Start from who really comes
Think of the last ten people who bought from you, or the segments in your analytics. If most come from a phone, most of the panel should use a phone. If half are first-time visitors, half the panel should be new to the site.
The library
The library has 18 ready-made people in five groups: ready to buy, comparing options, cautious, new here and buying for work. Each comes with a short description, a device and a browser language. Pick from the groups your customers belong to, not one of each.
Five traits change behaviour
Every persona has five traits: intent (just browsing, considering, ready to buy), tech comfort (new to this, everyday user, expert), patience (impatient, average, patient), trust (trusting, neutral, wary) and budget (tight, not a factor, comfortable). They are not labels for the report: they change what the persona does. An impatient persona leaves when the next step is not obvious within seconds. A wary one looks for who runs the site, reviews, refund terms and contact details before it goes on.
Device and language
A persona on a phone gets the phone layout of your site: the menu behind a button and less on each screen. A persona with a German browser gets the German version, if your site chooses the language from the browser. Use both to test the pages your customers really see.
Build a panel from your own notes
Paste customer segments, interview notes or support complaints, and the panel is generated from them, keeping their proportions: if a third of the complaints come from older customers on tablets, so does a third of the panel. Remove names, emails and phone numbers before you paste. With no notes yet, a panel can be suggested from your home page.
How many
Five to ten personas already show the problems that stop many people, and the free credits of a new account are enough for a first panel. After a fix, repeat the same panel with the same task, so the scores of the two runs can be compared.
A mismatch we made ourselves
In one of our own runs on senaste.ai we left shop personas in the panel. A gift shopper scrolled our home page for twenty steps looking for a product and a delivery date, on a site that sells neither. The note was honest, but it told us about our panel, not about our site.
