How to recruit user research participants when you cannot find customers

On-Off Group

October 2026

To recruit user research participants, write a short screener that defines who counts, then pull names from lists you already hold: recent customers, branch walk-ins, support queues, waitlists. Book six or seven to get five in the room, pay them, and be clear about what you will record.

If you have been told to test the onboarding flow with real customers and you cannot work out where those customers come from, this is for you. It is written for people running research in banks and insurers, where the customer list exists but is not yours to email.

You will get a way to recruit user research participants from lists your company already holds, a short screener, and a plan for the people who do not turn up.

You do not need to hire a research company or raise a purchase request.

Write the screener before you look for anyone

The usual version: Someone says find me five customers, and a coordinator starts phoning whoever answers. Three of the five turn out to be staff relatives. One has never used the app. The session happens, everyone is polite, and nothing in the recording is usable.

So write down who counts first. Three or four questions, no more. For an onboarding flow it might be: have you opened an account with any bank in the last six months, do you use a smartphone for banking, did you complete the application or drop out, what device will you have with you.

That last one matters more than people expect. If the flow breaks on a five year old Android and everyone you recruit has an iPhone, you will not see it.

Keep a mix. Two who completed, two who dropped out, one who has never tried. The drop-outs tell you the most and are the hardest to get, so start there.

Use the lists you already have

People go looking outside the company first. The names are usually already sitting in a few systems inside it.

Recent applicants, including the abandoned ones. Support tickets and call logs from the last month. Branch walk-ins. The waitlist for a product that has not launched. People who filled in a form and never heard back. Any of these is a warmer source than a panel, because they have already touched the thing you want to test.

The list usually exists. The hard part is that another department owns it. Ask the owner for a small, specific slice: thirty names from last month, one call each, a fixed script. A request that size gets a yes far more often than a general request for access to customer data.

Branch staff are often the fastest route. They can ask people in the queue instead of waiting on emails.

Recruit people who actually do the task

When customers really are out of reach, the fallback should not be your own team. Colleagues know the product and the shortcuts, and they want the session to go well. They fill in the form correctly and tell you nothing.

Look for people close to your customers instead. Someone who applied to a competitor. A customer of the same type who uses one of your other products. A relative of a customer who is in the right income band and has never used your app. For a business banking flow, a sari-sari store owner or a small trucking operator will get you closer than a product manager will.

Branch staff are worth including for a different reason. They watch customers get stuck all day. Test the assisted application with branch staff, since they do that part. Test the screens customers use with customers.

If you are unsure how to run the sessions once you have the people, we wrote up the format in How to run a usability test with five real users.

Pay people, and say what you will record

If you do not pay, the only people who show up are the ones with time to spare. That skews everything.

Pick an amount, make it a voucher or load rather than a promise of one, and say it in the first message. Then say plainly what happens: how long, where, that you will record the screen and the voice, that the recording stays inside the company, and that they can stop whenever they want.

In financial services, get compliance and data privacy in early. Not after you have booked the sessions. Agreeing the consent form, and how long you keep the recordings, takes about a week the first time and five minutes after that. Skip it and you end up with five good recordings nobody is allowed to show.

Write the consent wording once and reuse it. The second round of sessions then does not need approval again.

Over-recruit and hold the slots

People do not turn up. Traffic, or work, or a sick child. Book six or seven to end up with five.

Confirm twice: once when you book, once the day before. Send the link or the address in the same message, not in a separate email nobody finds. Keep one floating slot late in the day for whoever reschedules.

Run the sessions in one block over two or three days. Spread them over a month and people stop attending, and you end up reading a summary instead of watching someone get stuck.

While you are booking, get your baseline. Time to complete the flow, drop-off rate at each step, average turnaround time on an application. Without those numbers you will have a highlight reel and no way to prove anything changed. A large insurer ran five-day design sprints on journeys that used to take six months or more. Those journeys were designed, tested with customers and live in four weeks. Completion rates rose 80 percent. They had the old numbers written down, so they could show what changed.

If you would rather run this regularly than as a one-off, we compare the two approaches in One-off customer research or monthly insight.

What to do this week

Write the screener. Four questions, including the device they will bring.
List every source of customer names inside the company and note who owns each one.
Ask one owner for thirty names from last month, with a fixed call script.
Send the consent wording to compliance now, before you book anyone.
Write down the current numbers for the flow you are testing, even if they are rough.
Book seven people into two days, pay them, and confirm the day before.

Questions people ask

How many participants do we need?
Five people using the flow while you watch will show you most of the serious problems. Booking six or seven covers the ones who do not turn up.

Can we test with staff instead of customers?
Branch staff are useful for the parts of the process they handle themselves, but they know the product too well to stand in for a customer filling in an application for the first time.

What do we pay participants?
Pay something real, in a form people actually want, such as a load or grocery voucher, and say the amount up front. Unpaid recruitment quietly selects for people with spare time.

How long does recruitment take?
If you are drawing on a list you already hold, a few days of calls is usually enough for five sessions. Waiting for a panel agency or procurement adds weeks.

On-Off Group trains teams, tests products with real customers, finds where a transformation has stalled and builds what gets it moving, for banks, insurers and enterprises in the Philippines, since 2015. How we help with usability testing.