Audience research
Part of Build a repeatable audience research method for UK paid search
Why PPC agencies buyer personas need better data sources
Buyer personas for PPC agencies often rest on guesswork. This guide sets out which public data sources are worth using and how to keep personas current.
What to take away
- Three public sources carry most of the weightthe ONS Internet users dataset, Google Ads help on ad rank and Quality Score, and the Digital Markets, Competition and Consumers Act 2024.
- Map one buying question to one source, and delete any persona line you cannot evidence.
- Run three to five personas, each named by decision role rather than job title.
- Review quarterlyONS publication dates, Google Ads help revisions, CMA guidance and your own call notes.
Where persona research goes wrong
A job title from a LinkedIn export tells you little about how someone buys paid search management.
Most paid search decisions involve more than one person. A marketing manager shortlists, a finance lead approves spend, an operations director owns the account afterwards, and each wants different evidence.
For a repeatable way to split those roles, the audience research method for England segments buyers by decision role rather than seniority. Every persona line must answer what the buyer needs before releasing budget.
[figure 1]
Weak vs useful persona data
Weak data
- Job title
- LinkedIn export
- Sector
- Sector code
- Company size
- Size band
- Buyer count
- One buyer
- Evidence
- None cited
Useful data
- Job title
- Decision role
- Sector
- Buying triggers
- Company size
- Budget authority
- Buyer count
- Multiple roles
- Evidence
- Public source
Three sources that earn a citation
| Source | Exact document | Persona question it answers | Refresh signal |
|---|---|---|---|
| ONS | Internet users dataset, in the Internet access: households and individuals release | Is this buyer reachable, and on which device? | Annual, usually August |
| Google Ads help, About campaign settings | Which cost measures will the buyer challenge? | Changes without notice | |
| Parliament | Digital Markets, Competition and Consumers Act 2024 | Which compliance duties touch this buyer? | CMA guidance and enforcement |
Name the ONS release, not just the page
The dataset is titled Internet users and sits inside the ONS release Internet access: households and individuals. Its tables break use down by age group, sex and frequency.
The ONS reported that 96% of adults in Great Britain were recent internet users in 2024. Use it to test channel assumptions rather than buyer motives. If your persona is mobile-first, the device breakdowns should support the claim.
What the Google Ads pages give you
The About campaign settings page explains how expected click-through rate, ad relevance and landing page experience feed into Quality Score, ad rank and cost per click.
Budget holders argue in those terms. A persona that cannot speak to cost per acquisition fails in a finance call.
The DMCC and its enforcement powers
The Digital Markets, Competition and Consumers Act 2024 gives the CMA direct enforcement powers over consumer protection law, with penalties of up to 10% of global turnover. Drip pricing and fake reviews are named targets.
If a persona includes a compliance or legal stakeholder, that enforcement point belongs in their objections list.
What rival agencies claim shapes expectations before the first call. The competitor research method for UK search explains how to map that category context without copying rivals wholesale.
[figure 2]
What each source answers
ONS Internet Users
- Covers
- Who is online
- Answers
- Channel reach
- Use for
- Sanity-check mix
- Limitation
- Not PPC buyers
Google Ads docs
- Covers
- Ad rank metrics
- Answers
- Cost efficiency
- Use for
- CPA objections
- Limitation
- Not buyer intent
DMCC Act 2024
- Covers
- Digital regulation
- Answers
- Compliance questions
- Use for
- Legal stakeholders
- Limitation
- Sector-specific
Example: a completed persona card (figures illustrative)
- Decision role
- finance lead at a UK retailer with about 40 staff, approving marketing spend monthly.
- Budget
- an illustrative £4,000 monthly paid search budget, not a survey figure.
- Question asked
- how quickly will paid search show incremental return?
- Objection, in their words
- "the last agency reported clicks, not sales."
- Source behind the claim
- the Google Ads help page on campaign settings, ad rank and Quality Score.
- Owner and review date
- head of paid search, next quarterly cycle.
Before that objection goes into a persona, check what attribution methods can and cannot show so you promise only what you can measure.
Turn sources into questions
- Draft the question in the buyer's words, for examplewhich devices are our customers actually on?
- Match it to one sourceONS for reach and device, Google Ads help for cost measures, the DMCC for sector duties.
- Write the answer as a single sentence naming the source and the edition. Delete any persona line without one.
In practice: ONS device breakdowns answer whether a mobile-first message lands, Google Ads help answers what the finance lead will query, and the DMCC answers whether the buyer carries compliance risk.
Quarterly review: what to check
- ONS release calendarhas a new Internet users edition appeared, and does it shift the device or age split?
- Google Ads helpconfirm the campaign settings and ad rank page still lists the same quality components.
- DMCC and CMAscan for new guidance or enforcement notices touching this buyer's sector.
- Call notescompare objections heard this quarter against the persona's list.
- Ownerconfirm the named reviewer signed off the changes.
Update mid-cycle if regulation moves or campaign data contradicts a core assumption.
Common questions
Which ONS dataset should I start with?
Start with the dataset titled Internet users, inside the release Internet access: households and individuals. It gives reach, age and device breakdowns.
Can personas be built without client interviews?
Yes, but they will be thin. Public sources give the baseline; call notes give the wording that makes a persona usable.
How many personas does an agency need?
Three to five. Fewer usually means distinct decision roles have been merged. More than five is hard to maintain each quarter.
What is the biggest mistake in persona research?
Treating it as finished. Set an owner and a review date, then update as regulation and campaign results change.
