A thousand users before launch!
We develop personas inspired by your actual users, clone variants, and set them loose on your product. They report what they did and what they thought of it.
TRY SUPERR_testerBuilt from your users, then cloned by the thousand.
A persona is what a user knows, what they want, and what is in their way. We write them from your interviews, tickets, and analytics, and you review the roster before anyone is cloned.
- They differ in what they know and what they are allowed to do, not in their hobbies.
- Every persona carries its sources. You can open any one of them and ask it questions.
- If you have real research or usage data, the roster is built to match it.
- knows
- the menu by heart, not the app
- wants
- one order, ready at 12:30, no phone calls
- blocked by
- no company card on file
- knows
- nothing about the menu
- wants
- something in ten minutes
- blocked by
- will not create an account
- Rachel V.
- Marcus O.
- Nina P.
- Ana R.
- Tobias L.
- Halima S.
Nobody grades their own run.
A tester gets what a real user would have and nothing more. A separate observer scores every session against a fixed rubric, and where an outcome can be checked, it is checked.
- Hidden data cannot help a tester finish. If a real user could not see it, neither can they.
- A product defect, a tester mistake, and a broken test environment go in three different trays.
- Runs repeat with the product reset each time. How often a finding recurs is part of the finding.
What they did, and what they said.
One report, two halves: the recording of every session, and the panel explaining itself afterward.
What they did
Whether the button works is the first check, not the finding.
You see how far they got, where they slowed, where they stopped, and what they came back for.
- reached order status
- 712
- stopped at checkout
- 212
- stopped at the menu
- 61
- left on the first screen
- 15
- slowest step
- choosing a pickup time
- came back
- 380 the next day, most to reorder the same thing
What they said
They say what they liked, what confused them, and what they want next, ranked so that not everything can be important.
Answers are split by persona, so you can see which users hit which wall.
Show them two versions and they will tell you which one, and why.
- liked
- seeing the order stage without calling. every persona.
- confused
- where to set a pickup time. four of six personas, strongest among office orderers.
- asked for
- reorder the last order in one tap. a saved pickup time.
“I wanted 12:30 and could not see where to say so. I assumed it came after paying. It did not.”
A thousand testers stopping at the same step is a lead, not a statistic.
Every finding carries its replay, separates what was seen from what we think, and stays marked synthetic until a real person confirms it.
- A count in the report describes the run, not your customers.
- When you have real users, a handful of them walking the same flow tells us which findings to trust.
- If every tester agrees, we say so, and we treat it as a reason to look harder.
212 of 1,000 testers stopped at checkout. 180 of them were looking for a way to set a pickup time.
- where
- checkout, step 3 of 5
- replay
- 12 sessions attached
- recurrence
- 4 of 6 personas, every run
The pickup time is set on the menu screen and checkout does not show it. Testers assume it comes after paying.
what we think it means (business)Office orders are the largest baskets and the most likely to want a later time.
Three levels. Each one includes the one before.
You can stop at the report, at the ranked backlog, or at the shipped fix.
Test.
The run and the report. Every finding with its replay, a severity, and a next step.
Prioritize.
Findings across flows and releases, ranked by what they cost you and what they cost to fix. A backlog you can act on.
Improve.
We propose the fix, build it in a sandbox, run the same thousand against it, and show you both runs. A person approves anything that ships.
The same thousand, every release.
The personas persist, so the run repeats on each release and every finding is tracked from one to the next.
- closed
- 2
- still open
- 1
- new
- 1
What we need from you
-
A product we can reach. A staging link is ideal. Production with test accounts works too.
-
The people who know your users, or what they left behind: tickets, interviews, analytics.
-
What done means for the flow we are testing.
-
One person who can decide what to do with what we find.
Nothing gets bought, sent, or deleted on your side during a run.
If you are wondering
Keep them. They tell you the product does what the spec says. This tells you whether it did what you meant.
Better. The personas are built from it, and the findings are checked against it.
The panel can react to screens and mocks. The testers need a product that runs, so the other half of the report waits for one.
Web is where we start. Ask about the rest.
Test accounts, the least access that works, no real payments, no messages sent to anyone, and traces redacted. Your data is never pooled with anyone else’s.
No. Each one gets what a real user would have, and nothing in the application’s data can help it finish.
An hour to review the roster and an hour to walk the findings. The rest is ours.