# Header Search Field vs Menu Search A/B Test

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/header-search-field-vs-menu-search-ab-test

## Hypothesis

An always-visible search field in the header may make search easier to notice and bring more visitors to products; search inside the menu adds a step but frees header space for other navigation. Which option raises purchase conversion across all assigned visitors could depend on how central search is as a path on this site.

## When to run this test

Few search queries relative to visitors, session recordings of visitors who cannot find an item through the category menu, or key links missed in a crowded header are signs that header placement shapes how people search. The outcome to look for is a change in search use or navigation in one option that carries through to purchases among all assigned visitors. If searches with no matches leave users on an empty screen, fix that first, because a weak results page holds back the effect of either placement.

## Primary metric

**Purchase conversion rate (CR)**: Compare purchases among all eligible assigned visitors, not only those who use search.

## Other metrics

- **Search usage rate**: Compare search starts among eligible assigned visitors who can see the search control.
- **Product click-through after search**: Report clicks among search users as diagnostic; the primary decision includes all assigned visitors.
- **Search success rate**: Define a successful search event before launch and measure the share of submitted queries that meet it.
- **Purchase rate after search**: Report purchases among search users as diagnostic; keep non-searchers in the primary denominator.

## What to check before and during the test

- **Search usage**: Compare the always-visible header field with search opened from the menu.
- **Search conversion**: Report conversion among search users separately; they are not equivalent randomized groups.
- **Finding a product**: Compare time from search to product click using the same events.
- **Result quality**: Define search success in advance and report all assigned visitors separately from search users.
- **Device**: Include only devices where the search placement differs and report device results separately.

## What to watch when reading the result

- Decide on purchase conversion among all eligible assigned visitors. Because placement changes who searches, purchase rate after search and product click-through after search do not compare equivalent groups; read them only as diagnostics. A higher search usage rate is not enough on its own if it does not carry through to purchases.
- In the option where search moves into the menu, keep other navigation destinations reachable and keep the mobile search icon easy to tap. Keep placeholder text short and readable in both options; otherwise you end up measuring those details rather than the placement.
- Keep the search suggestion algorithm unchanged during the test so differences in search success rate can be tied to placement. Read results only on devices where the placement really differs, one device at a time, and validate again before applying them to a different page layout.
- Because only some visitors use search, its effect on total conversion is diluted and may be small. Try the current purchase conversion rate and the relative lift that would matter to you in the [sample size calculator](https://alidemirbas.com.tr/en/calculators/sample-size-calculator), then divide the sample per variant by daily traffic per arm on eligible devices to see how long the test must run. If no significant difference appears at the end, report it as a gap this sample could not detect, not as proof of no effect.

## Frequently Asked Questions

### Why is conversion among search users not the primary metric?

Placement can change who searches, so search users are not equivalent groups. Report their conversion diagnostically. Use overall purchase conversion among assigned visitors as the primary outcome.

### What matters when search moves inside the menu?

Preserve access to other navigation targets and a usable mobile touch target. Track time to a product click after entering search, since access now requires another step.

### How should search quality be measured?

Define successful search before launch and measure the share of queries reaching that event. Keep the suggestion algorithm fixed and provide useful recovery from zero results in both groups.