UI Elements
Imperative or First-Person Button Copy A/B Test Idea
An imperative label such as “Start” gives an instruction; a first-person label such as “I’m starting” can sound like the visitor’s own decision. Either may feel natural or awkward depending on context and language. Compare two accurate labels for the same action without assuming a lift, and follow clicks through to completion.
A/B test scenario. This page is a test plan, not the result of a real experiment.
- UI elements
- Action completion rate
Option A vs Option B
Two alternatives; neither is the incumbent.
- Changed
- Options
- Page
- UI elements
- Difference
- changed
Hypothesis
"Start" may read as a short instruction telling visitors what to do, while "I'm starting" may read as a decision they made themselves. Which one leads to a higher action completion rate depends on context and the page's language; the label that sounds more natural on the page could carry more visitors through the action.
Test plan
When to run this test
Consider comparing the two verb forms when the label seems not to fit the page, such as usability sessions where people find it stiff or bossy, survey comments about its wording, or few clicks relative to impressions. If one label feels more natural, more visitors in that arm complete the same action and fewer leave. If people drop off after the click, look at that next step first; a label will not fix it. Run the test only on buttons that do not complete an order, and never let first-person copy imply a decision the user has not made; a paid action keeps its confirmation step.
How to measure success
Action completion rate
Compare completion of the same action among eligible assigned visitors. Clicks are diagnostic, not the success outcome.
Also track
- Purchase conversion rate (CR)Compare completed actions per eligible assigned visitor; clicks alone are not success.
- Page exit rateCompare page exits after the CTA appears, using the same window.
- Repeat visit rateCompare assigned visitors over the same follow-up period; this is not a direct measure of brand perception.
How to set it up
- Verb formCompare an imperative label with a natural first-person label for the same action.
- ContextKeep the action and actual commitment the same; other commitment levels require a separate test.
- Brand toneChoose wording that fits the established tone of the page.
- FitKeep button size fixed and make sure both labels remain readable.
- Visitor historyReview new and returning visitors separately.
What to watch when reading the result
- 1Button clicks are diagnostic only; the share of visitors who complete the same action settles the test. Do not read a difference in repeat visit rate as a direct measure of brand perception.
- 2Keep button color and size identical in both arms, or you cannot tell whether the verb form or the look caused the difference. To keep the page consistent, run it on a screen with a single main button; if buttons on the same page use different verb forms, the experience becomes inconsistent and the result measures that inconsistency more than the label.
- 3The evidence covers this CTA and its action only; actions with a different level of commitment need their own test. Do not switch every button on the site at once; roll the winning form out gradually and review new and returning visitors separately.
- 4Before calling either label better, enter each arm's visitors and completions in the A/B test significance calculator to see whether the gap is beyond chance. Two labels that differ only in verb form may produce a small gap, so report an unclear result as inconclusive rather than as a preference.
Frequently Asked Questions
Should first-person wording be expected to perform better?
No. Compare two accurate labels without assuming a lift. Choose wording natural for the context, language and established page tone.
Which metric should decide the test?
Completion of the same real action per eligible assigned visitor. Clicks are diagnostic. Repeat visits are not a direct measure of brand perception.
Can I'm starting be problematic for a paid action?
If a button finalizes a paid order, its label must clearly convey the payment obligation. Run this comparison on buttons that do not finalize the order. Start or I'm starting must not conceal a financial commitment.
More tests in UI Elements
All testsFocusing a Promo Banner on a Single Message A/B Test
UI elementsCampaign action completion rate
Promo banner content under test: several benefits listed at once, or one message with one action. See both arms, the visibility trade-off and the KPI.
Highlighting One Option With Background Color A/B Test
UI elementsSelection completion time
An A/B test that changes only one option's background color to see whether visual emphasis speeds selection without harming overall completion.
Button Copy: Action Verb vs Benefit Phrase A/B Test
UI elementsNext-step completion rate
A microcopy experiment example comparing an action label such as Send with a benefit label such as Get my quote. See both arms and the trade-offs.