The Measurement Problem in Creator Economics

I have spent a meaningful part of the last fifteen years building measurement frameworks. At Publicis, during the digital-insights years, measurement was the product — figuring out what a marketing dollar actually bought, in a world where the tooling was changing faster than the accounting. At Xynteo, measurement was the vocabulary we used to tell purpose-led businesses whether their purpose was actually working. At Laminar and now at APMC, measurement is what lets us tell the difference between a streaming service that is growing and one that is running on momentum from a previous decision.

After fifteen years of this, I have developed one strongly held view.

Measurement frameworks outlast the businesses they were invented for. They persist, in dashboards and in decks and in the muscle memory of the people who came up in them, long after the underlying economics have moved. And the damage a stale measurement framework does is not that it produces wrong numbers — it is that it produces numbers that look right, and therefore do not get questioned, while the business they describe quietly turns into a different business.

This is what has happened to creator-economy measurement. The canonical metrics — CPM, followers, watch time, RPM — are all inherited from businesses that existed before the creator economy did. CPM is a television-and-display-advertising metric. Followers is a social-media-platform-growth metric. Watch time is a broadcast-network metric dressed up for a streaming environment. RPM is a YouTube-era adaptation of CPM with some platform take adjustments baked in.

None of those four metrics describe the business that most serious creators are actually running in 2026. Modern creator revenue is a composition of three things — long-tail attention rent, backlog-compounding library value, and direct-relationship subscription premium. The canonical metrics price none of the three components correctly. That is the measurement problem, and it is, I think, the most significant unresolved structural issue in the creator economy today.

What follows is a framework-level argument. I am not going to name specific creators or specific platforms beyond what is necessary. I am not going to publish revenue figures I cannot source. The argument is structural, not anecdotal. It is about measurement. It is, specifically, about why the measurement is wrong.

What The Inherited Metrics Were Built For

To see why the current metrics fail, it helps to go back to what each of them was originally built to measure.

CPM — cost per thousand impressions — was a mid-twentieth-century advertising accounting convention. It existed because the buyer (an advertiser) and the seller (a network or a publisher) needed a common language for the unit of inventory, and that unit was an impression. The metric assumed a few things. It assumed the impression was real. It assumed the environment around the impression was roughly uniform. It assumed the advertiser’s goal was aggregate reach, because that is what television and print advertising could deliver at scale and what the attribution tooling of the era could measure.

None of those assumptions hold in the creator economy. The impression may or may not be attended. The environment around the impression varies wildly — a ten-second ad read inside a thirty-minute trusted-voice podcast is doing something structurally different from a programmatic pre-roll on a random vertical video. The advertiser’s goal is increasingly not aggregate reach but resonant attention, which CPM does not measure at all.

Followers is a vanity metric that became an economic metric through platform politics. Originally, it was a crude proxy for audience size — how many accounts had explicitly opted to receive content from a given creator. That proxy worked, approximately, for about five years. Then the platforms moved to algorithmic feeds. Once the feed was algorithmic, the follower relationship decoupled from the distribution relationship. A creator with a million followers who was deprioritized by the algorithm was reaching fewer people than a creator with ten thousand followers who was being pushed to a broader non-follower audience. Follower count persisted as a metric anyway, because it was easy to count and because the platforms had an incentive to keep it in the conversation as a proxy for value.

Watch time was introduced by YouTube in 2012 as the replacement for view count in the algorithm — the platform was being gamed by creators who optimized for clickthrough at the expense of actual consumption, and the platform needed a metric that rewarded content which held attention rather than content which merely triggered a click. Watch time did that. It was a better metric than views, for the platform’s purposes in 2012. But watch time, like CPM, treats a minute as a minute. It does not distinguish between a minute of attentive, arousal-weighted consumption and a minute of ambient, half-attended background play. I have argued at length elsewhere that those two minutes are not economically equivalent, and the creator economy is a particularly acute place for that distinction to matter.

RPM — revenue per thousand views, or per thousand minutes watched, depending on the platform’s convention — is a creator-facing derivation of CPM with platform-share adjustments baked in. It is useful, as a rough input for a creator trying to model whether a given piece of content will cover its production cost. It is catastrophically inadequate as a representation of the value a creator is actually generating across their full body of work and across their relationship with their audience.

The Three Components of Modern Creator Revenue

The business that a serious creator is actually running, today, has three revenue components. They do not show up discretely on most dashboards. They are often bundled into a single aggregated “monthly earnings” number that hides the structure underneath. The structure matters.

Long-Tail Attention Rent

The first component is what I will call long-tail attention rent. A creator produces a piece of content. That piece of content, under the current platform economics, continues to be surfaced to new audiences for months or years after it was published. Each surfacing produces some small increment of attention, which converts into some small increment of revenue — through platform ad share, through evergreen sponsorship placements, through the indirect mechanism of driving new audiences to the creator’s other properties.

Long-tail attention rent is the recurring royalty a creator earns for content they already shipped. It is the closest analogue, in the creator economy, to the back-catalog revenue that a television studio earns from licensing reruns, or that a music label earns from catalog streaming.

CPM measures the first-day-surface of a piece of content. RPM measures the first-month or first-year surface. Neither measures the fifth-year surface. And for a creator with a meaningful body of work, the fifth-year surface is a very large fraction of the total lifetime value of that work.

Most creator-economy conversations treat long-tail revenue as a nice-to-have rather than a structural component. That is backwards. For a creator with three to five years of output, long-tail attention rent can be a material and growing share of total earnings — and it grows without any new production effort, which makes it the highest-margin part of the business.

Backlog-Compounding Library Value

The second component is adjacent to the first but meaningfully different. A creator’s backlog — the archive of content they have produced over years — is not just an annuity that spins off long-tail attention rent. It is also a discovery mechanism. It is the library that a new audience member, having found the creator through one piece of content, can work backwards through. It is the thing that converts a casual viewer into a subscriber, because it is evidence that there is more where this came from.

The economic value of that evidence is enormous and is not currently priced into any standard creator metric.

When a potential subscriber is deciding whether to commit to a creator — whether via a platform-native subscription, a newsletter sign-up, a Patreon tier, a Substack subscription, a YouTube membership — the single biggest determinant of that decision is whether the backlog is deep enough to feel like a worthwhile investment. A creator with five years of strong backlog is much more subscribable than an equivalent creator with six months of strong backlog, even if their most recent work is at the same quality level. The backlog is doing the work of de-risking the subscription decision.

A creator's backlog is not just an annuity. It is the thing that converts a casual viewer into a subscriber, because it is evidence that there is more where this came from.— the thesis

No current metric captures this. Watch time does not capture it. CPM does not capture it. Follower count does not capture it. And because no metric captures it, creators in the middle of their careers systematically underinvest in backlog quality — cleaning up old content, re-packaging older pieces, making the archive more discoverable — because the dashboard does not reward that work. The dashboard rewards new production.

The misallocation of creator effort produced by that measurement gap is, I think, the most expensive single thing in the creator economy today.

Direct-Relationship Subscription Premium

The third component is the one that has received the most rhetorical attention and the least rigorous measurement. The direct-relationship subscription premium is what a creator earns from the fraction of their audience who have opted to pay for ongoing access — through a newsletter subscription, a podcast membership, a Patreon tier, a Substack paid tier, a platform-native support mechanism, a private community access fee.

The subscription premium is structurally different from ad-supported revenue in ways that the standard metrics completely miss.

A paid subscriber is not a unit of audience that happens to also pay money. A paid subscriber is a different economic relationship entirely. They are much less churn-prone than a free follower. They consume more content per visit. They share more. They convert at much higher rates to the creator’s other offerings — books, courses, live events, merchandise. They are, in other words, the compounding asset. The free audience is the traffic. The paid subscribers are the business.

No creator dashboard I have seen represents this correctly. The dashboards aggregate paid subscribers into the same “audience” number as free followers, apply the same engagement metrics to both, and then report a blended revenue figure that makes it impossible to see which part of the business is compounding and which part is just churning through.

Why the Mispricing Is Structural, Not Incidental

It would be tempting to read the preceding sections as an argument that the platforms just need to add a few new metrics and the problem will be solved. That is not the argument.

The measurement problem is structural. It is baked into the revenue models of the platforms themselves. A platform that is primarily an advertising business — which most of the major creator platforms still are — has a strong incentive to optimize its measurement layer around ad-facing metrics, because those are the metrics that determine the platform’s own revenue. CPM, RPM, watch time, and follower count are all, in one way or another, metrics that the platform’s ad sales organization cares about. The platform publishes those metrics to creators because publishing them is cheap — the platform is already calculating them for its own purposes.

Long-tail attention rent, backlog value, and subscription premium are metrics the platform’s ad sales organization does not directly care about. They are creator-facing metrics that would require separate instrumentation, separate reporting, and separate engineering investment. The platforms have, with rare exceptions, not made that investment, because the return on that investment accrues primarily to creators rather than to the platform’s own P&L.

The mispricing, in other words, is not an oversight. It is the natural equilibrium of a system where the measurement infrastructure is paid for by one party and consumed by another, and the two parties do not have aligned interests.

This is the part of the creator-economy conversation that I think gets least attention. Everyone is arguing about platform take rates. Almost no one is arguing about the measurement stack on top of which the take rates operate. But the measurement stack determines what the creators can optimize for, and the creators’ optimization choices determine what kind of content gets made, and the content that gets made determines what the platform actually is.

What a Creator-Facing Measurement Stack Would Look Like

If we were starting from scratch — if a platform were being built today specifically to represent the creator economics that actually exist — the measurement stack would look different in at least three ways.

First, it would report long-tail revenue separately from current-period revenue, broken out by content-cohort age. A creator would be able to see, at a glance, what share of this month’s earnings came from content shipped this month, this year, two years ago, and five-plus years ago. That single breakdown would transform creator investment decisions. It would make the backlog visible as the compounding asset it is.

Second, it would report subscriber metrics and follower metrics as separate categories, with separate engagement profiles, separate revenue attribution, and separate lifetime-value modeling. The conflation of subscribers and followers into a single “audience” number is the single most damaging convention in current creator dashboards. Unconflating them would clarify, for every creator in the category, which part of their business is worth investing in.

Third, it would attempt some form of attention-weighting on the consumption metrics, even if the weighting were imperfect. Completion rate, rewatch rate, share rate, the ratio of active-visit time to passive-background time, the gap between autoplay-started views and user-initiated views — these are all fragments that exist in platform telemetry and that could, in combination, produce a directionally correct attention-weighted consumption number. Nobody is publishing that number today. No major creator platform — not YouTube’s public Studio reporting, not TikTok’s analytics, not Patreon, not Substack — surfaces anything beyond standard watch time and completion-rate derivatives as of this writing. The first platform that does will have an advantage that is hard to appreciate until you try to optimize a creator career without it.

What Happens If Nothing Changes

I do not want to end on an optimistic note that is not warranted. The most likely scenario, in my view, is that the measurement layer of the creator economy does not change meaningfully for the next several years, because none of the incumbent platforms have a direct incentive to change it.

If nothing changes, several things will continue to happen.

Creators will continue to over-invest in new production and under-invest in backlog, because the dashboard rewards the former and is silent on the latter. The careers that compound will be the ones whose operators intuit the backlog’s importance despite the dashboard, and those careers will be rare.

Creators will continue to treat subscribers and followers as interchangeable, because the platforms do. The subscription premium — which is the most durable revenue component in the creator economy — will continue to be systematically underdeveloped, because the tools for developing it are not well-instrumented.

Advertisers will continue to price creator inventory using CPM-derived conventions, which will continue to misprice the high-resonance inventory by a large margin. The creators running the most attention-dense formats — deep-dive podcasts, long-form video essays, serialized written work — will continue to extract less revenue per unit of earned attention than those formats are actually generating, and the category will continue to feel, to serious practitioners, slightly economically broken in a way they cannot name.

The creator economy, in other words, will continue to behave like a large, growing market that is also — at the level of the individual creator P&L — a little more precarious and a little less lucrative than it should be. The gap between the aggregate growth story and the individual-creator experience will persist, and it will be explained, as it has been for years, by the somewhat mystical invocation of “platform dynamics” rather than by the much more prosaic reality of a measurement stack that is describing the wrong business.

The Measurement Layer Is The Business

I want to land on a claim that I think is strong.

The measurement layer is not a technical artifact sitting adjacent to the creator economy. The measurement layer is the creator economy, because it is the thing that determines what creators can see, what creators optimize for, and therefore what gets made and how it gets monetized. A creator economy with the right measurement layer is a fundamentally different market than a creator economy with the wrong one — different content, different revenue distribution, different career paths, different power dynamics with platforms.

The industry has, for fifteen years, treated the measurement question as a detail. An engineering task. Something that would get solved eventually. I do not think it will get solved eventually by the incumbents. I think it will get solved — if it gets solved — by a new generation of tools, likely independent of the major platforms, that start to represent long-tail revenue, backlog value, and subscription premium as first-class metrics in a way that the platforms do not.

When those tools arrive, the creator economy will feel, retrospectively, like it had been badly mismeasured for its entire existence. Careers that looked thin will turn out to have been deep. Careers that looked dominant will turn out to have been shallow. The people who understood the three components — long-tail rent, library value, subscription premium — without being able to measure them cleanly will turn out to have been operating on a different and better map than the dashboards allowed.

I do not know exactly when that arrives. I do know the direction the map is moving in. And I know that the creators who internalize the three-component framework now — who build their work around all three, rather than just the one the dashboard rewards — will be the ones building businesses that compound while the rest of the category is still counting followers.

The measurement problem is solvable. Nobody is solving it yet.

That is the opportunity.