§01.01
Keyword volume isn’t a brief.
Most agencies write what’s high-volume. AI cites what’s authoritative. They’re rarely the same article — and the volume play loses both.
Service · Content Strategy
Write the source AI cites.
Most content gets briefed from keyword volume. We brief from the prompts your buyers run, the pages AI already cites in your category, and your own sales-call and support-ticket logs — then ship content deep enough to be cited.
What we put in one briefPer article
Buyer prompts47
Cited pages pulled193
Competitor keywords612
Sales-call themes36
Support-ticket questions84
Category positioning·
OutputShips in 5 days
One deep article — cited where AI looks.
Key takeawaysFAQ blockSchemaInternal links
350+
pieces shipped in 6 months
B2B SaaS · project mgmt
5 / day
deep outlines, sustained
Insurance · California broker
65%
organic growth, 6 months
B2B SaaS
5% → 25%
organic share of revenue
DTC wellness · Series A
§01 — Why most content missesThree reasons content gets
Three reasons content gets
ignored by AI.
AI doesn’t reward volume. It rewards depth, sourcing, and authority. Most content programs optimize for the wrong three things.
§01.02
Top-3 competitor research isn’t enough.
You see what your closest peers wrote. You don’t see the editorial citations, Reddit threads, and category guides AI actually pulls from when it answers.
§01.03
Generic content reads as AI-written.
Buyers can tell. So can AI models, increasingly. Generic content doesn’t earn citations, doesn’t earn trust, and doesn’t convert.
§02 — How we research before we writeFive inputs.
Five inputs.
One brief.
The brief is the product. By the time we’re writing, the decisions are already made — what to say, what to cite, what to skip, what voice your buyers actually use.
Brief intake5 sources · 1 output
- §02.01
Map the prompts — GEO and SEO.
30–60 buyer prompts your category actually generates across the AI platforms, plus the keyword space that drives Google traffic. Both surfaces, not just one.
- §02.02
Pull the citation graph.
For every prompt, we pull every page AI currently cites — competitor pages, editorial sources, Reddit threads, listicles. That’s the existing answer. To replace it, we have to read it.
- §02.03
Run competitor traffic analysis.
Which keywords your competitors win, which pages they’re building authority on, where you can outflank them with depth instead of breadth.
- §02.04
Mine your internal data.
Sales-call transcripts, support tickets, customer-service logs, churn interviews. The questions buyers actually ask — in the language they actually ask them in. This is where the unique voice comes from.
- §02.05
Find your positioning.
Industry patterns — X-vs-Y, X-alternative, best-of listicles, deep guides — are tools, not the strategy. We research what the category already says, then find the angle that’s yours.
§03 — How we shipFour steps from brief
Four steps from brief
to your CMS.
You stay in the loop where it matters. Everywhere else, the work ships without you having to chase it.
§03.01
Topic list, with you in the loop.
We propose 30–60 topics ranked by GEO and SEO impact. You pick what fits your roadmap and your conversations with users. Nothing ships you haven’t seen.
§03.02
Drafts with research depth.
Every piece is sourced from real data — your internal logs, primary research, third-party citations — not summarized from the top three Google results. Review every draft, or trust the editor and skip to publish.
§03.03
Human edit on every piece.
An editor reads, edits, and adds the final voice layer. That’s how content stops sounding AI-generated, even when AI handled the first 80%. Always a human, every time.
§03.04
Direct-to-CMS publish.
Key takeaways, FAQs, schema, internal links — formatted for Google ranking and AI retrieval. Pushed straight into your CMS via custom scripts. No copy-paste, no formatting tax on your team.
§04 — Receipts, in detail350 pieces in 6 months.
350 pieces in 6 months.
4% → 61% citation share.
Mid-market B2B SaaS · project management
A project-management SaaS in the $5–15M ARR band, losing the organic visibility race to bigger names. CAC was climbing. Demos were drying up. Their existing content program was producing high-volume keyword bait that nobody — including AI — was citing.
We mapped 240 buyer prompts, pulled the citation graph for every one of them, and mined their support-ticket database — which surfaced 87 specific buyer questions their existing content never addressed. Then we shipped: about 58 pieces a month, every piece human-edited, every piece schema-formatted for both Google and AI retrieval. By month 6: 200+ new page-1 keywords, 2.5x AI-platform visibility, 35% lower CAC.
Read the full case study →Pieces shipped · monthlyTotal · 350
32M1
48M2
56M3
62M4
70M5
82M6
Citation share · 4% → 61%−35% CAC
§05 — What makes a Geology article differentFour standards
Four standards
every piece meets.
01
Briefed from real prompts.
Every article maps to a specific buyer prompt or cluster — not a keyword string a tool returned.
02
Sourced from primary data.
Citations from your internal logs, plus at least five third-party sources we’ve actually verified. No summarized blog posts.
03
Formatted for retrieval.
Key takeaways, FAQs, structured data, AI-readable schema, internal links — every piece, no exceptions.
04
Reviewed by a human.
An editor reads every article before it ships. That’s the line between content that gets cited and content that gets dismissed.
§06 — Common questionsWhat buyers
What buyers
actually ask.
Pulled from the questions that actually come up in our intro calls — not the FAQ template every agency runs.
How fast can you ship?
Brief turnaround is 5 days from kickoff. After that, expected throughput depends on cadence — 8 to 16 deep articles a month is typical, 50+ when we run full-volume programs like the project-management SaaS case study above. Quality of internal data is the biggest variable; the more sales-call and support-ticket data we can mine, the faster the briefs lock.
Do you write everything from scratch, or use AI?
Both. AI handles the first pass on structure, citation pulling, and draft scaffolding. A human writer then rewrites for voice, adds primary-source citations from your internal data, and an editor does final pass. The deliverable doesn’t read as AI-generated because the last 20% is always human — and that’s the 20% that matters.
Do I have to publish what you draft?
No. You see the topic list, you see every brief if you want to, you see every draft if you want to. Anything you don’t approve doesn’t ship. Most clients trust the editor by month two and only review drafts on flagged topics.
What does the brief actually look like?
About 1,500 words. Buyer prompt or cluster the article targets, the existing AI citations we’re displacing, the angle and positioning, internal-data sources being mined, an outline with key data points, the schema markup we’ll attach, and the internal links. The brief is the product — by the time we’re writing, the decisions are made.
Can you publish to my CMS, or do I have to copy-paste?
We publish directly. Custom integration scripts per client — WordPress, Webflow, Sanity, custom CMSes — push the article in with schema, key takeaways, FAQs, and internal links already in place. We’ve never failed to integrate with a CMS yet.
What if my industry doesn't have the kind of internal data you mine?
Then we lean harder on prompt mining, primary research interviews with your team, and citation-graph analysis. The internal data is a multiplier when it’s there — early-stage clients without sales-call archives yet still get the prompt-first, citation-graph-first methodology, and we’ll backfill the internal-data layer once the data exists.
Get started
See how we'd brief your next 30 articles.
Run a Live Audit. We pull your prompt landscape, the pages AI cites for each, and the keywords your competitors win — then send the full report to your inbox.