Mixed pairs — one person from a customer-facing team, one from a team that never sees a customer — attempt real tasks in your own product or service from scratch, and write down every point of friction. It bonds the pair, and it produces a defect list that is usually embarrassing in a useful way.
What it is good for
- CollaborationWorking a shared problem where no one person holds all the information.
- Problem solvingStructuring an ambiguous problem under a time limit.
- Cross-team networkingConnecting people who would otherwise never share a meeting.
- CommunicationPractising clear hand-offs, active listening and saying the hard thing early.
What you need
- A clean device or fresh account per pair
- A one-page observation sheet
- A shared place to post findings
The run sheet
Write five real tasks (prep)
Things a genuine customer does in their first week. Specific and outcome-shaped: 'invite a colleague and give them edit access', not 'explore the settings'.
Pair across the org chart (10 min)
Deliberately pair finance with support, engineering with sales. The mix is the team building; the tasks are the excuse.
Fresh accounts, no shortcuts (5 min)
New account, no internal tooling, no admin access, no asking a colleague. The moment someone uses a back door the exercise is over.
Run the tasks, narrate out loud (60 min)
One person drives, the other writes down every hesitation, wrong turn and moment of doubt. Swap after task three. Record timestamps where you can — 'stuck for four minutes' is more persuasive than 'confusing'.
Post findings to one shared place (15 min)
One line per friction point, no solutions yet. Solutions in the room will crowd out findings.
Read the wall together (20 min)
Group the findings. Then — and only then — talk about what to do. Assign owners to the top three before anyone leaves.
Facilitator notes
- The pair is the unit of bonding here. Two hours of shared mild frustration builds a working relationship faster than two years of passing in a corridor.
- Protect people from defensiveness: say at the start that findings are about the product, never about the team that built it, and mean it.
- If the top three findings do not get owners and dates, do not run this again. The team will have noticed.
Debrief questions
Ask these immediately afterwards, not next week. Fifteen minutes of these is what turns the activity into something the team refers back to — see the debrief guide.
- What surprised you most about doing this without your usual access?
- What did your partner notice that you walked straight past?
- Which finding would you be most embarrassed for a customer to hit?
Variations
Competitor edition
Run the same tasks on a competitor's product. Produces a sharper list and fewer bruised feelings.
Support edition
Pairs listen to or read real support conversations instead of using the product. Better for teams whose product cannot be freshly signed up for.
Access and inclusion
Make sure at least one task is attempted with a screen reader or keyboard only, and pair someone who uses those tools day to day with someone who does not. That single pairing changes how a team thinks about accessibility more than any training session.
Running it remotely
Works identically over a call with screen sharing. The observer's job is easier remotely because they can watch the cursor hesitate.