B2B Website Content Strategy

A Set of Decisions, Not a Production Schedule

Content strategy is the set of decisions that determine what a website publishes: which questions this company is the best available source on, what depth each deserves, who is qualified to write them, what the site refuses to publish, who reviews what already exists, and when a page is retired.

Every one of those decisions has to be made by someone and defended.

A content calendar is a production schedule. It answers when a thing will be published and by whom. It never answers whether the thing should exist. Most organisations have the calendar, call it the strategy, and are consequently unable to decline anything — which is how a corporate site accumulates two hundred pages no buyer needed.

The working test is refusal. If a request from a product manager for a new page cannot be declined by reference to a written document, there is no strategy, only a queue.

Small Markets Change the Arithmetic

Consumer publishing can chase volume and let the numbers arbitrate. Business publishing cannot, for three structural reasons.

The addressable market is small. If four thousand companies could buy what you sell, a page read by two hundred people — six of whom sit on evaluation committees — can be the most commercially valuable page on the site. Traffic reporting will rank it near the bottom, and any strategy driven by session counts will delete it.

The reader is a committee with different jobs. A technical evaluator wants specifications and failure modes. Finance wants cost structure. Procurement and legal want terms and data handling. Those are separate pages, not sections of one page, and the person who quietly kills the purchase often never contacts sales.

Feedback arrives a quarter late. Content read in February influences an enquiry in July. You cannot test your way to a content strategy at B2B volumes, because the sample does not exist and the lag defeats the experiment.

Audit What You Own Before Commissioning Anything New

Nobody should approve a new page before someone has examined the existing ones. The audit is mechanical work with a specific output.

Crawl the site and export every indexable URL. Against each, join organic entrances and sessions over at least twelve months, the queries and impressions Search Console reports, conversions attributed to it, referring domains pointing at it, and its owner. Then classify every row into one of four dispositions: keep as is, improve, consolidate, or remove.

In most audits of established corporate sites the finding is the same. A minority of pages account for nearly all organic entrances and nearly all conversions, and a long tail has produced no organic entrance in a year and no conversion ever. That is what happens when publishing decisions are made one request at a time.

Record four things no traffic report shows: near-duplicate pages, sets of pages competing for one query, orphan pages with no internal links, and PDFs ranking in place of the page that should rank.

One Page, One Job, One Audience

Give every page a written purpose statement before it is commissioned: the question it answers, the single reader it answers for, and what that reader should do next. The three fields force decisions teams usually skip. The model separates topic from page purpose — what this page does about that topic — and from audience, which member of the buying committee it is written for.

The consequence is arithmetic: two pages answering the same question for the same reader are one page. Applied consistently, that rule prevents most internal competition and gives the audit its classification criteria.

Depth as a Deliberate Choice

A small number of thorough pages outperforms a large number of shallow ones. Three arguments support that, and they are not equally strong.

Internal competition is the strongest. Four thin pages on one subject split internal links, inbound links and topical signals across four URLs, and hand a search engine a choice you did not intend to offer. It is not reliably fixable afterwards: Google documents that indicating a canonical preference is a hint, not a rule, and that Google may choose a different page as canonical than you do. Consolidation plus a permanent redirect is deterministic. A tag is a request.

The reader argument is nearly as strong. A buyer comparing three vendors reads the page that answers the question completely, then forwards it to a colleague.

The crawl argument is weakest and routinely overstated. Crawl-budget management matters at large scale; a five-thousand-page corporate site is not large by search engine standards. If someone is selling you consolidation on crawl grounds alone, internal competition is the argument doing the real work.

Depth is not length. A nine-hundred-word page that answers the question completely is deep. A two-thousand-word page that circles is padding.

Who Writes It, and Who Owns It Afterwards

A subject-matter expert with an editor beats a generalist writer with a brief. What makes a B2B page worth reading is the detail only someone who has done the work knows: the step people skip, the failure that shows up in month three, the specification that decides the comparison. A generalist working from a brief and three competitor pages produces a synthesis of what is already published, which by definition adds nothing and cannot be cited.

The practical model is cheaper than commissioning blind. Record a forty-five minute interview with the expert. The editor drafts from the transcript, supplies structure and cuts, and returns it for correction.

Governance stops that decaying. Name an owner for each section of the site — a person, not a department. Set review cycles by volatility: annually for explanatory material, on-event for anything referencing statutes, versions or product detail. Make it a rule that no page publishes without a named owner and a next-review date.

Pruning, Consolidating, and the Courage to Delete

Removing pages is the highest-return content work available on most corporate websites, and the hardest to get approved. Three dispositions cover almost every case.

• Consolidate. Merge the useful material into the stronger page and point a permanent redirect at it. This is the answer for near-duplicates and for pages competing on one query.
• Redirect. Retire the page and redirect to the closest genuine equivalent, which may be a section page. Not the homepage, which tells a search engine that no equivalent exists.
• Remove. Return a 404 or 410 where there is no equivalent and no inbound link worth keeping.

What never works is the fourth option people choose: leave it, it is harmless. A page competing with the page you want ranking is not harmless.

Two facts to have to hand: Google classifies 301 and 308 as permanent redirects and treats a permanent redirect as a strong canonicalisation signal toward the target, while sitemap inclusion is a weak signal; and Google's site-move guidance is to keep redirects for as long as possible, generally at least one year. Keep a pruning ledger recording what was removed, when, why, and where it points.

Measurement, and What Google Documents About AI Features

Report by page and by topic cluster rather than by volume of output: organic entrances to the pages you decided matter, the queries on which you appear at all, conversions with the attribution model named, how often sales sends a page to a prospect. Ignore total page count, words published, and average time on page for a reference page.

Two traps are specific to small properties. GA4 data thresholds are system-defined, cannot be adjusted, and Google does not publish the numeric threshold, so demographic and query rows are routinely suppressed on low-traffic B2B sites. And because content read in one quarter influences an enquiry two quarters later, a pruning project needs an observation window agreed in advance.

On the machine-readable dimension, be careful, because a great deal is being sold here. Google documents that its AI features are grounded in core Search ranking systems using retrieval-augmented generation and query fan-out, states that you do not need to create new machine-readable files, AI text files, markup or Markdown to appear in generative AI search, and says structured data is not required for it. Search Console gained a generative-AI performance report in June 2026, reporting impressions rather than clicks. Claims that a particular formatting pattern causes inclusion are unproven, and Google explicitly says content chunking is unnecessary.

Frequently Asked Questions