Capability 03 — Content Strategy
Your experts know the answers. Almost none of it is on your website.
The goal is not to publish more. It is to get what your team already knows out of their heads and into the formats buyers and answer engines actually use. That is a knowledge extraction problem, not a writing problem.
The pipeline
From expertise to measured visibility.
Nine stages. Most content programmes skip the first two and wonder why the output reads like everyone else's.
- 01
Company expertise
the raw materialIdentify who actually holds knowledge worth publishing: solutions engineers, implementation leads, compliance specialists, the founder who has had the same argument with four hundred prospects. - 02
Extraction sessions
45 minutes eachStructured interviews against a prepared question bank, recorded and transcribed. We are mining for specifics: the failure modes, the numbers, the trade-offs, the thing they always have to explain twice. - 03
Search research
demand sideSearch data, Search Console queries, and the language buyers use rather than the language you use internally. - 04
Question discovery
many sourcesSales objections, support tickets, community threads, review-site comparison language, competitor coverage, and the follow-up questions answer engines suggest. - 05
Intent classification
format follows intentInformational, commercial investigation, comparison, transactional, navigational — and whether the reader is problem-aware or solution-aware. Getting this wrong produces a well-written page aimed at nobody. - 06
Content strategy
what earns a pageA prioritised map of topics to formats, with an explicit argument for each piece and what it is expected to influence. - 07
Expert-informed production
human ownedDrafting can be assisted. The substance comes from the transcripts, and a named subject-matter expert signs off before publication. - 08
Publication
structured for extractionAnswer-first, self-contained sections, entities named explicitly, facts in text, accurate structured data. - 09
Measurement
and pruningAgainst the tracked query set and organic performance — including the honest admission of what cannot be attributed.
Formats
What earns a page in B2B.
Not every topic deserves a page, and not every page deserves a thousand words. These are the formats that consistently justify the effort in a considered purchase.
Comparison and alternatives
Use-case and industry pages
Original research
Deep guides
Product documentation
Expert commentary
Standards
What we do and will not do.
How we work
- Every substantive piece traced back to an expert interview or primary source
- A named human reviewer accountable for accuracy before publication
- Competitor claims that are factually checkable and fairly stated
- Fewer, better pages, maintained and refreshed on a schedule
- Pruning and consolidating content that no longer earns its place
- Structured data that matches what the page actually says
What we refuse
- Hundreds of thin programmatic pages generated from a spreadsheet
- AI-written posts with no expert input, published at volume
- Publishing cadence presented as a result
- FAQ blocks written for schema rather than for readers
- Comparison pages that misrepresent competitors
- Keyword density as an editorial standard
The extraction session
How we get the good material out.
Experts rarely know which parts of what they know are valuable. They have stopped noticing. The interview is designed to surface exactly those parts.
- What do prospects consistently get wrong about this, and what do you have to correct every time?
- Walk me through the last implementation that went badly. What was the actual cause?
- When is our product the wrong choice, and what should they use instead?
- What question gets asked in every security or compliance review?
- What do our competitors claim that is technically true but misleading in practice?
- What would you tell a friend at another company to check before buying anything in this category?
Division of labour
Where AI helps and where it must not.
We use AI heavily in this process. We are specific about where, because the difference between assisted production and generated filler is the whole value of the programme.
Assisted
Transcript mining, question clustering, intent classification, competitor content analysis, brief construction, first-draft structure, editing passes, internal link suggestions.
Human-owned
The argument the piece makes. Any factual or technical claim. Anything about a competitor. Anything a regulator could read. Final sign-off, by a named person, recorded.
Measured honestly
We report what a piece did against the tracked query set and organic performance. Where influence on an AI answer cannot be cleanly attributed, we say so rather than claiming the credit.
See where you stand in AI search.
An AI Search Audit tells you how often AI systems name your company, who they name instead, and what is causing the gap. Every figure comes with the method behind it.