← All articles

Buyer Personas Built From Evidence (Not From a Workshop)

Buyer Personas Built From Evidence (Not From a Workshop)

Key takeaways

  • A persona is only useful if it changes a decision. Age, job title and a stock photo change none.
  • Nielsen Norman Group's position is blunt: design for the customers you actually have, not the ones you hope for.
  • Build from verbatim evidence — sales calls, support tickets, reviews, lost-deal notes. Not from a whiteboard.
  • The highest-value field is the job they hired you to do, and the second is the alternative they nearly chose.

Why most personas are never opened twice

The standard persona is produced in a workshop, given a name and a stock photograph, and described by age, job title, and a list of goals nobody disagrees with. It is then never referenced in a real decision, because nothing in it tells anyone what to do differently.

The Nielsen Norman Group's guidance on personas is unambiguous on the root cause: you should design for the customers you actually have, not for dreamed-up customers you might hope to have. Workshop personas describe hopes. Evidence-based ones describe buyers.

The test is simple: if two options are on the table, does the persona pick one? If not, it is a character sketch.

Where the evidence already exists

You almost certainly do not need new research to start. Five sources already exist inside most businesses:

  • Recorded sales calls. The objections and the questions, verbatim.
  • Support tickets. Where the product confuses people, in their own words.
  • Won/lost notes. Especially lost — what they chose instead.
  • Reviews, yours and competitors'. Unprompted language about what mattered.
  • Search queries that bring people to you, which is unfiltered demand language.

Read fifty of these before writing anything. The persona should be assembled from phrases you can point at, not adjectives you agreed on.

Same page count. Only one changes a decision. WORKSHOP PERSONA “Marketing Mary”, 34 Job title: Marketing Manager Goals: grow the business Frustrations: too busy Drinks oat lattes Decides nothing EVIDENCE PERSONA Job: stop rekeying invoices Trigger: bookkeeper resigned Alternative: hiring a temp Decides on: setup time Says: “I just need it to stop” Picks the headline for you
If the persona does not choose between two options, it is a character sketch.

The fields that change decisions

FieldWhat it decides
The job they hired you forYour headline, and what the product page leads with
The trigger eventCampaign timing and which lists are worth touching
The alternative consideredYour comparison content and objection handling
The deciding criterionWhich proof to put above the fold
Their own words for the problemKeywords, ad copy, subject lines
Who else must agreeWhether you need a one-pager for someone who never speaks to you

Note what is absent: age, gender, seniority, hobbies. Not because demographics never matter — they matter for ad targeting — but because they rarely change what you say.

Pro tip

Record the trigger event above everything else. Most buying is triggered by a change — someone left, a tool broke, a regulation shifted, a target moved. Knowing the trigger tells you when to reach people, which is usually worth more than knowing who they are.

Building one in an afternoon

  1. Pull 20 recent customers, including three you lost.
  2. For each, extract six verbatim quotes against the six fields above. If a field has no quote, leave it empty — do not fill it in from memory.
  3. Cluster on the job, not on job title. Two different roles buying for the same reason are one persona.
  4. Name each cluster after the job: "the owner whose bookkeeper just left", not "Finance Fiona".
  5. Write one page per cluster, with quotes attached and dated.

Empty fields are informative — they show what you do not actually know, which is precisely what workshop personas hide by inventing.

Using it, so it stays alive

A persona survives only if something references it routinely. Attach it to three artefacts: the landing page brief, the ad copy brief, and the sales call opener. If a new page cannot say which persona it serves and which job it addresses, that is the signal to revisit it.

The AI Marketing Automation Vault includes an ICP builder, a buyer persona generator and a voice-of-customer pain-mining workflow among its 50 systems, all built to run from evidence you already hold rather than from a blank template. Pair it with the Sales Vault when the same personas drive outbound, or take the bundle. Related: a competitor analysis framework and landing page copy that converts, which is where a good persona shows up first.

What should a buyer persona include?

Six fields that change decisions: the job they hired you to do, the trigger event, the alternative they considered, the criterion they decided on, their own words for the problem, and who else had to agree. Demographics matter for ad targeting but rarely change what you say, which is why they belong lower down.

How many buyer personas should we have?

As many as there are distinct jobs, which is usually two or three — not one per job title. Two different roles buying for the same reason are one persona; the same job title buying for two different reasons is two. Cluster on the job, not the org chart.

Can personas be built without doing new research?

Usually yes. Recorded sales calls, support tickets, won and lost notes, reviews, and the search queries that bring people to you are all existing evidence in customers' own words. Read fifty of them before writing anything, and leave fields empty where no quote supports them.

How do I know if a persona is actually useful?

Put two real options in front of it — two headlines, two campaign timings, two proof points — and see whether it picks one. If it does not discriminate between options, it is a character sketch and will not be opened again after the workshop that produced it.

Sources

  1. Nielsen Norman Group — Personas: Study Guide — the evidence-based position quoted here, including designing for the customers you actually have.
  2. Nielsen Norman Group — personas research collection — the broader body of persona research and application guidance.
  3. Christensen et al., “Know Your Customers' Jobs to Be Done”, Harvard Business Review — the jobs framing behind clustering on the job rather than the job title.
  4. UK Government Service Manual — user research — a public, practical standard for grounding user models in evidence rather than assumption.

Want this working
in your business?