What if our catalog genuinely serves very different buyers?

Then you may need more, but verify with the treatment test. A store selling both baby gear and pet supplies might justify six or seven. The question is never how different the buyers are in theory; it is whether you will treat them differently in practice.

Should personas be based on surveys or behavior?

Behavior first, surveys for color. Behavioral data tells you what people do; surveys tell you why they think they do it. Build the segments from purchase and browsing behavior, then use interviews to give each segment a human voice.

How often should we rebuild personas from scratch?

Rarely. Review quarterly, merge and split as needed, and rebuild only when the business fundamentally changes: a new category, a new geography, or a new customer base. Stable personas compound in value because the team learns them.

PersonaRail answers

How many personas does a store actually need?

Persona projects have a predictable failure mode: they start with four crisp archetypes and balloon into twelve overlapping segments that nobody on the team can keep straight. Marketing writes to all of them, which means it writes to none of them. The uncomfortable truth is that most stores need three to five personas, and the marginal sixth persona almost always adds more confusion than revenue. The right number is not a guess; it is the number your team can actually act on, and there are concrete tests for when to split a persona and when to kill one.

Why fewer beats more

A persona earns its keep when it changes a decision: different homepage hero, different email cadence, different product recommendations. If two personas would get the same treatment across every touchpoint, they are one persona wearing two names. Most stores find that three to five personas produce genuinely different treatments, and beyond that the differences become academic.

There is also a maintenance cost. Every persona needs its entry criteria reviewed, its content kept fresh, and its performance measured. Twelve personas means twelve of everything, and in practice it means none of it gets maintained. Personas rot fastest when there are too many to care about.

The three personas almost every store has

Strip away the industry specifics and most stores serve the same three archetypes. The Researcher compares carefully, reads reviews, and buys after deliberation; they need depth, comparisons, and risk reducers. The Replenisher buys the same things on a rhythm; they need shortcuts, subscriptions, and reminders. The Gift-Giver buys under time pressure for someone else; they need curation, deadlines, and easy returns.

These three behave so differently that they justify different treatments everywhere: different email timing, different homepage modules, different post-purchase flows. If your persona set does not include recognizable versions of these three, you are probably missing the highest-value splits.

When to split a persona in two

Split when the two halves would get meaningfully different treatments and each half is big enough to matter. The test is concrete: write down what you would do differently for group A versus group B across email, site, and ads. If the list is long and specific, split. If you struggle to fill it, do not.

The most common justified split is by price sensitivity within a persona. Bargain-hunting researchers and full-price researchers look identical in browsing behavior but respond to opposite incentives; sending discounts to the full-price half trains them to wait. Another common split is new versus returning within the gift-giver persona: the first-time gift buyer needs reassurance, the repeat one needs speed.

When to merge or kill a persona

Merge when two personas have converged in behavior. Personas are hypotheses, and hypotheses get tested: if the bargain researcher and the full-price researcher end up converting on the same offers at the same rates, the distinction was never real. Kill when a persona is too small to treat differently. A persona representing 3 percent of revenue does not justify its own email track; fold it into the nearest neighbor.

Review the set quarterly. Personas drift as the catalog, the customer base, and the economy change. The review is simple: for each persona, name the decisions it changed last quarter. The ones with no answer are candidates for merging. Ruthlessness here is a feature; a persona set that never shrinks is a persona set nobody trusts.

Making a small set stick

A small persona set only works if the whole team uses it. Give each persona a name the team will actually say out loud, one defining behavior, and one rule for how to treat them. Print it on one page. If it does not fit on one page, you have too many personas or too much description.

Then wire the personas into the tools, not just the slide deck. The email platform, the site personalization, and the ad audiences should all speak the same three to five labels. A persona that exists only in a strategy document is decoration. A persona that triggers real treatments is infrastructure.

Reviewed

Published Oct 3, 2026.