monetary. When did this customer last buy? How often do they buy? How much do they spend? RFM was a remarkable framework for its era and still works as a basic diagnostic. But it has a structural blind spot that becomes more expensive every year: all three of its variables are transaction variables. Recency means transaction recency. Frequency means transaction frequency. Monetary means transaction value. RFM cannot see a customer who has stopped opening emails, stopped clicking on push notifications, stopped opening the app — until the transactions also stop.

A customer who bought three times and is still opening every week is not the same as a customer who bought three times but has ignored the brand for ninety days. Same purchase count. Different future. Same RFM score, perhaps. Opposite trajectory.

In a world where attention decays before transactions stop, RFM is a rear-view mirror.

BRTN: the four canonical states

The first refinement is to collapse RFM scoring into four states that match how a CMO actually thinks about the customer base.

Best are the brand’s most valuable customers — three or more transactions with current attention. They are the profit engine.

Rest are customers who once mattered but are now drifting or dormant. They are not dead. They are simply no longer paying attention to the brand’s owned channels.

Test are early buyers whose future value is still uncertain. They have bought once or twice, but the relationship has not yet become habit.

Next are the future customers — identified non-buyers and genuine new acquisitions waiting to be converted.

BRTN is powerful because it shifts the marketer’s question from “Who bought?” to “Who is still listening?” But BRTN by itself still carries the RFM blind spot in a softer form: it tells the brand where a customer currently is, not where they are about to go. To make BRTN operational, the CRM team needs a simple grid — the equivalent of RFM for an attention-first world.

The Transactions-Attention Table (TAT)

Call it the Transactions-Attention Table, or TAT.

The rows measure transaction depth. The columns measure attention recency. The critical point is that attention recency means days since last meaningful attention event, not days since last transaction. A meaningful attention event is an email open, click, magnet interaction, app open, push tap, WhatsApp response, product browse, or wishlist action — any signal that shows the customer is still reachable through owned channels.

Transactions ↓ Attention →	0–30 days Strong	30–90 days Weakening	90+ days Lost
0	N	N–	L
1–2	T	T–	R2
3+	B	B–	R1

Nine cells. Each one a distinct managerial state with a distinct strategic prescription.

N — Next. Identified, no purchase yet, attention strong. The classic active lead. Convert.

N– — Weakening Next. Identified, no purchase, attention slipping. The brand still has a chance, but the strategy must shift. Another hard-sell campaign may accelerate the fade. This customer needs relevance, trust, utility — a reason to stay reachable before the lead goes cold.

L — Lost Lead. No purchase and no recent attention. This is not Rest. This person has never bought. Recovery investment should be low. Suppression, repermission, or low-cost attention rebuilding may be appropriate; heavy discounts and paid reacquisition rarely are.

T — Test. One or two purchases, strong attention. The acceleration cell. Drive the next transaction. This is where early-lifecycle marketing matters most — the second transaction is not just more revenue, it is evidence that the relationship may compound.

T– — Weakening Test. One or two purchases, attention slipping. A fragile state — proof of purchase, but not proof of habit. The wrong move is to keep shouting “buy again.” The right move is to preserve attention before pushing the next transaction.

B — Best. Three or more purchases, attention strong. The heart of the business. Best customers should receive the best personalisation, the best service, early access, recognition, and the deepest relationship investment.

B– — Weakening Best. The single most economically important cell on the grid. A high-value customer in the act of becoming Rest, flagged before the transaction signal would show it. The dashboard may still call them loyal because they have bought many times — but attention says something has changed. The cost of holding a B– is a fraction of the cost of recovering an R1. The brand that catches B– early avoids the AdWaste it would otherwise pay to recover them later.

R1 — Rest-1. Former Best customers who have lost attention. These are the most valuable recovery opportunities because the brand has proof of depth. They deserve priority recovery — winning them back protects the largest future LTV. Recovered R1 customers return to B– first — attention restored, transaction not yet re-proven — and graduate to B only when the next transaction lands on an owned route.

R2 — Rest-2. Lower-depth buyers who have lost attention. They matter, but the recovery economics must be more disciplined. Some will be worth low-cost reactivation. Some are better served by attention-monetisation if attention can be rebuilt. Some should simply be left alone. Recovered R2 customers return to T– on the same logic, graduating to T once the transaction follows.

Sell, Relate, Recover — the doctrine the table enforces

The power of TAT is not the labels. It is the action logic the columns impose.

Sell when attention is strong. The left column — N, T, B — is where transaction prompts pay off. The customer is listening. Move them forward: first purchase, second purchase, cross-sell, replenishment, upgrade, referral.

Relate when attention is weakening. The middle column — N–, T–, B– — is the danger zone. The brand must shift from Sell to Relate: fewer hard offers, more value, more utility, more recognition, more memory, more reasons to remain connected. The goal is not immediate conversion. The goal is to stop the attention slide. Most CRM teams do not have a Relate playbook; they have a Sell playbook with the frequency dialled up, which is exactly the opposite of what these cells need.

Recover when attention is lost. The right column — L, R2, R1 — is where the CRM channel has already failed. Recovery must be tiered: R1 deserves more investment than R2; Lost Leads deserve less than past buyers. A single grid prevents the common mistake of treating all inactivity as equal.

The one-line doctrine the table produces:

Sell when attention is strong. Relate when attention is weakening. Recover when attention is lost.

The grid is a velocity field, not a snapshot

A customer is never permanently in one cell of the TAT. Two forces act on every customer simultaneously, and they pull in different directions.

The brand’s CRM effort pushes customers downward through the grid — from N to T to B — by driving transactions. Each successful Sell play moves a customer down a row.

Entropy pushes customers rightward through the grid — from Strong to Weakening to Lost — by attention decay. Every day the brand does nothing, the entire customer base drifts rightward.

The job of a CRM team is to bend trajectories downward faster than entropy pulls them rightward. A healthy operation produces high downward velocity (Convert and Accelerate plays moving N → T → B) and resists rightward drift (Relate plays keeping customers in the Strong column). When the downward force wins, the brand grows LTV. When the rightward force wins, the brand grows AdWaste — because every customer who drifts into the Lost column becomes a candidate for paid reacquisition the brand will pay for next quarter.

This is what the Time lever from Part 1 means, operationally. Less Time between transactions is the downward arrow on the TAT. Compressing N → T → B is the lever in action. Less Tax per transaction is what the Revenue Tax Ladder from Part 2 governs — what each downward move actually costs when it happens.

A brand with many customers in B and T has momentum. A brand with too many in B– and T– has a silent attention crisis. A brand with a large R1 pool has allowed valuable customers to decay. A brand with heavy Repeat Direct Adtech and a large R1 pool is paying the price of having missed the warning signs months earlier — and is now paying the adtech tax to undo what cheap CRM attention could have prevented.

This is the hidden link between Part 3 and Part 4. The transaction taxonomy tells the brand what tax it paid. The TAT tells the brand why that tax may need to be paid again next quarter.

RFM looked backward: who bought, how often, and how much?

TAT looks forward: who bought, who is still listening, and who is about to be lost?

The answer determines the next action. And the next action determines whether the customer moves toward Alpha — or falls into AdWaste.

What the TAT does not yet describe is the intervention. The B– and R1 cells expose a structural problem the Revenue Tax Ladder also pointed to in Part 2 — the cliff between CRM (5–10%) and Adtech (20–25%) has nowhere economically viable for a brand to land a recovery transaction. The grid surfaces the customers who need that recovery. The ladder shows there is no rung to recover them onto. This is not a tactical gap. It is a missing engine. That is the work of the next part.

The Seven Transaction Buckets and the Offer Tax

Once the three questions are answered, every transaction can be placed into one and only one bucket. The classification is hierarchical — Q1 first, Q2 next, Q3 last — and it is mutually exclusive by design. A taxonomy is useful only if a transaction lands in exactly one cell. If the same sale can sit in Direct, CRM, Adtech, and Repeat all at once, the framework collapses into another attribution debate. That is not the goal here. The goal is managerial clarity — a single bucket per transaction, recorded the same way every week, comparable across quarters.

The hierarchy works like this. If the transaction happened on a marketplace, quick-commerce app, retailer, aggregator, or third-party commerce surface, it goes into Intermediated. Stop. Do not also classify it as Organic, CRM, or Adtech — those categories belong to Direct only. The money changed hands on someone else’s surface; that is the defining fact. If the transaction happened on the brand’s own site or app, it is Direct, and the next two questions decide which of six Direct cells it lands in: three demand drivers (Organic, CRM, Adtech) × two identity states (New, Known). Six Direct buckets plus one Intermediated bucket. Seven in total.

The seven buckets

Intermediated (30–40%+). Amazon, Flipkart, Myntra, Nykaa, Blinkit, Zepto, Instamart, BigBasket, offline retail partners, social commerce platforms. The revenue is real but expensive — commissions, visibility spend, platform pricing pressure, controlled delivery economics, restricted customer data. The brand gets a sale and loses the relationship. Not bad revenue; just costly revenue, and structurally unable to compound.

New Direct Organic (0–5%). The best New customer a brand can acquire — first transaction from unpaid demand: direct visit, branded search, SEO, word of mouth, unpaid referral. Brand pull converted into revenue. Tax near zero, identity captured, relationship begins clean.

New Direct CRM (5–10%). A first transaction from someone whose identity was already in the database before purchase — a subscriber, lead, quiz participant, app installer, wishlist creator, cart abandoner — converted by owned-channel nurture. Higher tax than pure organic, far lower than paid. Proof that the brand can convert Zero to One without renting attention every time.

New Direct Adtech (20–25%). A first purchase generated by paid media — Google, Meta, paid social, paid search, affiliate, prospecting retargeting, influencer boost. This is legitimate CAC when the customer is genuinely new and CAC sits comfortably below gross-margin-adjusted LTV. It becomes dangerous only when the brand fails to convert this customer into lower-tax repeat revenue afterwards.

Repeat Direct Organic (0–5%). The purest repeat. A known customer returns on their own — through direct visit, app open, branded search, or habit. No platform was paid; no auction was needed; no journey was triggered. The brand relationship did all the work.

Repeat Direct CRM (5–10%). The strategic core of D2C. A known customer returns because the brand used owned channels well — email, WhatsApp, push, SMS, app notifications, loyalty, recommendations, replenishment reminders. Tax modest, identity owned, relationship compounding. This is the bucket every D2C business should want a large share of repeat revenue to sit in.

Repeat Direct Adtech (20–25%) — the red-flag bucket. A known customer bought again through paid media. The dashboard shows revenue. The ad platform reports ROAS. The campaign manager celebrates the conversion. But the P&L should ask a harsher question — why did the brand have to pay Google or Meta to bring back someone already in its database?

Not all Repeat Direct Adtech is waste. Some categories have long consideration cycles, some customers benefit from external nudges, some retargeting flows are honest reactivation. But when this bucket is large — and in most D2C categories past their first eighteen months, it is — the brand has a relationship problem disguised as a performance-marketing success. Globally, B2C brands spend roughly $500 billion a year on exactly this — re-buying customers they already owned. That figure is not a forecast. It is what is currently happening on the P&L every quarter.

Route Tax is only half the story. Offer Tax is the other half.

The seven buckets account for Route Tax — what the brand paid to create or control the transaction. They do not yet account for what the brand gave away to close it: discounts, coupons, cashback, free shipping, loyalty burn, bundled offers, marketplace-funded promotions that come out of brand economics later. Channel tax and discount tax appear on different lines of the P&L — channel tax shows up as cash paid to a platform, discount tax shows up as foregone revenue — but the operating margin per transaction does not care which line.

The cleanest formulation:

Effective Transaction Tax = Route Tax + Offer Tax

A few examples make the arithmetic concrete.

A Repeat Direct Organic order with no discount carries an effective tax of 0–5%. A Repeat Direct CRM order with a 10% coupon carries 15–20%. A New Direct Adtech order with a 15% discount carries 35–40%. An Intermediated order with platform commission, visibility spend, logistics, and a discount easily crosses 45%.

This is where many D2C brands deceive themselves. A “20% off for our subscribers” email feels like an owned-channel win. The campaign attributes to CRM. The dashboard shows a 5–10% route tax. But add the 20% discount and the effective tax is 25–30% — worse than New Direct Adtech. The brand believes it is running owned-channel economics; it is in fact running adtech economics through its own email list, paying itself the platform fee and handing it back to the customer as a discount.

The diagnostic test is simple. If removing the discount would have lost the sale, the discount is tax. If removing the discount would not have lost the sale, the discount is a gift the brand chose to give for no operating reason at all. When D2C teams run this test honestly across a quarter of promotional campaigns, most discover that a meaningful share of their CRM-driven repeat revenue is sitting on the adtech rung in disguise.

What the seven buckets reveal

Once the table is filled in, the brand can finally read its revenue the way a CFO reads a P&L — not by total, but by composition.

How much New revenue is being bought, and at what effective tax? How much Repeat revenue is owned, and how much is reacquired? How much revenue is trapped inside intermediaries the brand cannot retain? How much margin is being quietly given away through discounts hiding inside owned channels?

These are not marketing questions alone. They are profit questions. The seven buckets turn revenue from a single number into a map, and the map shows where the leak is. A brand that cannot read its revenue at the bucket level cannot tell whether it is creating Alpha — or simply paying for Beta on credit.

Three Questions and the Revenue Tax Ladder

Before a brand can create Alpha, it must first learn to classify its revenue. Most D2C dashboards do the opposite. They cut revenue by campaign, channel, product, geography, or cohort, but they rarely answer the most important question — what tax did the brand pay to make this transaction happen?

The same $50 order can mean very different things. If the customer came directly to the website and bought without prompting, almost the entire gross margin is preserved. If the customer bought after an email or WhatsApp message, the brand paid a modest owned-channel cost. If the customer arrived through Google or Meta, the brand may have paid 20–25% of revenue to rent the moment of attention. If the customer bought through Amazon, Flipkart, Blinkit, or Zepto, the visible sale hides a much larger tax — commission, logistics, discounts, platform ads, fulfilment rules, returns, and the loss of identity itself.

Revenue is not equal. The route matters.

The first step, therefore, is a simple classification exercise. Every transaction answers three questions, in this order, with mutually exclusive answers.

Question 1 — Where did the money change hands?

The brand’s own surface — its website, its app, its checkout — or a third party: Amazon, Flipkart, Myntra, Nykaa, Blinkit, Zepto, Instamart. This single answer decides whether the brand captures the customer’s identity. Direct: identity captured, data complete, the relationship begins. Intermediated: identity stays with the platform, data is partial or absent, the relationship does not start because the brand does not know whom it sold to. A marketplace sale is not a relationship; it is a one-night arrangement. And because the relationship never began, every future transaction with that same person will also have to be paid for through the platform. The tax compounds across the customer’s entire lifetime.

Question 2 — What put the customer in front of the buy button?

Three answers, with sharply different Revenue Tax. Owned attention — direct visit, branded search, organic, app open, push, email, WhatsApp, SMS, loyalty, word of mouth — costs 0–10% of the transaction. Rented attention — Google, Meta, paid search, paid social, retargeting, affiliates, paid influencers — costs 20–25%. Platform-owned attention is the third answer — the Amazon search box, the Blinkit homepage, the Myntra ranking algorithm — but it already sits inside Intermediated from Q1. So for a Direct transaction, Q2 collapses to a binary: did owned attention or rented attention create the sale?

Question 3 — Was this identity already in the database before the sale?

New (the brand had never sold to this person before), Known (the brand sold to them recently and still considers them active), or Known-but-dormant (the brand sold to them once but lost contact long enough that what just happened was recovery, not retention). On Direct, the CRM gives the answer cleanly. On Intermediated, the brand often cannot answer at all — and that blindness is itself the finding. A brand that cannot tell a new acquisition apart from a paid reacquisition is running its CAC numbers in the dark.

The Revenue Tax Ladder

Stack the answers and four rungs appear, ordered low to high. Organic / Direct (0–5%) is brand pull — the customer arrived on their own. The route the whole system should be maximising. CRM / Owned channels (5–10%) is email, WhatsApp, push, SMS, app notifications turning owned attention into a transaction — the strategic core of Retained revenue. Adtech (20–25%) is Google, Meta, paid search and social, affiliates — identity captured, but the tax is 4–5× owned. Intermediated (30–40%+) is Amazon, Flipkart, Blinkit, Zepto — the transaction completes, but no identity transfers, so the full cost is channel tax plus every future transaction the brand will have to pay for again.

The ladder is not a blame chart. Every route has a role. Marketplaces create discovery. Quick commerce creates immediacy. Adtech creates new demand. CRM activates relationships. Organic reflects brand pull. The point is not to eliminate high-tax channels; the point is to know when a brand is using them for the wrong job. A healthy D2C business buys New efficiently and owns Repeat completely. A weak one keeps paying high taxes on customers it should already own.

But there is one structural problem with the ladder as it stands today. The jump from CRM (5–10%) to Adtech (20–25%) is a 15-percentage-point cliff, not a slope. There is no rung in between. When the brand’s owned channels fail to reach a customer in time — when attention decays, when the CRM journey stalls, when the email goes unopened, when the WhatsApp number goes stale — the next available route up the ladder is paid media at 20–25%. There is nowhere softer to land. Every customer the brand fails to hold on the 5–10% rung falls directly onto the 20–25% rung — or worse, onto the marketplace rung at 30-40+.

The cliff is not a bug in the ladder. It is the absence of a rung the brand does not yet have. We will return to that absence later in this essay.

Beta and Alpha

Every business has a Beta. Few create Alpha.  The vocabulary is borrowed from investing, and the borrowing is exact. Beta is the return a fund earns simply by being exposed to the market — if the index rises 10% and the fund rises 10%, that is not skill, that is exposure. Alpha is the excess return generated above the market by insight, timing, or execution. Applied to a brand, Beta is the revenue the business would have done anyway — the trajectory it was already on, set by category growth, prior brand equity, distribution already in place, and standard execution. Alpha is the incremental revenue produced above that baseline. Nothing else counts.

In the Age of AI, Beta belongs to everyone. Every meaningful capability that once created edge is being commoditised by AI within a single cycle: content variants, segmentation, creative testing, agentic execution, personalisation at scale, campaign optimisation. ERP did this in the 1990s. Cloud did it in the 2000s. Mobile did it in the 2010s. AI is doing it faster and across a wider surface than any prior wave. Saying “we use AI” will signal nothing about competitive position by 2027 — the way “we use cloud” signals nothing now. What was once Alpha becomes Beta in months, not years. A brand running on standard agents and standard tools will earn Beta returns — never Alpha.

For B2C and D2C, Alpha can only come from marketing. Product can be copied within months. Pricing is constrained by competitors. Distribution is rented from a small number of platforms. Capital costs the same on both sides of the deal. Supply-chain advantages collapse as supplier networks become shared. The only remaining surface where a consumer brand can produce durable separation is the relationship with its customers — who pays attention to it, who buys repeatedly, who returns without being paid for. Everything else is Beta. In B2C, marketing is now the only place where Alpha is made.

Marketing Alpha comes from two levers: less Time and less Tax.

The mistake most brands make is to treat revenue as one undifferentiated number. It is not. The first lever — less Time — compresses the gap between transactions, so the same customer delivers more revenue in the same window. If the next transaction arrives through an owned route, LTV compounds; if it arrives through adtech after a lapse, the brand pays for a relationship it had already earned.

The second lever — less Tax — routes each transaction through a cheaper rung of the Revenue Tax ladder. Sometimes the tax is near zero — the customer returned on their own. Sometimes it is modest — CRM created the sale. Sometimes it is punishing because Google, Meta, Amazon, Flipkart, Blinkit or Zepto controlled the moment of purchase — 20–25% to adtech, 30–40%+ to a marketplace or quick-commerce platform.

Both levers must be measured above Beta, never against zero. Alpha begins when a brand stops asking “how much did we sell?” and starts asking “how much tax did we pay to sell it, and how long did it take to get the next transaction?”

Pay less tax per transaction. Reduce the time to the next transaction. Never pay twice.

Published June 1, 2026

A new book about how the Pentagon put AI at the centre of warfare made an uncomfortable idea click into place. Marketing already has its own version of that machine, and it is called adtech. This essay argues that adtech is marketing’s kill chain — a find-fix-finish-feedback cycle pointed at customers — and that the alternative is not a faster or kinder chain, but a different shape entirely: the relationship loop that NeoMarketing is built to be.

1

What Project Maven Revealed

1. I have been reading Katrina Manson’s Project Maven, an account of how the United States military put artificial intelligence at the centre of how it fights. The programme began in 2017 as the Pentagon’s Algorithmic Warfare Cross-Functional Team, built to turn an unmanageable firehose of drone and surveillance footage into faster decisions. It later became contentious in public — thousands of Google employees protested their company’s involvement, and Google did not renew the contract. It is a gripping and uncomfortable book about the moment AI crossed from experiment into operating doctrine. I did not expect it to change how I think about marketing.
2. The book is built around four words: Find, Fix, Finish, Feedback. That sequence is the targeting cycle — what the military calls the kill chain. Find what matters in the data; fix its identity and location; finish the action; feed the outcome back so the next cycle is sharper. AI’s role was not to add intelligence in some abstract sense. It was to compress that cycle — to shrink the gap between a signal and an action until the machine, not the human, set the tempo. The human in the loop became the slowest step, and then the step to remove.
3. Marketing has borrowed military language for as long as it has existed. We run campaigns. We pick targets. We talk about acquisition, segments, conquesting, war rooms. For decades this felt like harmless metaphor — colourful borrowing, nothing more. Reading Manson, I stopped being sure it was metaphor at all. In one corner of marketing, the military vocabulary is not borrowed. It is literal. That corner is adtech.
4. My first instinct was the obvious one. If AI can become the operating doctrine of warfare, marketing needs the same thing — its own Project Maven, a benevolent one: AI moving from a feature bolted onto campaigns to the system running underneath them. It felt like a strong idea. A few chapters later I realised it was wrong — not the ambition, but the framing.
5. It is wrong because marketing already has its Project Maven. It was built over fifteen years, it runs continuously, and it has a name. Adtech is the find-fix-finish-feedback machine — and it has been pointed at customers all along. Adtech finds the audience, fixes the identity with a pixel, finishes with a conversion, and feeds the outcome back to sharpen the next cycle. Marketing did not need to build a Maven. It built one a decade ago and forgot to be alarmed by it.
6. So the idea that actually landed was sharper, and less comfortable, than the one I started with. Adtech is marketing’s kill chain. Not a turn of phrase — a structural description. The same four steps, the same compression of signal into action, the same logic in which the human is friction to be removed, applied not to combatants but to the customers a brand has already paid once to acquire.
7. That reframes the question entirely. Marketing’s problem is not that it lacks a Project Maven; its problem is that the one it has is a chain — and a chain ends on a Finish. The shift marketing needs is not a faster chain, or a kinder one. It is a different shape: a loop, where the customer is not the target at the end of the cycle but the partner the cycle exists to keep. The rest of this essay is about that shape.

2

Find, Fix, Finish, Feedback

8. Look at adtech as those four steps and the fit is exact, not approximate. Find. Adtech finds audiences — lookalikes, intent signals, behavioural segments assembled from activity tracked across thousands of sites the brand has nothing to do with. The brand does not know these people. The platform finds them, and rents the brand access.
9. Fix. Adtech fixes identity. The pixel, the cookie, the device graph, the marketplace identifier — each one pins a moving customer to a stable, trackable target. Fixing is the step that makes everything after it possible. You cannot finish what you have not first fixed in place.
10. Finish. Adtech finishes with the conversion — the click, the purchase, the terminal event the whole chain exists to produce. In the military kill chain the Finish is a strike. In adtech it is a transaction. In both, the Finish is the point of the exercise: the cycle is built to end on it, and the moment it ends, that cycle is complete and the next one starts cold.
11. Feedback. The outcome trains the model — and this is the step that matters most. The feedback compounds for the broker, not for the brand. Every cycle makes the platform’s targeting sharper, its next auction smarter, its grip on the attention tighter. The brand pays each time and owns none of the learning. In this system the customer is not a relationship. The customer is the target the machine gets better at hitting.
12. There is a moral structure underneath this, and the book makes it impossible to miss. Maven became controversial not because AI was analysing data, but because AI was being placed inside a chain that could end in force — and every protest around it came down to one question: should a human remain in the decision? Adtech’s stakes are not lethal, and the comparison should not be overdrawn. But the structural question is the same one. In the adtech kill chain, who decides what the customer sees — and who carries the cost of being targeted? The customer sits in neither seat. Consent was never one of the four steps.
13. This is why the answer cannot be a gentler chain. You cannot make a kill chain benevolent by attaching an ethics review to it, because the problem is not the intent — it is the shape. A chain is linear. It ends on a Finish. It treats whatever sits at the end as a target. A more considerate targeting chain still targets. To change what the system does to the customer, you have to change its geometry, not its manners.
14. So the part of my first instinct that was right was the part about operating doctrine — AI should become the system underneath marketing, not a feature beside it. What was wrong was the shape, and the target. In marketing, the customer was never the enemy. The enemy is attention decay; the enemy is reacquisition; the enemy is paying twice for a customer the brand already had. A kill chain aims the most powerful instrument marketing has ever had at the wrong thing. What marketing needs is a relationship loop — a cycle with no Finish, in which the transaction is a waypoint and the customer persists into the next turn. That is what NeoMarketing is built to be: not a faster chain aimed at customers, but the operating layer that recovers them before any chain can reach them.

Figure 1. Two shapes for AI in marketing. The adtech kill chain is linear and ends on a Finish — the customer is the target. The NeoMarketing loop is cyclical and has no Finish — the customer persists at the centre, and the learning compounds for the brand.

3

The Loop, Not the Chain

15. NeoMarketing is the counter-architecture, and it is defined by its shape before anything else. Where the kill chain is linear and ends on a Finish, the loop is cyclical and has none. The transaction is not the terminus — it is a waypoint, and the customer persists into the next turn. Same raw materials as adtech: AI, signals, decisions, speed. Opposite geometry — and the geometry is the doctrine. A chain is built to complete a conversion. A loop is built to keep a customer.
16. The loop also runs on a different question. The kill chain asks: who can we find, fix, and finish? NeoMarketing asks: whose attention is decaying, and what should we do before the relationship breaks? The first question treats the customer as something to be located and converted. The second treats the customer as a relationship to be kept. Every difference that follows — economic, operational, moral — descends from that single change of question.
17. The loop has four beats of its own, a deliberate inversion of Find, Fix, Finish, Feedback. Sense the customer’s state — read where a known customer sits on the attention axis, rather than find a stranger. Orient on the relationship context — the prior history the kill chain discards. Act with the next intervention — recover attention, rather than extract a conversion. Compound — record the Decision Trace and let it grow the brand’s asset, not a broker’s model. A chain ends. A loop returns, and is larger each time it does.
18. NeoMarketing occupies a specific place in the stack: Post-CRM, and Pre-Adtech. It begins where CRM’s reach has faded — where a customer has stopped responding to ordinary owned-channel messaging — and it operates before the brand pays the adtech reacquisition tax. CRM works while attention holds. Adtech works once the brand is willing to pay twenty to twenty-five percent. Between them sits the customer who has gone quiet but should never be handed to the kill chain at all. That customer is the loop’s whole reason to exist.
19. Seen step for step, the contrast between the two architectures stops being rhetorical and becomes structural. Every row in the comparison below is a design decision, and at each step NeoMarketing makes the opposite one — not because the opposite sounds better, but because the loop is solving a different problem from the chain. The chain exists to complete a conversion. The loop exists to keep a customer.

Figure 2. The kill chain and the loop, step for step. Every row is a design decision; the loop makes the opposite one at each step.

20. None of this makes the loop the easy choice. The kill chain is the path of least resistance — already built, already funded, already the default; a brand can buy into it this afternoon. The loop has to be assembled, and run, on purpose. But the two shapes build different things. Every turn of the chain rebuilds the broker’s asset. Every turn of the loop builds the brand’s. That is the trade this essay is really about — and it is worth the harder path.
21. A shape, though, is only a promise until something runs inside it. The loop has to sense real attention states, orient on real relationship context, and act through real channels — and that calls for engines built for exactly those tasks. NeoMarketing has two of them. The next part goes inside the loop: what it senses, why it is structurally cheaper than the chain, and what Atrium and Meridian each do to keep the wheel turning.

4

Inside the Loop

22. Begin with what the loop senses. The kill chain senses intent — signals that someone is in-market now. The loop senses something earlier and quieter: attention. Revenue decay begins as attention decay. A customer does not usually stop buying and then stop paying attention; they stop paying attention first, and the lost transaction arrives months later. By the time a customer registers as lapsed in the data, the relationship has been weakening, unnoticed, for a long time. The loop is built to see that early — to treat attention as its own axis, rather than wait for the transaction to fail.
23. This is why the loop tracks state, not just tier. A customer has a transaction tier — what they have bought — and an attention status — whether they are still listening. Rest is not a tier at the bottom of a ladder; it is the attention-lost condition, and it appears across every tier. A Rest-from-Best customer and a Rest-from-One customer have lost attention in the same sense but carry very different recoverable value. The loop senses both axes. The kill chain, and the dashboards built in its image, see only the transaction.
24. The loop is also structurally cheaper, and the reason is the starting point. The kill chain starts cold — it rediscovers the customer from behavioural exhaust, having discarded everything the brand already knew. The loop starts warm. It begins from prior relationship context: last category, message history, channel response, purchase tier, attention status. A warm start is both lower-tax and faster, because it does not pay to rediscover what the brand already owns. Where adtech reacquisition runs at a twenty to twenty-five percent tax, the loop runs at a fraction of it — on customers the brand already has.
25. Two engines turn the loop. The first is Atrium, the attention engine, for Rest and Next customers. It earns attention back through NeoMails and the units that ride inside them — Magnets, Mu, ActionAds — and through NeoNet, the cooperative recovery network. Atrium does not open with ‘buy now.’ It opens with a reason to engage, and it drives the cost of acquisition towards zero by recovering customers the brand has already paid for once — and should never have to pay for twice.
26. The second is Meridian, the outcomes engine, for Best customers. Where Atrium recovers attention, Meridian converts it: it uses M-Agents, Context Graphs and the Decision Trace to turn a recovered relationship into the next transaction, and the one after that. Meridian is not a targeting machine; it is an outcomes-underwriting machine — accountable for the lifetime value it produces, not the impressions it serves. Atrium compresses time-to-attention. Meridian compresses time-to-transaction. Together, they are the loop running.
27. And here the loop’s defining property appears. A chain is a cost: each turn starts cold, ends terminal, and the learning it generates compounds for the broker. A loop is an asset: it compounds for the brand. Atrium recovers attention at low cost; Meridian converts that attention into outcomes and lifetime value; the lifetime value funds the next round of attention investment; and every turn leaves behind a richer Decision Trace and a larger owned-attention surface. The wheel does not merely turn — it carries more each time it does.

Figure 3. The loop compounds. Atrium recovers attention; Meridian converts it into outcomes; lifetime value funds the next round of attention; and every turn leaves the brand a larger asset — owned attention, Decision Trace, lifetime value — than the turn before.

28. AI is going to become the operating doctrine of marketing. That much of the Project Maven instinct was right, and it is already happening. The only open question is the shape it takes. The kill chain is built and waiting; the relationship loop has to be chosen, and built, on purpose. That is the work NeoMarketing exists to do. Marketing does not need AI to target customers faster. It needs AI to stop losing them in the first place. Adtech already built marketing’s kill chain. The loop is what marketing has to build now.