Published July 24, 2026
1
The Trap: Why Sending Alone Cannot Save Email
How email service providers escape commoditisation by moving from delivery revenue to actions, outcomes and inbox media.
Email did not lose relevance. The companies that sell it lost imagination. For twenty-five years, email service providers built businesses around a single verb — send — and were rewarded for doing it reliably, at scale, with deliverability, routing, templates and reporting. That was not a mistake; it was exactly what the market needed. But the thing you are paid for is the thing you optimise, and an industry paid by the send spent a quarter of a century perfecting delivery while the email itself barely changed. The result is a category that is now judged by the very logic it taught the market to apply. This essay is in three parts: the trap, the ladder, and the business of climbing it.
The old bargain, and the ceiling it built
The original bargain was sound. Brands had databases they could not operate at scale and a channel — the inbox — that was the cheapest owned ground they possessed. They needed lists cleaned, domains protected, campaigns scheduled, events tracked, bounces suppressed, complaints monitored and messages delivered without breaking reputation or compliance. The email service provider became the operating layer for that channel, and it earned its keep. Email became the workhorse of digital retention because it combined three advantages almost nothing else could match: a known identity, a near-zero marginal cost, and genuine ownership — a brand could reach its own customer without renting an audience from a platform.
The business model followed the job. ESPs were paid for contacts, sends, volume and platform access; the input became the invoice, and the invoice quietly shaped the product. But the category made one consequential error: it confused the channel’s value with the provider’s value. The brand owned the customer. The mailbox provider — Gmail, Apple Mail, Outlook — owned the client software. The ESP owned only the sending system in the narrow middle. That position built a durable business and, at the same time, a strategic ceiling: the vendor could move the message but not easily change what the message was, could optimise the sending but not own the moment of opening, could report engagement but never guarantee a profit.
So ESPs did what infrastructure companies do — they made the infrastructure better. Faster sending, better routing, better templates, better deliverability monitoring, better APIs, better dashboards. All useful, all necessary, and all increasingly comparable across vendors. The better an ESP became at delivery, the more invisible it became — and invisible infrastructure is eventually priced like infrastructure. The ceiling was not a failure of execution. It was built into the position the category chose to occupy.
Five reasons the category under-imagined itself
Email did not become a commodity because email failed. It became a commodity because the vendor model under-imagined what email could become — for five reasons at once, each of which now points at its own way out.
The first was pricing. Paid on sends, ESPs optimised the world around sending: more contacts, more journeys, more triggered messages. Rational, and beside the point. The deepest customer problem was never “can this be sent?” but “is this worth opening, and will it cause the next profitable action?” A business priced on sends will never build the email that makes sends matter less.
The second was surface ownership. An email renders inside someone else’s client, so the ESP never owned the canvas the way an ad platform owns its unit. Concluding — correctly — that they could not control the client, vendors wrongly concluded they could not change the artefact, and improved the machinery around the email while the email stood still.
The third was the all-or-nothing mistake on interactivity. When interactive email arrived, support across clients was uneven, so the industry treated it as a campaign trick rather than a design principle. The better conclusion was available and never drawn: interactivity is one rendering path, not the whole strategy; the real job is to compose the best possible experience at open, with graceful fallback everywhere else.
The fourth was measurement. When open-rate reliability broke, the industry lost a familiar instrument and read the darkness as decline, instead of rebuilding around stronger signals — clicks, actions, replies, sessions, declared intent, transaction movement. The channel still held attention; the dashboard simply could no longer see it.
The fifth was cost. A genuinely useful email is not a template with a name inserted; it is a fresh decision made for a specific reader at a specific moment — what is true now, what to show now, what action to allow now, what to remember afterwards. For most of email’s history, composing that at scale was simply too expensive. A brand could handcraft one clever campaign, not operate millions of living messages a day. Put the five together and the verdict “email is tired” was a misdiagnosis: the patient was fine, the thermometer was broken, the treatment had not been invented, and the people who could have invented it were paid to do something else.
The commodity spiral
The trap has a cruel mechanism: excellence at the old job accelerates commoditisation. Deliverability, scale, compliance, routing, security, support — all of it matters, and all of it becomes harder to monetise the moment the buyer believes several vendors clear an acceptable bar. Then the conversation moves from value to benchmark. Procurement enters. Vendor diversification becomes policy. The customer asks for lower unit cost, more volume, more resilience and less dependency — and every one of those requests is reasonable.
Figure 1 — A single-rung business, priced on the input, is squeezed on the input. The pipe must be run well; it cannot be asked to carry the future margin.
The danger is that this position is comfortable for a long time. Revenue continues, renewals continue, campaigns and support tickets continue — and the category quietly loses altitude. The vendor becomes operationally important while becoming strategically replaceable: a line item to be optimised rather than a partner to be expanded. A company that sells only delivery will eventually be priced by delivery. Defending the pipe harder does not arrest the spiral; it deepens it, because every incremental improvement to an invisible utility is, by definition, hard to charge for. The escape cannot be found on the rung where the trap was built.
Why “more interactivity” is not the escape
The obvious response is to make email more interactive — more AMP, better templates, more widgets, forms and calculators. That is directionally right and strategically incomplete, because it confuses a capability with a business model. If interactivity is sold as one-time development, it stays a services line; if an embedded calculator is sold like a campaign asset, it stays a cost; if a living digest is sold as a template upgrade, it stays inside the old budget. The artefact becomes modern while the economics remain ancient.
The distinction is subtle and decisive. Interactivity as a feature says: pay us to build a better email. Interactivity as an action surface says: use the email to capture intent, complete actions, move customers and prove incremental value. The same artefact can sit in either model. A broker’s in-email application flow can be a paid development project, or an outcome instrument measured against a control. A retailer’s replenishment email can be a clever template, or a repeat-purchase engine. A publisher’s digest can be content, or monetisable inventory. The question is never what the email contains; it is what the vendor is paid for. Adding features to a per-send contract produces a more expensive pipe, not a new business — which is why a decade of “do more AMP” has not moved the category’s economics an inch.
Why now — three unlocks converge
If the diagnosis is twenty-five years old, why act now? Because three constraints that held the old model in place have broken at roughly the same moment, and their convergence is the opening the category has been waiting for.
Figure 2 — Three long-standing constraints break at once, and converge on a single opening: a living email, paid on what it proves.
The first unlock is artificial intelligence. The reason a living email was never operated at scale was cost: composing a fresh, relevant message per reader at the moment of opening could not be done economically. AI does not magically save email, but it changes what is cheap enough to attempt — the old email was written at send and guessed what would matter; the new email can be assembled at open and check what is true.
The second is the measurement reset. The collapse of the open rate, which once looked like a loss, is in fact the forcing function: with the old vanity metric gone, the only credible thing left to measure is action and lift against a control — exactly the basis an outcome business needs. The instrument that broke was the one keeping the category honest about the wrong thing.
The third is the rising cost of rented attention. As paid channels became more expensive and less certain, the economics of re-buying a customer you already own turned from wasteful to indefensible, and the owned inbox — identity-linked, low-cost, permissioned — became the obvious place to recover and retain rather than re-acquire. None of these three would be sufficient alone.
Together they make a living, accountable, owned-attention business not only possible but overdue. The conditions that made the old model rational have expired; the conditions that make the new one rational have arrived.
The surface is an asset
Step back and the reframe is simple. The surface was never the product. The surface is an asset. An owned email relationship has four properties that make it far too valuable to remain trapped inside per-send economics: it is identity-linked, it is low-cost, it is repeatable, and it sits in a place the customer returns to, reads, decides and acts. Unlike a paid impression it is not rented for a moment and gone; unlike a notification it can hold content, context, memory and choice; unlike a landing page it begins from a known relationship. Delivery is merely the first way to monetise that asset — and the category mistook the first way for the only way.
The next model for ESPs is therefore not another feature bundle but a migration from SEND to EARN. EARN stands for Email, Act, Run, Network. Email names the owned surface and the infrastructure that delivers it; Act makes that surface useful and interactive; Run takes responsibility for outcomes on it; Network turns trusted attention into media and cooperative acquisition. There is a deliberate symmetry with the framework brands already use. SNR — Sell, Notify, Relate — is the brand’s grammar for what an email should do. EARN is the vendor’s model for how the same surface gets paid. One describes the message; the other describes the business. They are two views of the same owned attention, seen from opposite sides of the table — and the only question that matters for Part 2 is this: if the future ESP is not paid primarily per send, then what is it paid for?
**
The old email business earned from volume. The next one earns from value — and the difference is not a feature, it is a business model.
2
The Ladder: How the Email Business Climbs
EARN is a business-model migration, not a product roadmap, and the distinction is the whole point. Product roadmaps list things to build; business-model migrations change the unit of value. Each rung of EARN changes the buyer, the competitor, the pricing logic and — the word that matters most — the accountability the vendor is willing to take. The same owned surface remains underneath the entire way up. What rises is what the vendor is trusted, and paid, to do.
Figure 3 — The EARN ladder. One owned surface, four ways to be paid; value and accountability rising with each rung. The climb is the strategy.
E Email — the infrastructure rung
Email is the floor: sending, routing, deliverability, reporting, compliance, APIs, rendering, suppression, authentication, reputation and operational support. Every serious vendor must run it well, because if this layer fails nothing above it matters. The buyer is procurement or marketing operations; the competitor is another ESP, an internal sending system or a cheaper delivery vendor; the pricing is input-led — per send, committed volume, platform access.
The strategic instruction here is counterintuitive and easy to get wrong: make Send efficient, and do not expect it to carry the future margin. There is no honour in pretending that sending is not infrastructure; it is. But infrastructure is not unimportant — roads are infrastructure, and everything travels on them. The error is not treating Send as essential; the error is worshipping it. The point of the Email rung is not premium pricing forever. It is to hold the owned surface through which the higher-value models can emerge, because an ESP that loses the send relationship usually loses the surface, the signals, the habit and the right to propose anything above it. So the pipe must be defended and run beautifully — and then deliberately treated as the cash engine and the distribution layer, not the destination.
A Act — the capability rung
One floor up, the email stops being a message and becomes a surface on which the customer can do something. It becomes live, current and able to remember: a digest assembled at open, a calculator personalised to the reader, a preference fork, a survey, a product selector, a renewal option, a claim status, a booking flow, a consent request, an intent signal. The point is not that every email becomes an app — it is that the inbox can now contain actions, not merely links to actions.
The buyer changes entirely: this is a marketing, growth and product conversation, not a procurement one. So does the competitor — here the vendor is up against agencies, dev shops, campaign studios, AMP specialists and the inertia of doing nothing, never another ESP. And so the pricing must change with it: a capability fee, a managed-innovation programme, a zero-development-cost pilot with upside participation, or a hybrid — anything but a return to per-send. The value is not the number of emails delivered; it is that the email can now capture an action, signal or preference that previously demanded a click-out, a login or a separate app session.
But Act has one discipline of its own: it is the on-ramp, not the destination. Its job is adoption and proof, not full outcome risk from day one. A new interactive surface usually needs to demonstrate that people engage with it, return to it and trust it before it can carry a revenue guarantee. A vendor that treats Act as the summit will over-invest in a thin-margin tier and call it transformation; a vendor that treats it as the path will use it to generate the evidence the next rung is priced on. Act helps the brand do more. Run takes responsibility for the result.
R Run — the outcome rung
Run is where the model changes character. Here the vendor stops merely enabling the brand and begins operating for a result: recover a dormant customer, restart a relationship, drive a second purchase, reactivate a subscriber, convert a declared intent, bring a lapsing buyer back before paid media has to. The buyer is the CMO and, increasingly, the CFO, because the conversation is no longer about campaign performance — it is about customer economics, about money that would not otherwise have appeared. The competitor is no longer an ESP or even an agency; it is the paid channels a brand reaches for when its owned attention runs out.
The pricing follows the accountability. It is not input-led; it is based on verified lift — a defined cohort, a concurrent, randomised control group rather than a prior-period baseline, an attributable outcome, and a payout only on what was added above what would have happened anyway. This is where the lost margin returns, because proven incremental value is the one thing a commodity pipe can never be. A brand does not need another dashboard confirming an email was sent and clicked; it needs to know whether a customer likely to be lost was recovered without paying to win them back, whether an in-email action created revenue rather than merely harvesting demand that would have arrived anyway, whether attention was rebuilt rather than spent.
Run is a fundamentally different business from selling software access, and it is more demanding. It requires operators, measurement discipline, creative judgement, experimentation and commercial courage; it will look services-shaped before it becomes a repeatable system, and that is acceptable — most outcome businesses begin as expert operations. The non-negotiable is honesty of measurement. Without a control group, every outcome claim is attribution theatre; with one, the vendor can say here is the baseline, here is the intervention, here is the lift, here is the payment. That sentence is the bridge from a marketing promise to a finance-grade fact — and it is the rung where an ESP becomes an Email Alpha company.
N Network — the media rung
Network is the top rung, and it arrives last because it must be earned. A brand’s inbox attention is valuable only while the customer keeps trusting it, and that trust is not created by inserting advertising into every available slot — it is created by making the emails useful enough that people keep opening them. Only a surface that has earned attention can become media.
When that condition is met, the surface becomes inventory: first for the brand’s own offers, then for carefully governed partner demand, and eventually as a cooperative network in which one brand’s earned attention can help another recover or acquire a customer in a permissioned, brand-safe way. The buyer changes again — partnerships, media, advertisers — and the competitor is not an ESP at all but retail media, commerce media and ad networks. The pricing is revenue-share, yield and media economics. But the sequencing is the discipline: first-party before third-party; utility before monetisation; trust before inventory; relevance before scale. A broker uses action modules for its own products before it carries anyone else’s; a retailer moves its own customers across categories before it sells a slot; a publisher serves its own subscription goals first. Network depends on everything below it — without Email there is no surface, without Act no interaction, without Run no proof that attention converts — which is exactly why it is the horizon and not the opening move. It is what an ESP becomes when it stops being a sender and becomes a marketplace for owned attention.
Where ActionAds belong — a bridge, not a rung
The most common confusion is where in-mail action units sit, and the answer is that they are a bridge across the ladder rather than a rung of their own. A first-party action unit — apply, renew, calculate, sample, upgrade, restart, declare intent — lives inside the brand’s own funnel: sold as a capability it belongs to Act; operated against a control and paid on lift it belongs to Run. The artefact has not changed; only the commercial treatment has.
Partner inventory is different. The moment a unit carries an outside advertiser or a complementary brand, it begins to become media, and it belongs to Network — provided the host brand keeps control of category, frequency, relevance and exclusions. The cooperative network is the endgame: many brands operating trusted surfaces, each able to carry relevant action units without degrading engagement, with the vendor coordinating demand, recovery and acquisition across them. The rule is short enough to remember: first-party proves value, partner inventory creates media, the network creates the marketplace.
The two laws that make EARN a strategy
A ladder of revenue models is only a menu unless two laws hold it together.
The first: a thing’s rung is set by how you sell it, not by what it is. The same living email can be Act or Run. Sold as a build, a capability or a managed experience, it is Act. Tied to a defined cohort, measured against a control and paid on attributable lift, it is Run. The artefact did not change; the commercial model did. The discriminator is the counterfactual: a clean baseline and payment on lift makes it Run; the absence of one makes it Act. And on any single audience you charge for the capability or you take a share of the outcome — never both, because no brand will tolerate paying twice for the same value.
Figure 4 — The same living email is Run or Act depending only on whether a clean counterfactual exists.
The second law: the rungs only compound if each one graduates customers to the next. Email funds Act; Act proves into Run; Run builds the attention density that makes Network possible. The number that tells a vendor whether the strategy is working is therefore not revenue per rung but the graduation rate between rungs — how many Email accounts adopt Act, how many Act pilots become Run programmes, how much trusted attention becomes Network inventory. Without that movement an ESP does not have a ladder; it has four disconnected product lines, and the commodity gravity of the ground floor will drag the whole structure back down to a per-send argument. EARN is a strategy only if each rung feeds the next.
3
The Business: Making the Climb Real
A model is only as good as the business that can be built on it. The architecture of EARN is clear; what decides whether it becomes a company rather than a slide is harder — the economics of each rung, the way the organisation is wired, what the buyer actually experiences, and the objections honest enough to break it. This part is about the climb in practice.
The economics of the climb
The four rungs do not merely earn different amounts of money; they earn different kinds of money, and the market prices each kind differently. Email is software-shaped at the commodity end: infinite scale, near-zero marginal cost, but benchmarked to the floor and valued as a utility. Act is closer to classic software economics — a capability sold repeatedly across a base — and earns a software multiple when it is productised rather than hand-built each time. Run is, at least at first, services-shaped: it carries real cost of delivery, demands talent and judgement, and is valued more cautiously until it becomes a repeatable system rather than a heroic engagement. Network, once it has density, earns the richest economics of all — media and marketplace yield with network effects — but only a handful of operators ever get there.
Figure 5 — Each rung earns a different kind of money. Value and the multiple the market pays rise as accountability rises.
Two consequences follow, and both are easy to miss. First, the rung that returns margin (Run) is also the rung that consumes capital and attention, because outcome work is funded in advance and collected in arrears, against proven lift. Working capital, not demand, is often the real constraint on how fast a vendor can scale outcome programmes — a queue of eager pilots can starve a balance sheet. Second, the rungs have opposing financial signatures — a high-multiple, low-touch floor beneath a lower-multiple, high-touch middle — which means running them on one P&L blends two businesses the market would value separately. The discipline is to let the software economics of Email and Act fund the services economics of Run until Run becomes systematic, and to ring-fence each so neither distorts the other. The climb is not just a value story; it is a cash-flow story, and the vendors that misjudge the second never finish the first.
The operating model
A new business model needs a new operating model, because the four rungs cannot share a single incentive and survive. The Email team runs infrastructure and is measured on deliverability, reliability, reputation and margin. The Act team runs experiences and is measured on adoption, action completion, data capture and time-to-deploy. The Run team runs outcomes and is measured on verified lift, recovery, control-group discipline and repeatability. The Network team runs media and is measured on fill, yield, advertiser repeat and the health of the audience’s attention. They can share technology, data, design systems and account relationships; they cannot share a scorecard.
The reason is gravitational. If a single team is measured only on send revenue, EARN dies inside the company — every quarter. The urgent renewal always beats the uncertain outcome pilot; the platform quota always crowds out the network experiment; the infrastructure mindset makes every higher rung look like custom work to be avoided. Spare time is Send time, and EARN never gets built in spare time. The structure has to make the higher rungs someone’s actual job, with their own targets, their own definition of success and their own permission to behave unlike the cash engine — services-shaped where the cash engine is software-shaped, patient where the cash engine is transactional. Without that separation, the new business is quietly strangled by the old one’s metrics.
If you are the brand
EARN is written from the vendor’s side, but it is at least as useful read from the buyer’s. For a CMO, the ladder is a way to stop having one undifferentiated argument about email — price — and start having four precise ones. The send is an infrastructure decision: settle it efficiently, keep it reliable, and do not let it consume the conversation. Everything above it is a growth decision, and it should be evaluated on growth’s terms, not procurement’s.
The practical implication is that a marketing leader should refuse to let the two conversations contaminate each other. Outcome work judged as a line-item cost will always look expensive; the same work judged against a holdout, paid only on proven lift, is the safest budget a CMO can hold — spend that, by construction, cannot lose money. So the buyer’s discipline mirrors the vendor’s: agree the measurement before the money, own the baseline and the control group, and treat “no lift, no fee” not as a vendor concession but as the brand’s protection. The brand that learns to buy outcomes instead of sends gets a partner whose incentives finally point the same way as its own. And the brand keeps the thing that matters most — the owned customer relationship — instead of renting it back from a platform every quarter.
The honest objections
A credible strategy names what could break it, and EARN has five real failure modes. The first is measurement. Outcome pricing rests on a clean control group living inside the brand’s data, and whoever owns the holdout, the baseline and the attribution effectively owns the invoice. The control must be concurrent and randomised, never a prior period — otherwise a seasonal swing or a market cycle gets mistaken for the vendor’s lift, in either direction. If measurement rights are not agreed up front, every payout becomes a debate — the vendor claims lift, the brand questions incrementality, finance delays payment. Measurement is not an analytics detail; it is the load-bearing wall of Run.
The second is that outcomes are capital- and trust-intensive, a services-shaped business beside a software-shaped one, which must prove itself repeatedly across accounts rather than once in a friendly pilot — a single success proves possibility, a business requires repeatability. The third is surface control: a vendor may concede delivery pricing and still win if it holds the strategic surface, but conceding the volume, data and events that feed the upper rungs is fatal, because the pipe is also the distribution layer for everything above it. Race the pipe to zero and you lose the right to climb.
The fourth is customer trust. Network is tempting because media revenue scales, and dangerous because the inbox is not a billboard; if monetisation degrades attention, the network eats the very asset it monetises. Control of category, frequency, relevance, labelling and exclusions is not optional. The fifth is organisational drag: incumbents resist model migration, and the easiest evasion is to rename old work in new language. That is not EARN. EARN begins only when the unit of value actually changes — when a vendor is paid, on at least one real account, for an outcome rather than a send.
The new scorecard
The migration shows up most plainly in what gets counted. The old scorecard measures effort and delivery — sends, delivery rate, opens, clicks, complaints, campaign revenue, throughput. Those numbers still matter; they are the telemetry of the Email rung. But they cannot describe the future, because they count what the vendor did, not what the customer’s business gained.
| The old scorecard |
The EARN scorecard |
| Sends, delivery rate, throughput |
Action completion inside the email |
| Open rate, click rate |
Declared first-party data captured |
| Complaints, unsubscribes |
Recovery and reactivation rate |
| Revenue per campaign |
Revenue proven above a control |
| List size |
Real Reach (the genuinely engaged base) |
| Cost per send |
Click Retention Rate; attention yield |
| — |
Graduation rate between rungs |
The last line is the one that tells a vendor whether it is escaping the trap at all. An Email customer who never adopts Act is still a send customer; an Act customer who never reaches Run is still a capability customer. When you change what you count, you change what the business is — the scorecard is not a report on the strategy, it is the strategy made visible.
The staircase of who values you
There is a simple way to see the whole migration: it is a staircase of who values the vendor. Stay on the ground floor and procurement prices you. Climb to Act and marketing values you. Reach Run and the CFO trusts you. Earn Network and you become a media business. The surface beneath your feet never changes — it is the same owned email relationship the whole way up. What changes is what you are willing to be paid for, and therefore who decides what you are worth.
This is not a call to abandon the send. The Email rung remains the foundation — it funds the system, protects the relationship, supplies the data and grants the distribution that makes every higher rung reachable. The instruction is only to understand it as the floor, not the ceiling. The vendor that refuses to climb will not fail dramatically; it will simply keep renewing, keep supporting, keep delivering, and keep losing altitude until it is absorbed, at an infrastructure multiple, into something larger. The vendor that climbs changes what the email business is for. The old ESP was paid to move messages; the new one is paid to make the owned customer relationship more profitable.
Figure 6 — Two frameworks, one surface. SNR is how a brand decides what to say; EARN is how a vendor decides how to be paid.
**
SNR is the brand’s grammar for what an email should do. EARN is the vendor’s model for how the same surface gets paid.
The next email company will not be paid to send more email. It will be paid to make every owned open worth more.
