Cost guide
What a Fact-Audited Page Costs to Produce
The short answer
What does a fact-audited page cost to produce?
A fact-audited page costs 5 stages of labor — research, drafting, fact audit, publish QA, and scheduled refresh — and in our production system the audit and refresh together are the largest share of lifetime effort: one site-wide audit on one of our builds checked 445 claims and corrected 40 [our data]. The dollar figure depends on wage assumptions, so this page ships the labor model and the worked math, not a fake universal price.
Ask what a page costs and most of the market answers with the price of writing. That answer counts 1 stage out of 5. A page that can survive being quoted by an answer engine — and survive a skeptical reader checking it — is researched, drafted, fact-audited, QA'd at publish, and refreshed on a schedule. Across our 3 production brands and 400+ published pages, those stages are the real unit economics [our data].
This page ships the model rather than a universal dollar figure, because the dollar figure is just your wage assumptions multiplied through the hours — and any vendor quoting one price for every buyer is hiding an assumption somewhere.
What are the 5 stages of a fact-audited page?
| Stage | What happens | Cost behavior |
|---|---|---|
| Research | Question inventory, source gathering, angle | Scales with topic difficulty |
| Draft | Answer-first writing to the content contract | Scales with length |
| Fact audit | Every claim traced, every figure recomputed | Scales with claim count |
| Publish QA | Frontmatter, links, schema, gate checks | Near-fixed per page |
| Refresh | Scheduled review and update | Recurring, annualized |
The stage most buyers have never been quoted for is the audit. Ours works claim by claim: every statistic must trace to a source on an approved list or to our own logs, every derived number is recomputed, and every hedge — "commonly," "roughly" — is treated as an unverified claim to verify or cut. When we ran a site-wide audit across one of our production builds, it checked 445 claims and corrected 40 [our data]. That is a 9% correction rate on content that had already shipped — our own content, which is exactly why the stage exists.
What does the worked math look like?
The hour figures below are illustrative — the labor mix is how our pipeline is structured [our data]; the specific hours and wages are yours to replace.
Illustrative page: research 1.5 hours, draft 2 hours, fact audit 2 hours, publish QA 0.5 hours, refresh 1 hour a year (annualized). Lifetime labor: 7 hours. Audit plus refresh: 3 of those 7 hours — about 43% of lifetime labor sitting in the 2 stages cheap quotes omit.
Now the multiplication anyone can finish: at a fully loaded $50 an hour, the illustrative page costs $350; at $100 an hour, $700. Compare that against any per-page or retainer quote — including the monthly retainers sold under the GEO label — by asking one question: which of the 5 stages does this price include, and at what depth? A $150 page is not a cheaper version of a $700 page. It is a different product with 3 stages missing.
Where this sits in a whole-site budget is a multiplication of its own: run the per-page model through your architecture's page count, then add the fixed engine work — platform, gates, analytics — that no per-page rate captures.
What changes at scale — and what refuses to?
Scale bends 3 of the 5 stages. Research gets cheaper per page once a question inventory exists for the niche: the second page in a cluster reuses the first page's sources. Publish QA collapses toward zero when gates are automated — on our builds, the checks run mechanically on every page, so page 400 costs the same near-nothing to validate as page 4 [our data]. Drafting compresses moderately as templates and voice rules harden.
The audit refuses to scale, and that is the honest center of content-ops economics. Claims do not verify faster because you have verified other claims; each statistic still has to be traced, each computation re-run. Our fleet's experience across 400+ pages is that audit labor stays roughly linear with claim count [our data] — which means the stage that differentiates the content is also the stage that never gets cheap. Any per-page price that falls dramatically at volume is discounting the one stage that cannot be discounted.
Why is refresh part of per-page cost?
Because recency is not cosmetic in AI search. Kevin Indig's State of AI Search Optimization analysis found content less than 3 months old was about 3x more likely to be cited in ChatGPT answers (Growth Memo, 2026) — a correlation from one analysis, not a law, but consistent with what freshness has always done in ordinary search. A library that is never revisited decays on both surfaces at once.
Refresh is also where fact-audited content earns compound interest: a page whose claims were sourced and dated at publish can be re-verified quickly, while a page of unsourced assertions has to be re-reported from scratch. Our freshness system enforces review dates by cluster — money pages every 182 days, reference pages every 365 — and the build fails when a page goes stale [our data]. That enforcement is part of the authority engine system, not a good intention.
Where does cheap volume actually lose?
The false economy runs through the audit. Skip it and the per-page price drops by roughly the audit's share of labor — while the pages carry whatever your version of our 9% correction rate is, unfound. Google's own guidance frames the quality bar as content that demonstrates first-hand depth and reliability (people-first content guidance), and an answer engine that quotes your page is republishing your claims under your name. Errors at scale are reputation spend, not savings.
Said against our own interest: if the audit stage is what your budget cannot fund, do not buy more pages — buy fewer. A 40-page library that survives checking beats a 200-page library that does not, on every surface described in the operator's guide. Volume is only an asset when every unit of it can afford to be quoted; if you are unsure your niche justifies even the smaller build, the fit check is the cheap way to find out.
Frequently asked questions
What does a fact-audited page cost?
It costs 5 stages of labor: research, drafting, fact audit, publish QA, and refresh. The dollar total is your wage assumptions times those hours — in our illustrative model, audit and refresh alone are about 43% of lifetime labor, which is the share cheap quotes leave out.
What is a fact audit and why is it expensive?
A claim-by-claim review: every statistic traced to an approved source, every derived number recomputed, every unverifiable sentence cut or hedged. One site-wide pass on one of our production builds checked 445 claims and corrected 40 [our data] — labor that scales with claims, not with word count.
Is a 9% error rate normal for published content?
It is what our own audit found on our own already-shipped pages: 40 corrections across 445 checked claims [our data]. We publish the number because it is the honest baseline — content produced without a dedicated audit stage ships errors, including ours.
Why does per-page cost include refresh?
Because a page is an asset with maintenance, not a one-time deliverable. Recency correlates with citation: content under 3 months old was about 3x more likely to be cited in ChatGPT answers (Growth Memo, 2026). A cost model with no refresh line understates real cost every year after the first.
Can you skip the audit to cut cost?
You can, and the market mostly does. But an answer engine quoting your page is quoting your claims, and a corrected claim rate of 9% on our own pre-audit content [our data] says unaudited pages carry errors. If the audit is unaffordable, publish fewer pages — volume without verification is the false economy.