Organic search in a market of a few thousand buyers
B2B SEO is the practice of earning organic search visibility among people buying on behalf of an employer. The mechanics are the ordinary ones: a page has to be reachable by a crawler, held in an index, and judged the best answer to a query. What changes is the shape of the demand. A consumer category can generate hundreds of thousands of searches a month. A category like industrial process instrumentation or claims-adjudication software may have a worldwide audience of a few thousand qualified people, a few hundred of them looking this quarter.
Two confusions are worth clearing first. Search engine optimization for B2B companies is not content marketing, though content marketing produces much of what search surfaces. Nor is it consumer SEO with different words in it. Consumer search work optimises for accumulated visits, because in a large market visits are a fair proxy for demand. B2B search optimises for presence at a small number of high-consequence moments, and has to be measured differently as a result.
Why ninety searches can outrank ninety thousand
Start with arithmetic rather than a tool. If a contract is worth six figures across its term, a phrase searched ninety times a month by specifiers with a budget code can produce more revenue than one searched ninety thousand times by students, competitors and job seekers. Volume counts typing. It says nothing about who can sign.
Audience structure makes volume-led selection worse still. Gartner's May 2025 survey of 632 B2B buyers, fielded in August and September 2024, describes buying teams ranging from five to 16 people across as many as four functions. Each function brings its own language: the engineer searches for a tolerance, the finance lead for a total cost comparison, the security reviewer for a certification. Gartner also finds the process is not a funnel — buyers loop, revisiting six buying jobs (problem identification, solution exploration, requirements building, supplier selection, validation, and consensus creation) at least once each. So volume-led selection fails twice. It picks terms typed by the wrong people, and it assumes a sequence the research does not support.
What engineering calls it, what the buyer types
Most wasted B2B SEO effort begins with a keyword list written by the people who built the product. Engineering names things by mechanism, procurement by category, and the buyer by the problem sitting on their desk.
| Who is naming it | Typical phrasing | What it signals |
|---|---|---|
| Engineering | rotary positive displacement blower | Mechanism and specification, from someone who knows the category |
| The operating buyer | low pressure air for wastewater aeration | A problem to solve; the category is not yet chosen |
| Procurement | blower package supplier, approved vendor list | A shortlist is forming and terms are being compared |
The gap closes by listening rather than guessing. Recorded sales calls, support tickets, the questions inside recent requests for proposal, and your own site search log carry the buyer's phrasing verbatim. A live Google Ads campaign helps too: Google documents that "By default, all keyword match types are eligible to match to close variants. There's no way to opt out," so the search terms report surfaces adjacent queries you never thought to bid on. Read it for language, not only for negatives.
That collected vocabulary is also where the uncontested positions are. Competitors optimise the same head terms at the top of a volume-sorted export and ignore everything specific, leaving failure modes, part and model numbers, standards references, integration pairs and end-of-life queries open. Search each phrase and read the results: forum threads and distributor catalogue pages mean nobody has claimed it.
What has to be true before any of this works
Organic search distributes something; it does not create it. When a programme underperforms, the cause is usually one of these dependencies rather than the search work itself.
- Pages that answer one specific question. A single Solutions page covering nine product lines cannot rank for any of them.
- A site a crawler can read. Content that appears only after a user interaction, key pages disallowed in robots.txt, or a template giving every page the same title will cap results before content quality matters.
- Performance measured on real users. The three Core Web Vitals are LCP (within 2.5 seconds), INP (200 milliseconds or less) and CLS (0.1 or less), assessed at the 75th percentile of real page loads and segmented separately for mobile and desktop. A perfect Lighthouse score is not proof of passing: Lighthouse cannot measure INP at all, and low-traffic B2B sites often have no field data to assess.
- A path from form fill to CRM record. Without it you will report sessions to a board that asked about pipeline.
The expensive mistakes
- Sorting the keyword export by volume and drawing a line. It is fast and produces a defensible-looking deck. Sort by plausible revenue instead, however crude the estimate, and be willing to target a two-digit phrase.
- Publishing broad educational content that ranks for people who will never buy. Definition pages attract students and competitors. Write them only where they serve your buyer's problem-identification stage, and judge them on assisted pipeline rather than sessions.
- Treating a first-place rank for the company name as an outcome. You rank for your own brand because you are the only entity with that name. Brand queries measure demand you already created: a useful trend line, and a worthless performance claim.
- Running paid and organic keyword research as two projects. The paid search terms report is the best query data most B2B companies hold. Leaving it inside the ads agency wastes it.
- Buying content restructuring for AI. Google states that "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in generative AI search," and names content chunking and rewriting for AI systems as unnecessary. Treat proprietary AI-visibility markup as a sales pitch.
Reporting that survives a four-quarter cycle
Report few things, consistently: rankings and impressions for the phrases you chose, phrase by phrase rather than averaged; non-brand organic sessions to the pages that matter; form fills and qualified opportunities carried into the CRM; and deal value when a cycle closes.
Then know where the numbers lie. Monthly conversion counts in a small market are single digits, so month-on-month percentages are noise — read rolling quarters. Self-reported source fields disagree with analytics and both are partly wrong. GA4 keeps user-level and key-event data for 14 months at most on a standard property and defaults to two months, shorter than a typical enterprise cycle, though that setting affects explorations and funnel reports rather than standard aggregated reports. GA4's behavioural modelling for consent-declining users needs at least 1,000 events a day with analytics storage denied and at least 1,000 daily consenting users — thresholds most B2B sites never reach, so the property absorbs the full loss from consent refusal with none of the modelled recovery a large consumer site gets.
When to spend the money elsewhere
B2B SEO is the wrong first investment in several honest situations. If the category is genuinely new and nobody has a name for what you built, there is no demand to capture, and the money belongs in demand creation, analyst relations and direct outreach until a vocabulary exists. If contracts are awarded through framework agreements and approved-supplier lists, organic visibility may influence a shortlist but will not create one. If the site cannot change for two quarters because of an unfinished replatform, wait: publishing into a site you are about to migrate pays for the work twice.
Where it does apply, the sequence is unglamorous. Collect real buyer language for two weeks before opening a keyword tool. Map it to the six buying jobs and pick the two where you are weakest. Build or rewrite one page per job, deep enough that a specifier would forward it. Fix the technical faults that cap the whole domain. Then wait a quarter before judging anything.
Frequently Asked Questions
How long does B2B SEO take to show results?
Expect leading indicators within one to two quarters and revenue signal only after a full sales cycle has run. Impressions and rankings for the phrases you targeted move first, often within weeks of a page being indexed and improved. Form fills follow. Closed revenue lags by whatever the sales cycle is, which in enterprise B2B is frequently three to four quarters.
Agree the reporting up front: which leading indicators count as evidence early, and when the pipeline question becomes fair to ask.
Is it worth targeting keywords with very low search volume?
Yes, when the searchers are qualified and the phrase describes something you sell. In business markets the addressable audience is often a few thousand people, so a phrase reported at 10 to 90 searches a month can represent a real share of everyone who will ever look. Volume figures are modelled estimates in any case, and Google Ads reports no Quality Score at all — a dash — when there are too few exactly matching searches.
The test is not volume. It is whether the person typing could plausibly sign or specify a purchase.
Should our engineers or our marketing team choose the keywords?
Neither alone. Engineering supplies accuracy — terminology, standards, model numbers, failure modes — and will reliably produce a list nobody outside the company searches for. Marketing supplies the buyer's phrasing and the judgement about which phrases deserve a page. Assemble the list from recorded sales calls, support tickets and paid search query reports, then have engineering check it for factual correctness rather than draft it.
One rule keeps it honest: every phrase must have been observed somewhere real, not invented in a meeting.
Do we need to do anything special to appear in AI Overviews?
No, according to Google's own documentation. Google states that its AI features in Search are grounded in core Search ranking and quality systems, and that "There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary." No new machine-readable files, AI text files, markup or Markdown are needed, and Google names content chunking and rewriting for AI as unnecessary.
What is documented is the mechanism: retrieval from the Search index plus query fan-out across related subtopics. Claims that a proprietary schema or an llms.txt file drives inclusion are unproven.