Published June 21, 2026
Seventeen objections a sceptical CMO or CFO should raise — steel-manned, then answered straight, conceding where the critic is right.
A framework that cannot survive its own strongest critics will not survive a procurement committee. So here is the case against NeoMarketing, made as well as a sceptic would make it — and answered straight, conceding wherever the criticism lands. The objections arrive in three families, and each is answered in a different register. The first family says the model is too crude — those are conceded and sharpened. The second says the proof is rigged — those are conceded and governed. The third says the business will not hold — those are answered at the level of principle.
1
The model is too crude
- “It is just RFM with a new coat of paint.”
Recency, frequency and monetary value (RFM) is forty years old. Renaming the cells Best, Rest and Next adds words, not insight. The charge is fair about half the grid: the depth axis really is RFM — how often someone buys, and how much. The difference is the second axis: attention. RFM is built only from past purchases, so it can only tell you what a customer has already done. The TAT adds a forward-looking signal — is the customer still reachable and engaged? — and that is what lets you spot a Best customer going quiet months before any sales report would show the loss. RFM tells you the past; the attention axis tells you what is about to happen. A real second dimension is not a relabel.
- “Your attention signal is built on a sensor that broke in 2021.”
The axis leans on email opens, and opens have been unreliable since Apple’s Mail Privacy Protection and bot prefetch — many recorded opens are now machine-generated, not a real person. Agreed, completely. So opens are weighted down to almost nothing, and attention is read as a mix of signals — weak ones like opens and views, medium ones like clicks, app sessions and message reads, and strong ones like site visits, wishlist and cart activity, and purchases. The axis reads the strongest evidence it can find, precisely because the easy signal is now junk. One point keeps it honest: a purchase is the surest sign a customer is still engaged, but the attention axis earns its keep on the signals between purchases — the quiet drift a sales report cannot see yet. Depth and attention are kept as separate axes on purpose, so the grid never collapses back into a single ‘did they buy’ read.
- “It breaks for long purchase cycles.”
For furniture, insurance or cars, a loyal customer may not buy — or even engage — for a year or more. Fixed 30- and 90-day windows would file them as Lost when they have not left at all. Agreed: the default windows suit short-cycle categories, and on a furniture brand they would flag a perfectly loyal customer as gone. The fix is to set the attention clock to the category’s rhythm, not the calendar. The real question is not ‘did they engage in the last 90 days?’ but ‘can I still reach them on an owned channel when their next purchase is due?’ The thresholds scale to each category — the Alpha Audit asks for the typical gap between purchases at setup, so a skincare brand and a furniture brand never run the same clock — and ideally to each customer’s own pattern. For subscriptions and policies, the clock is set to the renewal date. And the point cuts the other way too: a high-value, slow-cycle category is the strongest case for Never Pay Twice, because losing reach there means buying the customer back through an expensive search auction when they finally return.
- “It misreads whole types of customer.”
Some customers buy without ever opening a message; others only engage when there is a discount. An opens-based view would file the silent buyer as lost and the discount-only buyer as healthy — both wrong. Agreed on both, and the same move fixes both. The silent buyer lands in the right place the moment a purchase counts as attention; their real risk is not being misread but having no owned channel at all — a Capture and Relate gap, not a flaw in the axis. The discount-only buyer is caught by a brand-versus-deal read of each customer’s Offer Tax: someone whose every order carries a deep discount has ‘deal’ attention, winnable only by giving away margin, while a full-price buyer has ‘brand’ attention. So attention has two parts — how reachable a customer is, and what kind of attention it is. The grid shows where a customer sits; the overlay shows what kind they are.
- “It ignores offline entirely.”
Half of retail happens in stores you cannot track; you cannot watch someone browse and walk out. But an offline purchase places the buyer correctly with no tracking at all — the purchase is the attention. The real gap is offline engagement that is not a purchase: the shopper who comes in, looks, and buys nothing, invisible if you only count transactions. Much of that is recoverable through signals the customer has agreed to share, which nobody is feeding into the axis yet — loyalty-card scans, app check-ins, click-and-collect, e-receipts, service visits. The honest limit: a pure-cash, no-identity business falls back to ordinary RFM, because there is no attention to read between purchases. That is a real boundary — and it lies outside the digital-first brands this is built for.
- “Only the top decile matters — so why model the whole base?”
Customer value follows a power law: the top 5 to 10 per cent earn most of the money and the long tail adds almost nothing. So just look after the top and ignore the rest. Right about the top, wrong as a way to grow. The power-law curve is a photograph; the framework runs the film. Give the top tier velvet-rope treatment — that is Meridian’s job. But today’s top customers were yesterday’s middle, moved up; stop developing the middle and the curve quietly hollows out while the photograph still looks fine. The tail gets its own answer — earn a little revenue from its leftover attention and otherwise spend less on it, rather than nursing it. That is three decisions, not one. And the curve is steeper than it needs to be partly because the model never develops the middle it writes off.
2
The proof is rigged
- “You set your own baseline, and you get paid for beating it.”
Progency’s value claim — Alpha above Beta — and its fee both rest on a number the vendor can influence. Set the baseline low and the Alpha is invented. It is a fund manager choosing his own benchmark. This is the most important objection here, and the answer is built into the method, not promised. For recovery there is no baseline to set, because the yardstick is a control group the brand can check for itself. Split the lapsed customers at random into three groups: one Progency works, one left completely alone, one given to adtech. The untouched group shows how many come back on their own. Progency is paid only on the extra recoveries above that group — the lift, not the gross total — and at a fraction of what the adtech group cost for the same result, measured the same way.
- “The suppression test cannot see what it is breaking.”
A 30-day test that pulls active customers out of retargeting will show the spend was not adding sales — but it will miss the brand awareness that was quietly feeding those same customers. Cut too hard on a short test and you damage the demand that made the owned channel look good. Agreed — a fair warning. A suppression test reads short-term, lower-funnel effects well and brand effects poorly, so the cuts are phased, not flipped overnight. Repeat-retargeting goes first, because it is the easiest to replace and does the least brand-building; broad prospecting is cut last and slowest, over a longer window, watching for knock-on effects on organic and new demand. The rule is to move only as fast as the measurement can actually see.
- “Your holdout is contaminated — those customers still see your other marketing.”
A control customer is not sealed off. They still see your social posts, your marketplace listings, an influencer, a billboard. So how do you separate Progency’s effect from everything else reaching the same person? Agreed: perfect isolation is impossible in live marketing, and any vendor who claims it is bluffing. The honest method is random assignment: because the worked group and the control group are picked at random from the same pool, every other influence — social, marketplace, organic — hits both groups equally, so the gap between them is what Progency added and nothing else. A log records what each customer saw and what they did, and the result is reported as a range, not a falsely exact figure. Other marketing does not bias the comparison; it just widens the range — and we show the width rather than hide it.
- “Last-click attribution is too crude to build a tax on.”
The whole tax calculation rests on last-click attribution, which everyone knows gives too much credit to the final touch and too little to everything before it. Agreed — last-click is rough, and we do not pretend otherwise. The defence is consistency, not accuracy: the rule is fixed for the period and applied the same way to every channel, so the comparisons hold even if the absolute numbers are soft. The findings that matter — repeat revenue is leaking to paid, this group is drifting — hold true under any consistent rule, because they are about which way things are moving, not exact credit. A better attribution model sharpens the numbers; it does not change the diagnosis.
- “The numbers are unproven, and the eventual proof will be self-selected.”
The headline figures — the leakage percentage, the recovery rung — come from your own analysis, not from proof at scale. And when proof does come it will be skewed, because the brands that adopt are the ones that both have the problem and can fix it. Both true, and both said out loud. The benchmark numbers are provisional, the recovery rung is marked medium-to-low confidence, and the early evidence is thin and self-selected. The answer is the same throughout: a control group on every play, a shared record of what was done and what resulted, and a habit of publishing the pilots that failed alongside the ones that worked. A framework that only ever shows its wins has told you nothing; the controls are what make the numbers mean anything.
3
The business does not hold up
- “My own stack and a smart analyst could just do this.”
The TAT is smart segmentation; the plays are good CRM practice; none of it is patented. Why pay a vendor for something a capable team can build? Largely true — and on purpose. The framework is open by design; owning the language of the category is the strategy, not the moat. What a team cannot easily copy is everything around it: the outcome-priced model that takes the risk off the brand, recovery engines tuned across many brands, a cooperative network no single company can build alone, and the first-party data scale that makes the predictions sharper. If the framework were the defence, there would be no defence. It is the execution and the economics — and a framework you can read for free is doing its job.
- “You are on rented land while criticising rented land.”
NeoMails and the whole attention play depend on Gmail, Apple and the AMP rules — the same gatekeepers, able to change the terms overnight, that the doctrine attacks adtech for relying on. The dependency is real and belongs in any honest risk list. The difference is what you keep when the ground moves: with adtech you rent the audience and lose it the moment you stop paying; with an owned channel you keep the identity and the relationship even if one channel gets worse. That is why the doctrine spans channels by design — email today, WhatsApp and others next — and why Capture Identity is a play of its own. The channel can change; the relationship does not reset to zero.
- “Putting ads in my customer emails will cheapen my brand.”
My customers gave me their attention to hear from me, not to be sold to someone else. ActionAds inside NeoMails could cheapen the brand, lose trust, and train people to ignore the very channel I am trying to bring back to life. The risk is real, and we grant it without hedging. Attention is monetised under the brand’s control or not at all: ActionAds are permission-based, relevant, frequency-capped and brand-safe, and the brand sets every exclusion. Premium brands can limit them to partner offers, samples, surveys and co-branded tools rather than open third-party ads. Monetisation starts in lower-value and non-buying groups, never in Best. And the number that governs it is not ad revenue but net attention yield — revenue earned minus the cost in unsubscribes, complaints and falling engagement; if that is negative, the ads come out. A funding method that damages the thing it funds is not self-funding — it is borrowing against the brand.
- “Cooperative recovery across brands is a privacy and consent problem waiting to happen.”
NeoNet moves customers between brands — how do I know it does not break consent, platform rules, or the next privacy law? This would block any procurement review if it were not answered in the design, so it is answered there. No raw personal data passes between brands; matching uses hashed, consented identifiers, and each customer’s opt-in and unsubscribe travel with them. Every brand controls its own eligibility, category exclusions, frequency caps and suppression lists, and a customer who has opted out of one brand is never shown to another. NeoNet is deterministic but permission-based, built privacy-first rather than bolted on later — because a network that treats consent as an afterthought is one law away from being switched off.
- “My data is too messy and my teams too siloed to run this.”
The Alpha Audit wants transactions, channels, customer IDs, paid spend, discounts and marketplace economics — owned by finance, commerce, CRM, paid media and IT, none of whom sit in the same room. For most brands this is simply true on day one. The answer is to start with a Lite Alpha Audit — transactions, channel and customer ID, which almost every brand already has — and add the engagement, discount and paid-spend layers as the data becomes available. Every number comes with a confidence level and a data-readiness score, not false precision. And not being able to get the data owners in a room is not a reason to wait; it is the first finding, because a brand that cannot assemble its own customer picture is exactly the one leaking the most through the gaps between its teams.
- “Where is the offence? This is all defence.”
Almost every play protects or wins back the existing base. The only real new-customer engine is NeoNet, which is the least proven part of the system. A brand that has to grow fast hears ‘stop spending on paid’ as ‘stop growing.’ Fair, and better said plainly than glossed over. This is a profitable-growth doctrine for brands whose growth has become inefficient — not a land-grab playbook for the all-out acquisition stage. It makes the existing base do more before the brand buys the same customers back at full price, and it adds new customers through Capture and NeoNet — but NeoNet is described as early, not oversold. For a brand whose only lever is more paid acquisition, this is the wrong year to adopt. For the far larger group whose acquisition costs keep climbing while their repeat business leaks away, defence is the offence.
**
The objections that weaken on inspection
Two charges weaken under a close look, though neither fully dies — and saying so is the point. ‘It is just RFM’ weakens because the attention axis is a real second dimension, not a relabel; but half the grid is RFM, so the honest claim is that the framework builds on a proven tool, not that it replaces it. ‘It is a vendor pitch’ weakens because a model that holds back a control group and takes its fee only on proven lift is the opposite of a vendor chasing the biggest fee; but it is not fully dead until that governance is shown in the field, not just on paper. Both are weaker after a close look. Neither is gone.
The honest case also draws the line around where the doctrine fits, and a framework that says who it is not for is easier to trust than one that claims to be for everyone. NeoMarketing needs identifiable customers, repeat or renewal business, real owned channels, and enough volume to run a control group. A pure marketplace seller with no way to reach customers directly, a one-purchase category with no repeat, an early-stage brand still finding product-market fit — for these it is the wrong doctrine, or the wrong year.
So the honest case does not weaken NeoMarketing; it defines it. The conditions it names — signal quality, category rhythm, control-group governance, privacy by design, brand-safe monetisation, and outcome pricing — are not weaknesses to hide; they are the specification. NeoMarketing should not win by claiming there are no objections. It should win because it has built the objections into the operating system.
