How long a test takes

What actually governs the clock: results land when the committed sample is reached, not on a deadline, and the test closes itself when it gets there.

Most tests fill within hours. None of them run on a clock, and that difference is worth understanding before you launch one the afternoon before a presentation.

Results land at fill, not at a deadline

You commit to a number of people per version when you launch, and that number is frozen there. The test ends when every version has its people, whenever that happens. Nothing stops early because one side looks ahead, which is the most common way a test like this misleads the person who paid for it.

When the last person lands, the test settles on its own: the credit hold reconciles against what actually filled, the people who took part are paid, and your results are ready. There is nothing to press.

What fills faster

• A general-population audience. The widest pool gathers quickest.
• Evenings and weekends, when more people are around.
• A smaller sample per version.
• A shorter piece, which is a smaller ask of anyone's evening.

What fills slower

• A narrow audience. A specific age band or gender competes for a smaller pool, and the test waits for the right people rather than the next people. See Choosing an audience.
• A large sample per version.
• A long piece.
• Weekday working hours.

Typical, and deliberately not promised

Typically a wide audience at a modest size is done in a few hours, and a narrow audience at a large size takes considerably longer. We do not put a predicted finish time on your test, because the pool changes hour by hour and a promised duration is a guess wearing a watch. What you get instead is the live count: responses per version as they arrive, so you can watch the fill rather than wonder about it.

Plan backwards from the decision

If a test has to inform a call on Thursday, launch it Tuesday rather than Wednesday night, and prefer the wider audience when the schedule is tight. If the deadline will not move, run a smaller sample: a small test that finishes beats a large one still filling when the meeting starts. Read that result with the care it deserves, because a smaller sample earns a softer claim. See Reading your report.

If a fill stalls

A fill that has crawled to a stop is usually the audience, not a fault. Three moves, cheapest first:

1. Give it another evening.
2. Widen toward the general population, or trim the sample size a little.
3. Send the test to your own viewers with a share link, which costs nothing to recruit.

If it stays stuck for far longer than any of that explains, contact us from the Support tab in the bottom-right corner of HoldFast with the test and roughly when you launched it. See Audience testing isn't behaving.