Priced on articles a month.
5 plans — $1 once, then $89 to $1,200 a month. Every figure below is read from the same file the server bills from.
Every plan runs the whole pipeline and the measurement above it. What changes is how many articles, how many sites, and how small a change in AI citations the measurement can resolve.
A one-time charge, not a subscription — nothing renews. No star rating or customer logos in this space: we do not have customers yet, and we are not going to borrow anyone else’s.
Starter
Measurement resolves citation changes of about 8pp or larger; smaller ones read as inconclusive.
- Site analysis, brand-voice extraction and a banned-phrase list
- Keyword research with intent and clustering
- A 30-day calendar you can reschedule or trim before generation
- 1,500–3,000 words, images with alt text, hard SEO validation
- Auto-publish to WordPress, Webflow, Shopify or a webhook
- Backlink exchange, opt-in per site
- Monthly AI-citation measurement across 4 engines
Not included. No experiments — at 80 tracked questions we’d be selling you noise.
Start writingGrowth
Measurement resolves citation changes of about 4.6pp or larger; smaller ones read as inconclusive.
- Everything in Starter, at twice the volume
- 2 sites instead of 1 site
- Held-out control questions and difference-in-differences readouts
- Pre-registered predictions — written down before the follow-up data exists
- The raw before/after number shown beside the controlled one
- A running log of what was published, predicted and learned
Scale
Measurement resolves citation changes of about 2.9pp or larger; smaller ones read as inconclusive.
- Everything in Growth
- 60 articles a month across 4 sites
- 600 tracked questions, so ~40% smaller detectable effects
- 8 runs per prompt instead of 6
- For crowded categories where a few points is a real quarter’s work
Agency — 90 articles a month across 5 sites
Each client is a separate project — profile, voice, keywords, calendar and CMS, nothing shared. Measurement runs per client at ~2.1pp resolution, with a held-out control and a prediction written down beforehand. Not yet shipped: white-label branding, and a public API. If you need either to sell this on, wait for them.
All five plans, side by side.
| Plan | Price | Billing | What’s included |
|---|---|---|---|
| Trialtrial | $1 | Oncenothing renews | 3 articles · 1 site · 20 tracked questions · 6 runs each · 4 enginesThe whole pipeline on one site, ending in 3 finished articles you can read. Drafts carry a one-line watermark. One measurement round — enough to see where you stand, not enough to detect a change. |
| Startermeasure | $89 | Monthlyuntil you cancel | 15 articles/mo · 1 site · 80 tracked questions · 6 runs each · 4 enginesMonthly AI-citation measurement across 4 engines. No experiments: you will see movement and will not be able to attribute it. |
| Growthprove | $349 | Monthlyuntil you cancel | 30 articles/mo · 2 sites · 240 tracked questions · 6 runs each · 4 enginesWhere proof starts. Pre-registered experiments with held-out controls, resolving citation changes of about 4.6pp or larger. |
| Scaleprove_plus | $899 | Monthlyuntil you cancel | 60 articles/mo · 4 sites · 600 tracked questions · 8 runs each · 4 enginesThe same experiments on twice the questions, resolving about 2.9pp — for categories where the change you are looking for is small. |
| Agencyagency | $1,200 | Monthlyuntil you cancel | 90 articles/mo · 5 sites · 1,200 tracked questions · 6 runs each · 4 enginesOne project per client — profile, voice, keywords, calendar and CMS, nothing shared — measured at about 2.1pp each. White-label branding and a public API are not shipped. |
No annual plan, no volume discount, and no crossed-out anchor price. Each price above is the price: $1 once for the trial, and every other plan billed monthly in the same currency. Checkout will accept a promotion code if we have given you one; that is the only discount that exists today. If we add annual billing you will see the annual number in this table, not a percentage in a badge.
What one article is, and what depends on your own configuration.
- word-count
- 2,140 words
- keyword-density
- 1.8% of 2.5%
- internal-links
- 3 of 2–5
- no-fabricated-specifics
- 0 unsourced claims
- title-length advisory
- 63 chars — over 60
An illustration of the scorecard every draft carries — the numbers in it are made up for the picture, the rules are not. An advisory costs 5 points; a hard failure costs 15 and the draft is held for review rather than published.
- An article is 1,500–3,000 words with one H1, at least two H2s, density capped at 2.5%, 2 to 5 internal links to your own pages, a hero plus one or two in-body images with alt text, and a fitted meta description.
- Regenerating spends one from the month. It writes a new version rather than overwriting the old one, so the count is articles produced, not articles you liked.
- Real keyword volume needs a DataForSEO account. Without one, volume and difficulty stay empty and show “—” — never a made-up number, and never a zero standing in for one.
- Photographic images need an image provider (Replicate). Without one every article still gets illustrations — deterministic abstract SVGs rather than generated art.
- Detectable effects are indicative, not promised. They are estimated from your own baseline variance, and power comes from the number of distinct questions tracked, not from asking one of them more times.
- Plan ids are trial, measure, prove, prove_plus and agency — the same five tiers as the table above. An invoice showing a tier you did not choose is a bug worth telling us about.
The questions a careful buyer asks.
What counts against my article allowance?
Every article generated for the project that month, including regenerations — you keep both versions, but it costs one. The allowance is counted the same way as the number in your workspace header, so a refusal always matches what is on screen.
Is there an annual plan, a volume discount or a founder price?
No. There is one price per plan and one billing interval, both in the table above. We would rather leave money on the table than show you a crossed-out number we never charged anyone.
Do the articles go straight to my live site?
Only if you set the connection that way. Each connection publishes as a draft or as a live post, and nothing publishes until we have verified the credentials can actually create posts. A retry never leaves a duplicate.
Why is the Trial only 3 articles?
Because the thing you need to decide is whether it sounds like you, and three articles answer that. Generating thirty before you have read one spends a month on a voice you would have corrected after the first piece. Trial drafts carry a one-line watermark; paid output does not.
What happens if I cancel?
Your account drops to Trial limits — 20 tracked questions, 1 site, no experiments — and nothing is deleted. Your articles and your measurement history are the thing you came for; destroying them over a lapsed card would be indefensible. Billing is managed in Stripe, from Settings → Billing.
Where do keyword volume and difficulty come from?
From a DataForSEO account, which is configuration rather than something every plan switches on. Without one, those columns render an em dash and say why. We never print a zero where we have no number — a zero is a measurement, and we did not take one.
Will every article have a photograph?
Only with an image provider (Replicate) configured. Without one, articles still get a hero and in-body illustrations, but they are deterministic abstract SVGs rather than generated photographs. Both carry alt text and both count as a finished article.
Is the backlink exchange included?
On every plan, and off until you turn it on for a specific site. A relevance floor of 0.35, 1 link per article, 3 placements per site per week and no reciprocal pairs are enforced in code, not in a policy page. The risk is real and the penalty would land on your domain, so leaving it off is a legitimate choice.
Why does the measurement run monthly?
Content needs a window to be indexed before re-measuring means anything, and identical questions return materially different answers about a quarter of the time. Weekly deltas are mostly run-to-run variance wearing a trend line.
Can I get white-label reports or an API?
Not today. Both are named on the Agency plan as not shipped, and they will stay named that way until they exist. If reselling under your own brand is the reason you are here, that plan is not yet worth $1,200 to you.
Can I check your statistics?
Yes. We scored our estimator against 150 simulated worlds with known ground truth and published the failures too: our 95% intervals covered the truth 92.7% of the time, not 95%. Read the calibration.
Read three articles for a dollar.
The full pipeline on your own site, ending in 3 finished articles. If they do not sound like you, you have lost a dollar.