For twenty-five years martech has been paid for what it carries. Its next growth engine is getting paid for what it causes, from two pools of money it has never been able to reach.
The argument in one line. Martech companies have billed through two meters, messages sent and users tracked, and both are reaching a ceiling. Three engines take them past it: Premium Sends, which make customer interactions more valuable; Email Ads, which turn earned attention into media; and Alpha, which recovers the dormant database at the vendor’s cost, earns only on verified lift, and puts a floor of six dollars of incremental revenue under every dollar the brand spends. Together they move martech beyond the technology budget and into the media budget and the profit pool.
A note on scope. This essay is written for martech companies that run email and customer engagement infrastructure for brands. Two of the three engines are email businesses, and all three depend on the sending, consent and measurement machinery these vendors already operate. The principles may reach further across martech; this is the segment with the clearest path.
PART I
The Ceiling
1 Two Billing Units Built an Industry
Martech was paid for what it carried, not for what it caused.
For twenty-five years the martech industry has run on two meters. Email and messaging platforms charge per thousand messages sent. Customer engagement platforms charge per monthly active user, monthly tracked user or stored profile. Every innovation of the period, from segmentation and journeys to personalisation, analytics and now AI, was packaged so that it could be billed through one of those two meters.
The meters made sense. They were easy to count, easy to audit and easy to forecast. They grew with the customer: a brand that grew its database grew its user bill, and a brand that sent more grew its messaging bill. Vendor and customer grew together, and nobody had to agree on what any particular message had achieved.
But the meters carried an assumption that was never true: that value is proportional to volume. A reminder that recovers a $90 order and a promotional blast ignored by almost everyone cost the same per thousand. A customer who buys every month and one who has not opened anything in two years cost the same under user pricing, and the dormant one is often still on the bill. The meters measured activity, and activity became the product.
There is a quieter problem underneath. Most efficiencies a vendor delivers reduce the units it bills. Better targeting means fewer sends. Pruning inactive profiles means fewer users. An agent that does the work of five campaigns means fewer campaigns. A martech company that makes its customers more efficient often shrinks its own revenue line. Growth in the customer base can mask the effect, but it does not remove it, and it is no surprise the industry has been slow to do it.
2 Why the Ceiling Is Arriving Now
Martech competes for the smallest budget in marketing, and that budget is getting cheaper to serve.
Four forces are pressing on the two meters at once.
- Production is approaching zero cost. Copy, creative, variants and segments that once took a team a week now take an AI system minutes. When producing a message costs almost nothing, the price per unit of production falls with it.
- User-based pricing is under pressure. Brands are pruning databases, consolidating tools and asking why they pay to track customers who do nothing. AI agents that do the work of marketers compress seats here as they do everywhere else in software.
- The inbox has a second reader. Mailbox providers now summarise, sort and decide what the human sees, the shift I described in The Inbox Has Two Readers. A send no longer guarantees a glance, so a price per send buys less than it used to.
- Investors are repricing software. As AI compresses seats and units, investors are asking which software revenue will last. As I argued in B2B Software Needs a New Revenue Playbook, software companies need revenue tied to outcomes and attention rather than to inputs.
Meanwhile, look at where the money is. In the consumer businesses I meet, paid media is typically several times the technology budget. A large part of that media buys back customers the brand already had: people whose email addresses sit in its own database, whom it stopped engaging, and whom it now pays Google and Meta to reach again. I have called this the reacquisition tax.
That is the irony at the centre of this essay. Martech holds the customer relationships that would make much of that media unnecessary. It is paid from the smallest pool in the marketing budget, while the largest pool flows to the companies that rent attention back to brands.
3 The Crux
Martech can grow faster than its customers’ technology budgets only by earning from money outside them, without becoming adtech and without breaking the brand’s trust.
Richard Rumelt’s test of a good strategy is that it names the crux: the one problem which, once solved, unlocks the rest. For martech, the crux can be put as a question: how does it escape the budget that bought it, without becoming adtech? It is not a product gap. Vendors do not lack features. They lack access to money. As long as their revenue comes only from the technology budget, they can grow no faster than brands are willing to spend on technology, and AI is teaching brands to spend less.
There are three pools of money a martech company can reach (Figure 1).
- The communications budget, which it already serves, and which can be enlarged when email does more and covers more moments.
- The media budget, which belongs to advertisers, and which can be reached when a brand’s email earns attention worth placing an action inside.
- The incremental profit pool, which does not exist until someone creates it, and which can be shared when the vendor takes accountability for a measurable result.
Two constraints come with the crux. The vendor must not become adtech: it does not sell customer data, auction one brand’s customers to another, or own the relationships it operates. And it must not break trust: every new revenue line has to leave the customer’s experience better, or at least no worse, or the attention it depends on disappears.
Figure 1. Three engines, three pools of money. Martech has billed only the first pool. The second has belonged to adtech; the third has belonged to nobody.
Readers of From SEND to EARN will recognise the shape. There I described EARN (Email, Act, Run, Network) as a ladder of provider accountability. Here the same ladder becomes a business model: Email and Act become Premium Sends, Network becomes Email Ads, and Run becomes Alpha.
PART II
Three Engines
4 Engine One: Premium Sends
Email revenue has two levers, the number of valuable sends and the revenue per send, and most vendors pull only one.
The first engine improves the business martech already has. Its arithmetic is simple:
Email revenue = Valuable sends × Revenue per send
Both terms can grow, and they grow for different reasons.
Coverage is the first lever. Consider a bank, an insurer, an airline or a large retailer. Each has hundreds of customer moments: onboarding, verification, activation, statements, renewals, reminders, service updates, feedback, recovery. Many of these sit on SMS, WhatsApp, app notifications or call centres. Some are not handled at all. As the price of paid messaging channels rises, email can win the moments where it carries an action rather than just a notice. The rule is owned first: route each moment to the channel that does the job at the lowest cost, and use paid channels where speed or reach demands them. This is not about sending more email. It is about email holding a larger share of the customer’s interactions with the brand.
Yield is the second lever. A conventional HTML email is a commodity: it informs, and asks the customer to click away to do anything. A NeoMail lets the customer act inside the message, to pay, confirm a renewal, book a slot, answer a question or update a preference, and shows content that is current at the moment of opening rather than the moment of sending. Combined with measured improvements in inbox placement, that is a different product, and it can carry a different price. I have argued in Paid to Send. Paid to Keep. for a premium in the region of 50% on interactive sends, applied only where the interactive version renders; the fallback HTML version is billed at the standard rate.
Two cautions keep this engine sound. Every new moment must be useful and permissioned: sending more email to generate volume accelerates the attention decay that makes every send worth less. And the premium must reflect a measurable gain, such as completions, data captured or cost avoided on another channel, not a new format name. 10+1 Email Innovations sets out what those gains can be.
Premium Sends are the most immediate engine and the least defensible, because interactive capabilities can be copied. But this engine is the foundation. Its revenue is recurring and predictable, it is billed today, and it pays for the two engines that follow.
5 Engine Two: Email Ads
The vendor stops only charging brands to send attention and starts helping them earn from it.
Today a brand pays its vendor to send an email. If the customer opens and engages, the attention that results stays on the brand’s books at a value of zero. Nobody earns from it except, indirectly, the brand, and only if the customer buys.
The Brand Digest changes that. It is a regular, useful edition, with something to read, something to do, a card to collect and Mu to earn, sent between transactions to customers who have nothing to buy today. Customers return because there is something worth opening beyond another offer. And a surface people return to voluntarily is inventory.
The unit of that inventory is the ActionAd: one clearly labelled partner action completed inside the email, such as a sign-up, a quote, a sample or a booking. The advertiser pays per completed action. The number that matters is the yield: net revenue per thousand verified engaged editions, after third-party fees and before the split between the publishing brand and the vendor. That net is shared, 50:50 in the model I use, so a yield of $12 per thousand editions means $6 for the vendor. The vendor’s revenue is its share of the net, never the advertiser’s gross spend.
Yield is not a given. It has to be built from demand, and it is the product of three numbers: how often an edition carries a paid action (fill), how often a reader completes it (action rate), and what the advertiser pays per action. At 50% fill, a 1.5% action rate and $1.50 per action, a thousand engaged editions earn about $11 gross and $10 after third-party costs. Each of those numbers improves with scale. An advertiser who needs a million qualified readers will not buy across five Digests but will across two hundred, so fill rises. Matching across more categories puts the right action in front of the right reader, so action rates rise. More advertisers competing for the same actions raise the price. Modest gains in each, to 65% fill, a 1.6% action rate and $1.60 per action, take the yield to about $15. That is why this engine is a network business: it is worth more at three hundred publishing brands than at fifty, and may be worth little at five.
The media business should start with engaged audiences. A brand that already has customers opening its email can run a Digest from the first month, and the inventory exists from the first edition. The fastest start may be with media companies, which already have advertisers, ad sales teams and readers who open every day; brand Digests follow. Dormant recovery adds inventory later, but the media engine does not need to wait for it.
Across many brands, NeoNet becomes the cooperative layer: matching advertisers to Digests, managing frequency and category exclusions, and pooling demand so that an advertiser can buy at a scale no single brand offers. Its sharpest product may be CRM Ads, which I described in CRM Ads: Your Customer, Someone Else’s Attention: a brand reaching its own customers inside another brand’s Digest when its own emails are going unread. That is neither owned media nor rented media. It is friendly media: a customer the brand already knows, reached through a relationship the customer trusts, with an outcome the brand can measure.
Three consequences follow. Brands become publishers without becoming media companies: they do not hire editors, build ad servers or employ an ad sales team. Email can become self-funding, because a Digest earning enough from ActionAds pays for its own production. And the vendor reaches the media budget for the first time, the shift I described in From Messages to Media.
The danger is equally clear. A Digest stuffed with irrelevant advertising is just another promotional email, and it will be ignored like one. The discipline has to be built in: at most one partner action per edition, category exclusions the brand controls, and an automatic stop when complaint rates rise. The metric that matters is net advertiser revenue per thousand verified engaged editions, read alongside reader retention. Some categories, banking among them, will exclude third-party advertising altogether, and the engine has to work without them.
6 Engine Three: Alpha
“Give us the customers you have stopped operating. We fund their recovery, we earn only on verified lift, and every dollar you spend returns at least six.”
Every brand database has a dormant half: customers who have not opened or bought in a year or more. The brand stopped sending to them for good reasons. They hurt deliverability, there was no budget to engage them, and there was nothing new to say. Then the acquisition team went to Google and Meta to find customers, and paid to reach some of the same people again. The Dormant Goldmine describes that database as an unoperated asset rather than a dead one.
The third engine takes responsibility for that dormant half. The brand grants the vendor operating rights, not ownership, over its eligible dormant customers, under clear consent and privacy rules. The vendor funds the recovery. The brand pays only when recovery produces incremental transactions.
The operating sequence matters more than any single tactic (Figure 2). Eligibility comes first: consent, suppression and list validation, so that recovery never damages the brand’s active programme. Then attention: short Digests, interactive Magnets, cards and Mu, built to earn a first verified human action, because a dormant customer asked to buy on the first contact simply stays dormant. Then habit: regular Digests for those who respond. Only then conversion, where agents choose whom to contact, what to show, when to make an offer and when to stop. The last step is verification: transactions are counted only against a concurrent randomised holdout.
Figure 2. The recovery sequence. The vendor funds the first four steps. The brand funds only incentives agreed in advance, and pays Carry on the fifth.
Agents appear here deliberately, and not as a fourth revenue line. Many vendors will price agents as a surcharge, per seat, per user or per task. That is the old meter with a new name. In this model the agents are the machinery that makes Alpha cheap enough to guarantee. Their benefit shows up as the vendor’s margin on the commitment, not as another line on the brand’s invoice.
What ROAS 6 means
ROAS, return on ad spend, is the language performance marketers already speak, which is why the commitment is expressed in it. But a ROAS promise can be empty in two ways, and the contract has to close both.
The commitment, in three parts.
No upfront cost. The vendor funds the sends, the content, the cards, the Mu and the agents.
No lift, no bill. The vendor earns Carry only on incremental revenue verified against a concurrent randomised holdout, over a 90-day window for each recovered cohort.
A 6× floor on the brand’s money. Everything the brand spends on the programme, meaning the vendor’s fee plus any incentives the brand funds, returns at least six times in incremental revenue. If the return falls short, the vendor reduces its fee, to zero if necessary, and then reimburses brand-funded incentives up to an agreed ceiling.
The first way a ROAS promise can be empty is arithmetic. If the fee is simply set at one-sixth of incremental revenue, ROAS is six by construction and the vendor has committed to nothing. So the fee has to be the residual: Carry is capped at one-sixth of incremental revenue minus whatever the brand spent on incentives. On $100 of incremental revenue with $5 of brand-funded discounts, the most the vendor can bill is $11.67.
The second is that waiving the fee is not always enough. If the brand spends $20 on discounts to produce the same $100, the return is 5× even with no fee at all, and the vendor has to reimburse $3.33 of the brand’s costs to honour the floor. That is why the contract fixes the incentive budget in advance, sets a minimum cohort size so that the holdout can be read with confidence, and caps the vendor’s make-good. A vendor that only waives its fee is offering a 6× pricing hurdle, and should call it that. A vendor that also makes good on the brand’s costs is offering a guarantee.
The floor protects the brand’s money; it does not promise volume. Volume is the vendor’s own incentive, since it earns nothing on customers it fails to recover and carries the cost of trying. The 90-day window keeps the claim bounded and avoids disputes with the brand’s own retention team over purchases a year later. The holdout keeps it clean: customers who would have returned anyway belong to the baseline. In the language of No Lift, No Bill, that baseline is Beta, the verified lift is Alpha, and the vendor’s share is Carry.
Figure 3 shows the commitment on $100 of incremental revenue for a brand with a 40% contribution margin that funds $5 of incentives. The brand earns $40 of contribution, spends $5 on incentives and $11.67 on the fee, and keeps $23.33 on revenue it would not otherwise have had. A comparison with paid media means something only on the same basis: the same customers, the same holdout, the same margin. If buying those customers back through paid media achieved a 4× incremental return, the brand would keep $15. Platform-reported ROAS usually runs higher than incremental ROAS, so the real comparison is rarely kinder to media than this. The way to settle it for any brand is a three-arm test: a control group, a group reached through paid media, and a group recovered through the programme.
Figure 3. A matched, hypothetical comparison. Both routes are measured as incremental revenue against the same holdout; the paid-media return is illustrative, not a benchmark.
Margin matters. For a retailer with a 20% contribution margin, spending a sixth of incremental revenue on fee and incentives leaves little behind. For low-margin categories the same floor should be expressed to the CFO as a share of incremental contribution, while the CMO still hears ROAS. And “recover for free” means the brand pays nothing upfront, not that recovery costs nothing. The vendor carries the cost of attention, content and operations before a single fee is earned. That is why this engine needs the other two, and why it needs a portfolio.
The dormant database is the wedge, not the limit. Once a vendor can price verified incremental recovery, the same machinery and the same measurement can protect customers who are fading before they go dormant, and grow one-time buyers into repeat buyers. Recovery is where outcome underwriting starts; it is not where it ends.
PART III
One Relationship, Three Ledgers
7 How the Engines Combine
One customer relationship can produce three legitimate kinds of revenue over its life, provided each has its own ledger.
The engines are not assigned one per customer. They are layers of monetisation applied across the customer’s lifecycle (Figure 4). An engaged customer generates Premium Sends revenue and, through Digests, advertising revenue. A dormant customer generates no routine sends; recovery is funded by the vendor and paid for through Alpha, and advertising begins once attention returns. A recovered customer generates Alpha within the 90-day window, Digest inventory as a reader, and Premium Sends revenue once they graduate back to the brand’s active list.
Figure 4. Three engines across the customer lifecycle. No customer is forced into all three, and no outcome is counted twice.
That last step closes a loop. A recovered customer becomes an active customer again, and an active customer is billed through the first engine. Alpha does not just earn a fee; it grows the base on which Premium Sends are billed.
The safeguard is that the same economic outcome is never counted twice. An advertiser-funded action belongs to the advertiser’s ledger. A transaction counted as Alpha cannot also be used to justify a send premium. A send to a customer inside a recovery programme is not billed at all, because the vendor is already paid through Carry. Three meters can coexist on one relationship only if their claims do not overlap.
The funding hypothesis
The most interesting possibility is that the second engine helps finance the third. Conventional recovery spends money on attention and hopes for a transaction. Here, a recovered reader earns advertising revenue before they are ready to buy again. If that revenue is material, it lowers the net cost of recovery, and a lower cost of recovery makes the 6× floor affordable in more categories. This is the potential innovation: making attention profitable before the customer transacts.
It is a hypothesis, not an assumption. Ad sales costs, low early engagement, deliverability limits and thin initial yields may overwhelm the revenue. The first financial experiment has to measure precisely whether the offset is material. If it is not, Alpha still stands on its own; it simply costs more to underwrite.
8 Why Martech, and Not Adtech or Agencies
The advantage is not a feature. It is a position across thousands of brands’ owned customer relationships at once.
If these three engines are valuable, someone will build them. The question is who is best placed.
Each candidate has real strengths. The largest advertising platforms have vast first-party relationships of their own, but with consumers inside their own products, not inside a brand’s relationship with its customers. Agencies know their clients’ strategy and can build infrastructure or partner for it. Brands own their customer relationships outright. But a single brand rarely has the scale: running a dormant database, producing a Digest every week and selling its inventory are uneconomic for one brand and economic for a platform that does it for a hundred.
A martech company already sends for hundreds or thousands of brands. It holds integrations, authorised access to engagement infrastructure, deliverability expertise and consent systems. Those are delegated rights, not ownership. They do not by themselves permit sharing data, matching identities across brands or running third-party advertising; each of those needs the brand’s explicit permission and the customer’s consent. What the position gives the vendor is a head start.
The head start becomes an advantage only if the portfolio turns into a shared operating factory. This is the logic I drew from Bending Spoons in The Dormant Goldmine: a shared platform makes each additional asset cheaper to run than the last. In recovery the test is concrete. If the vendor’s cost of running a programme does not fall from something like $120K a year for the first brands to about $70K at portfolio scale, the economics of the floor do not close. The analogy also carries a warning: the operator runs the asset and never owns it, and an operator that over-monetises a recovered customer loses that customer a second time.
The factory also accumulates memory. Every programme records whom it was eligible to contact, what it did, what the holdout did, what it cost and what it produced. That record of decisions and outcomes, showing which customers respond to what, in which category and at which point of dormancy, is the part a competitor cannot copy by building the same software. The model is not the moat. The traces are.
The portfolio also changes the risk. One Alpha contract is volatile: a category slump, a bad season or a weak offer can wipe out its Carry. Two hundred contracts behave like a book, in which variance across brands averages out. The same is true of advertising demand: pooled across many brands’ Digests, it reaches a scale no single publisher can offer. The engines that look risky one brand at a time become steadier across a portfolio. That does not turn performance fees into recurring revenue, and a downturn that hits every category at once still hits the whole book. But only a company that already serves the portfolio can assemble it.
9 The Rule That Holds the Portfolio Together
Optimise incremental customer profit and retained attention, never a single revenue line.
Assigning engines by customer state reduces the conflicts between them. It does not remove them. Each engine rewards a different behaviour, and four conflicts need naming.
- Sends reward volume. A vendor paid per send benefits from more sends, even when they erode attention. The rule: frequency follows engagement, and unresponsive customers are suppressed from routine sends and moved into recovery.
- Ads reward ad load. More placements raise short-term yield and damage the habit that produces it. The rule: one partner action per edition, with reader retention as the metric that gates any increase.
- Agents reward conversion. An agent chasing transactions can buy them with discounts that destroy margin. The rule: Alpha is measured on contribution for low-margin brands, and brand-funded discounts sit inside the ROAS denominator, so a discount-led recovery fails its own test.
- Coverage cannibalises. Moving a service moment from a paid messaging channel to email may grow email revenue while shrinking the vendor’s own messaging revenue. The rule: route to the channel that does the job. A vendor that will not cannibalise itself will be cannibalised by someone else.
All four rules rest on the same apparatus: a shared ledger that records every action and its outcome, and concurrent holdouts that separate what the vendor caused from what would have happened anyway. A vendor running three engines needs one measure of success: whether the brand’s customers are worth more, and pay more attention, than they would have without it.
In practice that means one portfolio review each quarter, with four numbers on the table: incremental customer profit, retained attention, the vendor’s contribution by engine, and any incident of consent or deliverability.
PART IV
The New P&L
10 Three Businesses, Three Sales Motions
The engines share technology and customers. They do not share a go-to-market.
Premium Sends, Email Ads and Alpha are not three products on one price list. They are three businesses with different buyers, different budgets and different economics (Table 1).
Table 1. Three engines, three businesses.
| Engine | Buyer | Budget | Sales motion | Revenue character |
| Premium Sends | CRM and email team | Communications | Account management, as today | Recurring, usage-based |
| Email Ads | Advertisers’ media buyers | Media | Ad sales and yield management: a new capability | Variable, network-driven |
| Alpha | Growth head and CFO | Incremental profit | Commercial negotiation and holdout design | Performance-based, paid in arrears |
The products can share a platform. The company still needs new muscles: ad sales and editorial for the second engine, and recovery operators, measurement scientists and commercial structurers for the third. A martech CEO reading this should conclude that the obstacle is not the product roadmap. It is the organisation chart.
The finance function changes too. Premium Sends are billed monthly. Advertising revenue depends on demand and yield, and arrives when advertisers pay. Alpha is funded upfront, with the vendor carrying the cost of recovery for months, and paid in arrears, after the 90-day window closes and the holdout is read. The three engines need separate working-capital plans even when they share customers.
Investors will value these revenues differently. Variable media revenue and performance fees do not earn the multiple of recurring subscription revenue, and the answer is not to pretend they do. It is to show contribution margin by engine, repeatability across brands, diversification across the portfolio, and an advantage that grows as the portfolio grows.
One consequence follows that the largest vendors will not enjoy. A company that reports recurring revenue to public markets every quarter will find it hard to fund recovery upfront and wait to be paid. A profitable, founder-controlled company can make that trade more easily. The first vendors to build all three engines may not be the biggest ones.
11 An Illustrative Three-Year Transformation
A vendor growing at 7% on messages and MAUs can roughly double its growth rate by adding two engines, and the engine with the most upside is also the one that costs most to run.
To make this concrete, consider an illustrative martech company with $100M of revenue, 45% of it from email sends and the rest from engagement platforms priced per user. Its legacy business grows at 7% a year. Table 2 sets out a base case and an upside case. Both count only the vendor’s net revenue: never advertisers’ gross spend, and never the brand’s recovered revenue. Alpha fees are modelled as realised Carry after brand-funded incentives, not as a flat one-sixth of incremental revenue.
Table 2. Illustrative assumptions for a $100M martech company.
| Assumption | Base, Year 1 | Base, Year 2 | Base, Year 3 | Upside, Year 3 |
| Premium Sends: additional valuable moments, as a share of email sends | 4% | 8% | 12% | 15% |
| Premium Sends: sends upgraded to interactive, at a 50% premium | 8% | 15% | 20% | 25% |
| Email Ads: brands running Digests, each with 100,000 engaged readers | 40 | 120 | 240 | 300 |
| Email Ads: verified engaged editions per reader per year | 36 | 36 | 36 | 36 |
| Email Ads: yield per thousand engaged editions, before the 50:50 split | $10 | $10 | $10 | $15 |
| Alpha: recovery programmes, each with 1M dormant customers | 15 (half year) | 50 | 110 | 150 |
| Alpha: incremental revenue per programme per year | $1.0M | $1.0M | $1.0M | $1.2M |
| Alpha: brand-funded incentives, as a share of incremental revenue | 5% | 5% | 5% | 5% |
| Alpha: vendor fee per programme, the maximum consistent with 6× | $117K | $117K | $117K | $140K |
In the base case, revenue grows from $100M to about $152M in three years (Figure 5), a compound rate of 15%, roughly double the legacy rate. Growth rises from 12% in the first year to 17% in the third as the engines move from pilots to portfolios. By year three the new engines contribute about $30M, a fifth of revenue.
Figure 5. Revenue by engine, base case, years 0 to 3, with the upside-case total marked. Vendor net revenue only.
The bridge in Figure 6 shows the base case by engine. Premium Sends and Alpha each add about $13M in year three; Email Ads add $4.3M. Premium Sends get there with little risk, half from new moments and half from higher yield. Alpha gets there from 110 programmes earning about $117K each.
Figure 6. From $100M to $152M in the base case. Premium Sends and Alpha contribute equally; Email Ads matter more than their size suggests, because they can help fund recovery.
The Alpha number rests on one hypothesis that has to be tested, not assumed. A programme with a million dormant customers earning $1M of incremental revenue at a $100 average order needs 10,000 purchases that would not otherwise have happened: an incremental conversion of 1%. That may be achievable in categories with natural repeat purchase and out of reach in others. It is the first number the pilots must read.
The model is also sensitive to how recovery is bought. With no brand-funded incentives, the same 110 programmes would earn $18.3M; at 5% of incremental revenue, $12.8M; at 10%, $7.3M. A discount-led recovery shrinks the vendor’s fee before it shrinks anything else, which is the behaviour the 6× floor is designed to produce.
Table 3. Three scenarios for year three.
| Scenario | Year-3 revenue | Three-year growth rate | New engines, share of revenue | What it assumes |
| Stress | $137.5M | 11.2% | 11% | Each new engine delivers half the base case |
| Base | $152.4M | 15.1% | 20% | Table 2, base columns |
| Upside | $167.8M | 18.8% | 27% | Table 2, upside column; ad yield rises with scale |
Even the stress case, in which every new engine delivers half its base-case revenue, adds four points to the legacy growth rate. The upside case adds nearly twelve, and more than half of the gap between base and upside comes from Alpha.
The ranking changes again when the measure is capital efficiency. Premium Sends need almost no upfront investment and carry the highest margin, because they are billed on infrastructure the company already runs. Email Ads need editorial capacity and an ad sales team before the yield arrives. Alpha needs the most capital: at $100K or more per programme in the early years, mostly spent in the first months, the 50 programmes of year two require about $5M before most of their fees are collected. And at a base-case fee of $117K, a programme earns a real margin only once the portfolio drives its cost towards $70K. The engine with the most upside needs the strongest balance sheet and the most operating discipline.
The point of the illustration is not the precise numbers. It is that a mature martech company does not need to abandon its platform to change its growth rate. It needs to add two businesses on top of the one it has, fund them from it, and prove the one number that decides the rest.
PART V
Tests
12 What Would Prove This Wrong
Each engine carries its own kill condition, and each must be tested before the next is scaled.
An argument that cannot fail is not a strategy. These are the conditions under which I would conclude that this one has failed (Table 4).
Table 4. Kill conditions, engine by engine.
| Test | Kill condition | What it would mean |
| Premium Sends | Fewer than half of pilot brands renew interactive sends at a premium of at least 30% after twelve months | Interactivity is a feature, not a revenue line; the first engine reverts to volume |
| Email Ads | Net yield fails to cover Digest production and ad sales at a scale of 100 brands, or reader retention falls in Digests that carry ads | Media cannot fund attention; Alpha must carry recovery costs alone |
| Alpha: the brand | Fewer than half of the first 20 programmes clear 6× on all brand spend against the holdout | Most recovery would have happened anyway, or costs too much to guarantee |
| Alpha: the vendor | After two years at portfolio scale, the vendor’s contribution on Alpha stays below 25% of fees | The floor is affordable only by losing money; the factory has not formed |
| Recovery operations | Recovery sends breach mailbox-provider complaint thresholds, or the brand’s active domain loses more than two percentage points of inbox placement for a sustained period | Recovery damages the active programme; the sequence needs redesign |
| Organisation | Within 18 months, the company cannot build ad sales and recovery teams alongside its platform business | The engines need separate companies, not separate product lines |
Three of these tests matter more than the rest. The two Alpha tests decide whether the engine with the most upside exists, and whether the vendor can afford it. The yield test decides whether media can help pay for it. All three can be read within two years, from a few dozen brands.
If the engines pass, the result restates the Three NEVERs from the vendor’s side of the table. Useful sends, Digest attention and timely recovery mean a brand need never lose customers it could have kept. Rebuilding owned attention before reaching for paid media means it need never pay twice to reach customers it already has. And the Alpha contract, with its 6× floor, means it need never pay fixed for an outcome it cannot see.
For twenty-five years, martech companies competed for the technology budget by finding more messages to send and more customers to track. Those meters built an industry, and now they cap it. The vendors that grow past the cap will be the ones that make customer interactions worth more, turn the attention they earn into media, and take responsibility for the profit they create.
Adtech gets paid when brands need attention. Martech can get paid for making sure they need to rent less of it.
**
Key points
- Martech has billed through two meters, messages sent and users tracked, and AI, inbox agents and pricing pressure are capping both.
- The crux is access to money: vendors can outgrow their customers’ technology budgets only by earning from the media budget and the incremental profit pool.
- Premium Sends grow both the number of valuable sends and the revenue per send. The engine is recurring and immediate, but copyable.
- Email Ads turn Brand Digests into inventory for ActionAds, starting with engaged audiences. Yield is fill × action rate × price per action, and it improves with the scale of the network.
- Alpha recovers dormant customers at the vendor’s cost in three commitments: no upfront cost, no lift no bill, and a 6× floor on everything the brand spends, backed by a capped make-good.
- The engines are layers of monetisation across the customer lifecycle, with separate ledgers, held together by one rule: optimise incremental customer profit and retained attention.
- Martech’s portfolio across thousands of brands is a head start, not a moat. It becomes one when a shared factory lowers the cost of each programme and accumulates the traces of what worked.
- An illustrative $100M vendor growing at 7% reaches about 15% a year in the base case, 19% in the upside and 11% in a stress case. Alpha carries the most upside and needs the most capital.
