# Autocomplete Suggestions in Form Fields A/B Test

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/autocomplete-suggestions-in-form-fields-ab-test

## Hypothesis

If a suggestion list appears while visitors type in fields such as address, city or company, they can pick a value instead of typing it in full. Less typing and fewer spelling errors could mean a higher form completion rate, while picking a wrong suggestion could produce bad data.

## When to run this test

Visitors may be struggling with long or hard-to-spell values when abandonment clusters at the address or company field, the same value is saved under different spellings, or typos trigger support requests. If the suggestions help, visitors get through these fields faster and more forms are submitted without a drop in accuracy. If visitors leave because they do not see why a field is asked for, review the field copy first. Verify before launch that the suggestion source covers the eligible users well enough; otherwise the test measures its gaps, not the suggestions.

## Primary metric

**Form completion rate**: Compare successful submissions to see whether autocomplete suggestions affect completion.

## Other metrics

- **Submitted-data accuracy**: Check selected suggestions and typed values against the same accuracy criteria.
- **Form completion time**: Measure time from field entry to successful submission consistently in both arms.
- **Abandonment at suggested fields**: Track visitors who leave at a field with suggestions.
- **Correction and support workload**: Compare recorded corrections and support requests caused by inaccurate suggestions.

## What to check before and during the test

- **Suggestions**: Enable suggestions for the predefined set of fields while preserving free-text entry.
- **Keep constant**: Keep fields, validation, suggestion source and the variant's list length unchanged.
- **Accuracy**: Check selected suggestions against the same quality criteria as manual entries.
- **Input**: Verify keyboard navigation and compatibility with the mobile keyboard.
- **Separate tests**: Test suggestion count or data source separately.

## What to watch when reading the result

- Do not count a higher form completion rate as a win without checking submitted-data accuracy. If completion rises while the correction and support workload from inaccurate suggestions grows too, the variant may have sped up the form at the cost of the data.
- Nothing here speaks beyond the field set, suggestion source and list length you tested; changing the number of suggestions or the source is a separate test. Corrections and support requests arrive after submission, so keep tracking them for a while after the test ends.
- Test the completion gap between the two arms with the [A/B test significance calculator](https://alidemirbas.com.tr/en/calculators/ab-test). If no clear difference shows up, do not count the suggestion list as a winner, and note any differences in abandonment at suggested fields and in data accuracy separately.

## Frequently Asked Questions

### Should manual entry remain available when suggestions do not fit?

Yes. Preserve manual input and validate suggestion coverage for eligible users. An incomplete source affects the implemented experience. Keep fields, validation and the chosen source stable.

### Does higher completion make autocomplete the winner?

Use valid form completion as the primary outcome and monitor incorrect matches, corrections and support contacts. Faster entry is not a gain if it produces unacceptable data errors.

### How can mobile suggestion-list problems be detected?

Inspect field abandonment, keyboard navigation and mobile-keyboard compatibility. Compare completion time using identical events to determine whether suggestions reduce or add work.