The data layer under Draftcamp —
and why combining it is the whole point
Four independent signals about every article.
Any one of them, alone, misleads.
4 Signals
Per article, per audit pass — judged together, not separately
Search
GSC performance history
Market
Live keyword intelligence
You
Your library + your ICP
Single-signal tools see traffic. That's one-quarter of the picture
Every signal, read alone, has a failure mode
Traffic alone lies by omission. A page holding position 4 looks healthy — until you know the query's volume doubled and competitors' coverage moved past it. And it says nothing about whether the article still argues your current case, to your current buyer.
Keyword data alone lies by abstraction. Volumes and difficulty scores describe the market, not your page in it. Optimizing to keyword data without your own performance history is how teams rewrite pages that were quietly winning.
Content analysis alone lies by vacuum. An article can be beautifully structured, technically clean, on-brand — and rank for nothing, or rank for queries that bring the wrong buyer. Quality without performance context is decoration.
The failure modes cancel out when the signals are read together. That's the design decision everything else follows from.
SIGNALS 1 & 2 — THE MARKET SIDE
What search actually did,
and what the market actually wants
The performance half of the picture: your page's real history in Google, put in context by live market data.
Signal 1 — Search performance (GSC)
Per page, per query, per day: clicks, impressions, CTR, position. Synced continuously and kept as history — so the audit sees trajectories and drop deltas, not snapshots. Including the drop that matters most: crossing off page 1.
Signal 2 — Keyword intelligence
Live search volumes, related and long-tail terms, intent classification — plus the questions AI answer engines surface for the topic. Pulled fresh per audit and per brief, not cached from last quarter.
Why they're read together
GSC says where you stand; keyword data says what standing there is worth, and what nearby ground is unclaimed. Position 8 on a dying query and position 8 on a growing one are opposite priorities.
Signal 1 — Search performance (GSC)
Per page, per query, per day: clicks, impressions, CTR, position. Synced continuously and kept as history — so the audit sees trajectories and drop deltas, not snapshots. Including the drop that matters most: crossing off page 1.
Signal 2 — Keyword intelligence
Live search volumes, related and long-tail terms, intent classification — plus the questions AI answer engines surface for the topic. Pulled fresh per audit and per brief, not cached from last quarter.
Why they're read together
GSC says where you stand; keyword data says what standing there is worth, and what nearby ground is unclaimed. Position 8 on a dying query and position 8 on a growing one are opposite priorities.
SIGNALS 3 & 4 — YOUR SIDE
What your library actually says,
and what your business actually needs
The alignment half: the content as it exists today, benchmarked against the company as it exists today.
Signal 3 — Your library, crawled and versioned
Every article fetched, parsed, and profiled: full text, titles, metas, heading structure, links, images, schema — kept as versions, so the audit knows what changed on the page and when.
Signal 4 — Your ICP and brand definition
The one signal no vendor can sell: who you sell to and how you talk. Compiled from your own articles and style guide, confirmed and corrected by you — the explicit benchmark drift is measured against.
Why they're read together
Signal 3 without Signal 4 is a text dump. Signal 4 without Signal 3 is a strategy doc. Together they answer the question dashboards can't: does this page still make our case?
Signal 3 — Your library, crawled and versioned
Every article fetched, parsed, and profiled: full text, titles, metas, heading structure, links, images, schema — kept as versions, so the audit knows what changed on the page and when.
Signal 4 — Your ICP and brand definition
The one signal no vendor can sell: who you sell to and how you talk. Compiled from your own articles and style guide, confirmed and corrected by you — the explicit benchmark drift is measured against.
Why they're read together
Signal 3 without Signal 4 is a text dump. Signal 4 without Signal 3 is a strategy doc. Together they answer the question dashboards can't: does this page still make our case?
The novelty isn't any one dataset. It's what four of them say together
Each signal exists somewhere else. The judgement across them doesn't
You can export GSC. You can subscribe to keyword data. You can crawl your own site. You can write your ICP on a wall. Teams have all four — in four tabs, examined by four people, on four schedules. The insight lives in the intersections, and intersections are exactly what tab-switching loses.
A concrete example of the intersection logic: an article holds position 3, so traffic tools stay silent. The keyword signal shows the query's intent has shifted commercial. The library signal shows the page hasn't changed in 19 months. The ICP signal shows it targets the segment you moved away from last year. Read together: your best-ranking page is your biggest liability — a conclusion no single signal supports, and the audit reaches it automatically, for every article, on every pass.
That's also why the output isn't a dashboard. Cross-signal conclusions come with reasoning attached — flagged as touch-up, rewrite, or retire — because a judgement you can't inspect is a judgement you can't trust. The system shows its work; your team makes the call.
What we found running this on our own library first
Draftcamp was built inside Socialinsider. Our blog was the first dataset
Before this was a product, it was the system we ran on our own blog — a library built over years that had quietly started losing rankings we'd spent those years earning. The first full audit was the argument for productizing it.
⚠ [PLACEHOLDER — pull real figures from Socialinsider audit runs: articles analyzed, % flagged by classification (touch-up / rewrite / retire), count of still-ranking-but-drifted pages, count of page-1 exits detected, most common technical finding. 3–4 real numbers beat 10 vague ones.]
⚠ [PLACEHOLDER — one concrete anonymized example: the strongest 'ranking fine, completely wrong for our ICP' article the audit caught, and what the reasoning said. This is the canvas's dream quote, from our own data.]
The data layer, answered
The honest answers.
See the four signals read your library
Book a 30-minute demo — connect GSC and watch the cross-signal audit run on your real articles, reasoning included.
✓ 30 minutes ✓ Read-only access ✓ Every flag shows its reasoning