Checkout feedback — Experiment

Running
Traffic splitA 60% · B 40%
Goal: completion rate
Variant A

How satisfied are you with the checkout experience?

Responses412
Completion74%
Variant B

Rate your checkout experience today.

Responses274
Completion81%

Phased rollout

Stage 2 of 4
10%
25%
50%
100%

2

Variants

686

Total responses

B +7pp

Leading

Products › Experiments

Test your forms.
Ship the one
that works.

Run A/B/N experiments on any feedback form — different
questions, different flows, any traffic split you choose.

Test different question sets, wording, or form length
Fully configurable traffic split — any weight across any number of variants
Phased rollout from 10% to 100% at your own pace

How it works

Run rigorous experiments. Ship with confidence.

Experiments gives you the controls to test meaningfully — not just flip a coin. From variant design to phased rollout to winner selection, every step is in your hands.

Form variants

Different forms for different groups.

Create multiple versions of the same form — vary the questions, the wording, the flow length, or the entire structure. Each variant is a completely independent form configuration.

Test short vs long forms to find the right balance between depth and completion.

Compare different question wording to see what gets more honest responses.

Test entirely different form flows — NPS vs CSAT vs open-ended.

Form variants

3 variants
A

5-question CSAT

5Q
Overall satisfaction?What went well?What could be better?Likelihood to recommend?Any other feedback?
B

3-question CSAT

3Q
Rate your experienceWhat can we improve?Would you recommend us?
C

Single NPS

1Q
How likely are you to recommend us? (0–10)

Traffic split

Any split. Any number of variants.

Decide exactly how much of your traffic each variant receives — from a conservative 10/90 to an equal three-way split. No fixed options, no rounding. You set the numbers.

Fully configurable weights — 50/50, 70/20/10, or any other distribution.

Add a control group (no form shown) as one of your allocation buckets.

Use segments to exclude specific user groups from the experiment entirely.

Traffic allocation

Variant A

50%

Variant B

30%

Variant C

15%

Control (no form)

5%

Any split across any number of variants — down to 1% granularity.

Winning metric

You define what winning looks like.

Set the goal that matters for your specific experiment — response rate, completion rate, average score on a specific question, or your own custom metric. Results are always visible side by side.

Choose from completion rate, response rate, average score, or custom.

View variant results side by side at any point during the experiment.

You decide when to call a winner — no auto-selection, no black box.

Experiment goal

Completion rate

Response rate

Average score

Custom metric

Current results — completion rate

A
74%412r
B
81%274rLeading
C
68%198r

Phased rollout

Start small. Expand when you're confident.

Launch a winning variant to 10% of your audience first, monitor the results, then ramp up at your own pace. No pressure to go 100% until you're ready.

Define rollout stages: 10%, 25%, 50%, 100% — or any stages you choose.

Advance to the next stage manually — no auto-scheduling required.

Real user data at each stage before full exposure.

Phased rollout — Variant B

10%

Done

~840 users

No issues detected

25%

Done

~2,100 users

Completion up +4pp

50%

Active

~4,200 users

Running now

100%

Pending

All users

Pending

Advance to next stage anytime — no schedule required.

Control group

Measure lift against the baseline.

Assign a slice of your audience to a control group — they see no form at all. This lets you understand the true impact of your feedback collection, not just which variant is relatively better.

Control group users receive no form — zero interference with natural behaviour.

Compare variant metrics against a true no-form baseline.

Use segments to create permanent holdout groups across all experiments.

Experiment allocation

Variant A

5-question CSAT

45%

Variant B

3-question CSAT

45%

Control

No form shown

10%

Why use a control group?

A control group never sees a form. Compare variant metrics against natural behaviour — no form shown — to understand the true lift your feedback collection creates.

Full capabilities

Every control you need to run experiments confidently.

Experiments is built for teams who want data to drive form decisions — not gut feel. Full control over who sees what, how much traffic each variant gets, and what counts as winning.

A/B/N testing

Test any number of form variants simultaneously — different questions, flows, or lengths.

Granular traffic split

Allocate traffic to any weight across variants — down to 1% granularity.

Admin-defined goal

Set your winning metric: completion rate, response rate, average score, or custom.

Phased rollout

Start at 10% and expand manually — test safely before full exposure.

Control group

Assign a percentage of users to see no form — measure the true baseline.

Segment-based holdout

Use segments to permanently exclude certain user groups from all experiments.

Per-form experiments

Each feedback form has its own independent experiment configuration.

Side-by-side comparison

Compare variant performance on your chosen metric — you decide when to pick a winner.

Instant activation

Experiments activate immediately on save — no deployment or code change needed.

Start experimenting

Stop guessing which form works best.

Run A/B/N experiments with any traffic split, any goal metric, and phased rollout — all without touching your code.