Audience research
Part of How technical SEO audience research improves audits
Why technical SEO buyer personas need data, not guesswork
Build technical SEO buyer personas from documented sources: search demand, performance data, lawful interview notes and internationalisation rules.
What to take away
- The Office for National Statistics reported in November 2024 that 55% of UK businesses had a website. Persona work has to separate those firms from the rest.
- Technical SEO personas sit on four evidence typessearch demand, measured site performance, interview transcripts and legal constraints.
- Primary data beats desk assumption because technical buyers approve budgets against risk, not slogans.
- Every persona claim needs a source and a date, or it is decoration.
Search demand as the first layer
Keyword data shows what a market asks for, not who is asking. Pull Google Search Console query data around audits, migrations, Core Web Vitals and hreflang, cross-check volume and related terms in Ahrefs or Semrush, then group the queries by job title and company size where the data allows.
From queries to buying window
- Pull audit, migration, CWV, hreflang queries
- Group by job title and company size
- Find clusters implying a trigger event
- Note buying-cycle position, not just volume
Look for clusters that imply a trigger event: a replatform, a traffic drop, a new market. Those triggers define the buying window better than firmographics alone.
Volume tells you how many searches happen, not how urgent the need is. A busy term can carry weak commercial intent, so note where the searcher sits in the buying cycle.
For background on how audience segments and research methods fit together, read our technical SEO audience research guide for England, which sets out the wider process this article supports.
Performance data as evidence
A persona that claims a buyer cares about speed should be tested against measured page experience. Core Web Vitals set good, needs improvement and poor bands for three metrics: Largest Contentful Paint (LCP) for loading, Interaction to Next Paint (INP) for interactivity and Cumulative Layout Shift (CLS) for visual stability. Those bands give a shared language for audit conversations.
If most prospects sit in the poor band, the persona's pain is load and layout, not crawl budget. If they sit in the good band, the pitch shifts to maintaining gains during releases.
Field data from real visits and lab data from a controlled test can disagree, so label which one you used. The Chrome UX Report (CrUX) supplies field data, surfaced through PageSpeed Insights; Lighthouse and PageSpeed Insights run lab tests. A persona built on lab scores alone can overstate or understate the problem.
Record the date of every measurement. A performance figure without a date is not evidence.
Interview notes and lawful use
Interviews turn segments into people. Ask what triggered the last audit, who signed it off and what nearly stopped it. Our technical SEO customer interview notes for England cover how to run those sessions and what to record.
Personal data collected during research needs a lawful basis. The Data Protection Act 2018 sets the statutory framework for processing, and the Information Commissioner's Office (ICO) is the regulator that publishes the guidance on lawful basis, retention and individual rights. Consent, retention periods and anonymisation belong in the research plan from the start.
Consent to record is not the same as consent to quote. Keep audio on a system with restricted access, and delete it once the notes are checked.
Store quotes without names unless the participant agrees otherwise in writing. Anonymised notes still support persona claims.
Multi-market and international signals
Personas for firms selling abroad need an extra layer. Google's guidance on international and multilingual sites explains hreflang and locale handling, which tells you whether a prospect runs separate market sites or one site with translated pages.
That distinction changes the audit scope, the required skills and the likely budget. A single-market persona will misread a business operating in four countries.
Ask which markets the prospect bills from and which currencies they price in. That says more about operational complexity than a translated homepage does.
A worked example persona
A persona built on the layers above reads like this for a mid-market retailer:
- Segmentin-house SEO lead at a business running more than one market site.
- Triggera replatform, or a Core Web Vitals regression visible in field data.
- Evidencedated Google Search Console query clusters around migration and Core Web Vitals terms; CrUX and PageSpeed Insights readings labelled field or lab; anonymised interview notes on who signs off the audit budget.
- Legal notelawful basis and retention recorded under the Data Protection Act 2018 before interviews start.
- Open gapsanything without a source stays a hypothesis until it is tested.
Tools that supply the evidence
Each layer of evidence has a named source. Search demand comes from Google Search Console, Ahrefs and Semrush. Field performance comes from the Chrome UX Report through PageSpeed Insights; lab runs come from Lighthouse. Behaviour and conversion context comes from GA4. Crawl and rendering problems that shape the persona's pain appear in Screaming Frog, Sitebulb or JetOctopus. Interviews supply the rest, kept in a dated research log.
Persona checklist
- One named source and date per claim.
- Search clusters mapped to trigger events.
- Performance bands recorded with measurement dates.
- Interview notes anonymised and lawfully held.
- Market footprint confirmed before scoping work.
- Evidence gaps marked as hypotheses, not findings.
- Persona reviewed when the evidence changes.
Common questions
Short answers to the questions that come up most when teams build technical SEO personas.
How many personas does a technical SEO team need?
Most teams can work with two or three. More than that usually means the segments overlap and the evidence is too thin to separate them.
Can personas be built from keyword data alone?
No. Search data shows demand but not decision rights, budget or risk appetite, so it needs interview and performance evidence alongside it.
How often should personas be reviewed?
Review them when a source changes: new performance data, a new market, or a shift in search demand. An annual check is a reasonable default for most teams.
What makes a persona claim credible?
A named source, a date and a clear link to a buying trigger. If any of those is missing, treat the claim as a hypothesis to test.
