1
The Gap No One Gets Paid to Close
A cart gets abandoned. A KYC form is left half-filled. A lead goes quiet after showing real interest. In most companies, the same thing happens next: CRM tries a few times, then stops. Weeks later, the same customer starts turning up in retargeting ads — the brand paying a platform to remind someone of a relationship it already owns.
This is the pattern behind NeoMarketing’s central complaint: brands routinely pay twice for the same customer — once to acquire them, again to remind them they exist. NeoMarketing’s Three NEVERs name the fix directly: Never Lose Customers, Never Pay Twice, Never Pay Fixed.
Turning that into a working system starts with a segmentation few brands do explicitly — BRTN: Best, Rest, Test, Next. Best customers — typically the top fifth, engaged within the last month — already deliver outsized value; the job is protecting and growing it. Rest customers have gone quiet over the past one to three months — not lost, just drifting; the job is stopping the drift before it hardens. Test customers have been silent for ninety days or more — the dormant base every brand carries and few will name; the job is reclaiming them on owned channels before adtech sells them back. Next customers haven’t been acquired at all yet. Most brands serve only Best and Next — they reward loyalty and celebrate acquisition — while the middle two segments, where sixty to seventy per cent of the base usually sits, are exactly where the leak lives. (Operationally, Test sits within Rest: two stations on the same slide from attention to silence.)
NeoCore is the engine built to run this structure. Meridian serves Best customers, underwriting outcomes to maximise lifetime value. Atrium serves Rest and Next, running an attention marketplace built to push the cost of reactivation and acquisition toward zero.
Why the relationship goes quiet in the first place
It helps to be precise about why CRM’s messages stop working, rather than treating drift as an unavoidable fact of life. Email has four jobs — SNDR: Sell, Notify, Digest, Relate. Sell is the commercial ask: an offer, a promotion, a conversion nudge. Notify is the service layer: an order confirmation, a shipping update. Digest curates the customer’s world — markets, destinations, ingredients, whatever the category knows well — with the brand as editor rather than subject. Relate builds the relationship itself: recognition, rituals, content that asks for nothing. Most brands send only the first two — and both are withdrawals from the customer’s attention, because every one either asks for something or merely reports a transaction. Digest and Relate — the two deposits — are the emails most brands have never sent. That explains something CRM dashboards usually hide: a customer can be technically “reachable” and still be drifting, because nothing in the inbox has given them a reason to keep opening. BRTN tells you where a customer has drifted to. SNDR tells you why.
What it costs to look the other way
The alternative to closing this gap isn’t free. At a typical 4-5x return on ad spend, a brand hands back roughly 20-25% of the revenue generated on that campaign, in fees, to the platform that delivered it. For a genuinely new customer, that’s simply the cost of acquisition. For a customer the brand has met before — whose email is sitting quietly in the CRM, unresponsive but not deleted — that 20-25% is money paid a second time for something already owned.
Between how CRM behaves and how Meridian and Atrium operate sits a gap most martech stacks quietly ignore. CRM’s automation gives up on a customer after a fixed number of retries — that’s how campaign tools are built. Adtech has the opposite economics: it takes over exactly when CRM stops trying, and starts charging the brand a media fee to reach the same person again. In between — after CRM has stalled, before the brand pays adtech to re-acquire — sits a stretch of unfinished value that neither system is actually built to close.
Call it the post-CRM, pre-Adtech gap. It’s also pre-call-centre, pre-agency, and pre-manual-escalation in a great many cases — any point where the brand still owns the identity, the history and the context, but has no accountable operator for what’s left unfinished. Nobody in the standard stack is accountable for it, priced for it, or even measuring it, because it falls between systems, each designed to hand it off rather than close it.
This is the specific space Progency exists to close. Progency is NeoCore’s managed-service layer, run by Martech Growth Engineers (MGEs) working with M-Agents — the agent collective that does the machine-scale work of cohort discovery, message variants, and journey decisioning, while the MGEs supply judgement, governance, and the accountability line back to the brand. The pairing matters: without the agents, a managed-service team is simply the client’s own CRM team outsourced — same bandwidth ceiling, same economics, same give-up point. With them, a small number of MGEs can run outcome-based interventions across pools no in-house team has the capacity to touch — exactly the customers CRM has stopped trying and adtech hasn’t yet re-bought. It isn’t a new platform and it isn’t a new segment. It’s a dedicated operating function sitting precisely in the space the rest of the stack leaves empty.
Naming the gap explicitly turns an invisible leak into something measurable, sellable, and priceable. Once you can point to this customer, this unfinished journey, this exact moment CRM gave up, you can ask a sharper question than “how do we get more customers?” You can ask: how much of the value we already have is leaking through this gap right now, and what would it cost to close it?
The answer differs depending on who the customer is. A customer who has gone quiet entirely is a different problem from one who was actively transacting right up until the journey broke. Progency treats them as two distinct books of business, priced against two different alternatives — which is where the next part picks up.
2
Two Books, Two Alternatives
Not every customer who falls into the post-CRM, pre-Adtech gap got there the same way, and treating them as one undifferentiated pool is the fastest route to underpricing the harder cases and overpaying for the easy ones. Progency splits the gap into two books.
Finish is for the engaged-but-stuck base — customers mid-journey when things broke: an abandoned cart, an incomplete KYC form, a lead that cooled after real interest, a quote never followed up. These customers haven’t drifted away; a specific transaction simply never completed. Here the alternative isn’t adtech at all — it’s whatever the brand’s own fallback happens to be: a call-centre follow-up, an agency retainer, a manual outreach queue, or in a lot of cases, nothing.
Recover is for the non-engaged base — lapsed buyers, dormant subscribers, one-time purchasers who never returned. For this book, the alternative the brand is weighing is almost always Adtech: pay a platform to put this same customer back in view via a paid ad. Recover’s job is to win the customer back on owned channels first, at a fraction of that cost, before the brand ever reaches for its ad budget. It’s priced against what paid reacquisition would have cost — a return equivalent to roughly 6x has proven realistic, or roughly half of what an equivalent paid campaign would run.
That distinction corrects a claim that’s tempting to make and doesn’t survive scrutiny: that these plays “never go to adtech, because adtech can only target cohorts.” That isn’t accurate. Dynamic retargeting — the kind that shows someone an ad for the exact product left in their cart — already does precise, one-to-one targeting. Adtech is entirely capable of chasing an abandoned cart.
The honest differentiator isn’t that adtech can’t do this. It’s that owned channels do it better, on four counts. Identity: Progency already knows exactly who this customer is, deterministically, with no probabilistic ad-matching involved. Context: it knows precisely what was left unfinished — which product, which form field, which step — not just a broad interest signal. Cost: another message to a known customer costs a fraction of a paid impression. Proof: because the intervention runs on owned channels against a proper holdout group, the lift is measured cleanly, rather than inferred through an ad platform’s own attribution model.
The declared leakage pool
Neither book starts with Progency simply being handed a customer list. It starts with a declared leakage pool — a specific, named group of customers where the brand, the in-house CRM team, and Progency agree on three things upfront: the current journey has genuinely stalled, the desired outcome is measurable, and the comparison baseline (what the brand’s own current best effort achieves) is clear. Incomplete KYC applications from the last quarter. Leads that went cold after a product demo. A suppressed email segment nobody has touched in six months. Each is a pool with a name, a size, and an agreed definition of success — not a vague mandate to “help with retention.”
That boundary does two jobs at once. First, it protects the in-house marketing team politically: Progency isn’t taking over BAU CRM, brand, or strategy — it’s working the specific pools the team has already agreed are stuck, dormant, or too costly to chase through existing methods. Second, it protects Progency from becoming generic services. Without the declared-pool discipline, any outcome-shaped request could get pitched as “Progency,” which dilutes the model into something that looks like an agency retainer with extra steps. A declared pool, an agreed baseline, and a measurable outcome are what keep this a specific, underwritten service rather than marketing-as-a-vague-favour.
Put simply: adtech can chase these moments. Progency finishes them — faster, cheaper, and with cleaner proof of what actually worked.
The practical upshot for a CMO is that Finish and Recover aren’t two versions of the same pitch dressed up differently — they solve different problems, against different competitors, on different customer states. Most brands run both problems simultaneously: a live stream of engaged customers falling out of half-finished journeys every day, and a Rest segment slowly decaying toward being repurchased by adtech.
Because the alternative each book displaces is different, the price for each has to be different too — and that’s where an earlier temptation, pricing everything the same way, runs into a problem the next part addresses directly.
3
Paying for Progress, Not Attention
One of the Three NEVERs is Never Pay Fixed — the idea that brands shouldn’t pay a platform for exposure regardless of whether it produced anything. It’s the core complaint against adtech’s CPM model: pay for the impression, hope for the result.
It would be easy for Progency to quietly reintroduce the same problem from the other side — charging a flat fee per email opened, regardless of what happens next. An open is a rendering event. It tells you the message reached an inbox and someone glanced at it. It doesn’t tell you whether the customer moved a step closer to finishing the KYC form, replying to the follow-up, or returning to the cart. Billing for it anyway would be paying for attention, not outcome — the exact pattern NeoMarketing exists to end, wearing a different logo. It would also make the pricing indistinguishable from a CPM by another name.
So opens stay where they belong: on the diagnostic dashboards NeoMarketing already runs to track attention health (Click Retention Rate, Real Reach, and the rest of the NEVER Metrics) — not on the invoice. What Progency prices instead is a ladder of named, verified commercial units, each closer to revenue than the last.
Pay-for-Action covers two of the rungs. At its weakest, that’s a click, an in-message tap, or a reply — a real, verified action, but not yet meaningful progress. At its strongest, it’s a completed form step, a qualified lead, or a quote requested — the customer visibly moving through the exact journey that had stalled. Pay-for-Data sits alongside it: a preference answered, a renewal date confirmed, a missing KYC field filled in — first-party information the brand didn’t have before, written back into the Customer Context Graph, priced by qualified usable fields rather than raw data points. At the top of the ladder, Pay-in-Email prices a transaction completed inside the message itself, and Carry pays a share of verified uplift on larger pools with a clean holdout.
Why the ladder needs more than one rung
Pay-in-Email is the cleanest outcome — money moves, the result is visible — but it can’t be the only rung, because plenty of brands hesitate to commit to it on day one. The hesitation is rarely about the economics; it’s operational. Finance wants to know who owns the payment flow. Legal wants sign-off on a new transaction path. Whoever runs the dormant base doesn’t want the first message a lapsed customer receives in months to ask for money — that reads as pushing them further away, not winning them back. None of that is irrational, and Progency shouldn’t need a brand to clear every internal hurdle before a pilot can start.
That’s the entire argument for the ladder existing: a brand that isn’t ready for Pay-in-Email can start on Pay-for-Action or Pay-for-Data, see the mechanism work on a small, safe slice of its Rest base or a single stuck-lead pool, and climb toward Revenue and Carry once trust is established. Underneath all four tiers sits the same contract logic NeoCore uses for Meridian’s Best-customer outcomes: a small baseline payment (Beta), an upside tied to verified lift above what the brand’s own current effort would have achieved (Alpha), and a payout on that lift (Carry) — always measured against a randomised, concurrent holdout, never against doing nothing at all. That comparison to the brand’s actual current best effort, not to silence, is what makes the lift figure defensible in front of a CFO rather than a marketing team’s own optimistic assumption.
What doesn’t change is the underlying discipline: every tier is still an outcome, verified against a holdout, never a payment for reach alone. The ladder makes the model easier to start. It doesn’t make it easier to cheat.
That leaves the practical question of where to start climbing — which pool to run first, and why the obvious answer usually isn’t the right one.
4
The Operating Layer
Progency shouldn’t begin by building a new customer engagement platform. Netcore already has the rails — email, CE, CPaaS, WhatsApp, RCS, CDP integrations, Unbxd. The right move is an intelligence and operating layer above what already exists, not a rebuilt stack underneath it. Every new idea is tempted to become a new platform; Progency has to resist that temptation deliberately, because the first version isn’t software sold to the brand — it’s an accountable operating system, run by people, that gets more automated as it proves itself.
The declared leakage pool is where every engagement starts — named, sized, with an agreed stall point and a known current best effort to measure against. From there, the action surface follows a simple discipline: email as the low-cost owned room for attention, data capture, and completing the outcome in place, with WhatsApp, RCS, and SMS held back as escalation rails for the moments that genuinely need immediacy — an urgent KYC deadline, a lead about to go cold for good. The rule is channel-fit, not channel-loyalty: use the cheapest owned or cooperative surface that can actually finish the job.
Underneath that sits the utility layer — the Living Email Factory, AMP components, Pay-in-Email infrastructure, tracking middleware, and the Attention Processing Unit (Magnets, Mu, ActionAds, and the Ledger that records it all). These tools are necessary. They are not, on their own, defensible, because a competent competitor can build broadly similar tooling given enough time and budget.
M-Agents and Martech Growth Engineers run the pool day to day. M-Agents handle the repeatable work — cohort discovery, message variants, channel selection, response scoring, test monitoring. MGEs make the judgement calls: what a specific customer’s context actually means, when to escalate a channel, how to read an ambiguous response, how to keep the brand’s tone and consent rules intact. Automation follows this pattern, rather than preceding it — the point isn’t to declare an all-automated system on day one, but to automate the runbook one proven outcome at a time: finish KYC, then finish leads, then finish renewals; recover dormant buyers, then recover old leads.
The Decision Trace Graph is what all of this writes back to — customer state, pool, context, channel, message, offer, holdout status, cost, response, outcome, and the customer’s next state, every time. This is the actual compounding asset, sitting inside NeoCore’s wider Context Graphs alongside the Customer CG and Product CG. Integrating many tools is the cost of building Progency. The corpus of verified decisions and outcomes is the moat — the record that makes the tenth brand’s first pilot smarter than the first brand’s tenth pilot, because every trace sharpens what the system already knows about which action moves which kind of stuck customer.
Choosing the first pool
A serious brand almost always already has an abandoned-cart flow, and often dynamic retargeting alongside it — which makes cart recovery the most familiar leak, and, for that exact reason, usually the wrong one to lead with. The brand’s current best effort there is already strong, so the incremental lift a holdout can prove is thin. Alpha is widest wherever the brand’s current best effort is weakest, not wherever the leak happens to be most visible.
Incomplete KYC and stuck leads make better first pools precisely because they’re less discussed: the current alternative is often a call centre, a manual queue, or nothing at all, which gives a randomised holdout real room to show a difference. Renewals and form completion follow — clear events, clear value, clear counterfactuals. Dormant-base recovery comes after: strategically the larger prize, but it requires re-earning attention before anything resembling revenue is realistic, which makes it a slower pool to prove first.
None of this is about permanently ignoring cart recovery or dormant reactivation — both matter, and both eventually run. It’s about sequencing the first proof where the brand’s own current effort gives Progency the least competition, so the first holdout comparison is unambiguous rather than marginal.
5
Where Progency Lives
It would be tempting to describe Progency as a third pillar of NeoCore, sitting beside Meridian and Atrium as an independent offering with its own sales motion. That would be the wrong way to think about it — worth saying plainly rather than leaving vague.
Progency isn’t a third engine. It’s the delivery arm of the two engines that already exist.
Finish operates inside Meridian’s domain. Meridian’s job is protecting and maximising the value of Best customers — and a Best customer whose renewal quote goes unanswered, or whose KYC step stalls mid-upgrade, is exactly the leak Meridian is built to prevent. Finish is how that protection gets delivered day to day: MGEs and M-Agents running the specific interventions that stop a valuable, engaged customer’s journey from quietly dying in a queue.
Recover operates inside Atrium’s domain. Atrium exists to push the cost of reactivation and acquisition toward zero for Rest and Next customers. Recover is Atrium’s operating layer for the specific customers who have gone dormant and are drifting toward being re-bought by a paid channel — the same mission as Atrium, delivered as a hands-on managed service rather than a self-serve marketplace mechanic.
One name, one delivery team, two engines it plugs into, two pricing ladders reflecting the two different alternatives each book displaces. That’s a materially different structure than treating Progency as a stand-alone product line — and it has a real consequence: Progency doesn’t need a sales pitch separate from Meridian and Atrium. It’s the answer to “how does this actually get done” for both.
There’s a sequencing question worth being direct about, because the honest answer isn’t the tidy one. Recover — reactivating a customer before a platform re-buys them — is the story that gets a sceptical CMO into the room in the first place. No adtech vendor will ever offer to make itself unnecessary; that’s a structurally unique pitch, and it should stay the headline.
But inside an actual engagement, Finish often ships first. Its cycles are shorter — a stalled KYC form can resolve in weeks, while winning back a genuinely dormant customer takes longer to prove and to fund, since outcome-only pricing means the cost of delivery is carried upfront and collected only once results land. Running Finish first improves how quickly a pilot pays for itself, while the longer Recover proof cycle plays out alongside it. The story that opens the door and the plan that ships first don’t have to be the same thing — and pretending otherwise is how good doctrine quietly drifts.
Put together, the pitch is simple enough to say in one breath: before a known customer is handed to a call centre, an agency, a manual queue, or an adtech platform, give Progency the first right to finish or recover the outcome. You pay only for verified results.
That’s the whole idea. Not a new department. Not a new promise. Just the place in the stack where NeoMarketing’s outcome discipline finally reaches the customers who had been falling through the middle all along.
