# Segments & Breakdowns

Slice your results down to a group of visitors — like mobile users or email traffic — to spot patterns. Powerful, with one important rule to use them honestly.

Look at your results for just one group of visitors — mobile users, email traffic, returning customers — to spot patterns the overall number can hide. Powerful, with one golden rule to keep it honest.

Your headline result mixes everyone together. Sometimes the interesting story is in a slice: maybe a new layout wins big on mobile but loses on desktop. **Segments** let you narrow the whole results page down to one group of visitors so you can see those patterns. The numbers, the chart, and any export all follow your filter while it's on.

## What you can break results down by

| Dimension | Examples |
| --- | --- |
| Traffic Source | Direct, Search, Social, Referral, Email, Paid |
| Device Type | Mobile, Desktop, Tablet |
| Visitor Type | New vs Returning |
| Identity | Identified (logged-in) vs Anonymous |
| Browser | Chrome, Safari, Firefox, Edge, Opera |
| Operating System | Windows, macOS, Linux, iOS, Android |
| Day of Week | Weekday vs Weekend |
| Country | Wherever your visitors are |
| UTM tags | Source, Medium, Campaign on your links |
| Custom dimension | Any custom property you send with the SDK |

You'll only see the breakdowns Otter actually has data for — the list grows on its own as different kinds of visitor arrive. And you can combine several at once, like **Mobile** + **Email**.

## How to use it

1. **Open the Filter control.** On the results page, open the **Filter** control and choose the values you want — for example, Device Type: Mobile. The page updates instantly.
2. **Narrow further, or clear.** Add more filters to slice tighter, or use the clear (✕) button to remove one and go back to everyone.

> **The golden rule**
>
> **Don't pick your winner from a segment.** When you slice results into a small group and hunt for a standout, you'll often find one purely by chance — the smaller the group, the more likely the "win" is a fluke. Make your ship decision on the overall result, and treat anything exciting you spot in a segment as an idea to confirm with a fresh, dedicated test. The Experiment health card shows a reminder whenever a segment filter is on.

> **Segment tips**
>
> **Use segments to ask better questions, not to declare answers.** "Mobile loved this" is a great hypothesis for a mobile-only follow-up test.
>
> **Clear filters before you read the real result.** It's easy to forget a filter is on and misread a slice as the whole. Check the [Experiment health](https://www.otterab.com/docs/analyzing-results/experiment-health) card — it tells you when you're looking at a slice.
>
> **Exports match your filter.** A CSV or PDF taken while a segment is on contains only that group — handy when you want to share a specific cut, easy to misread if you forget.

## Frequently asked questions

### What is a segment breakdown?

It's a way to look at your results for just one group of visitors instead of everyone. For example, you can ask "how did variant B do for mobile visitors only?" or "just for people who came from an email?" The whole results page — the numbers, the chart, and any export — narrows down to that group while the filter is on.

### What can I break results down by?

Quite a lot: Traffic Source (direct, search, social, referral, email, paid), Device Type (mobile, desktop, tablet), Visitor Type (new vs returning), Identity (identified vs anonymous), Browser, Operating System, Day of Week (weekday vs weekend), Country, the UTM tags on your links (source, medium, campaign), and any custom property you send with the SDK. You can pick more than one at once — like Mobile and Email together.

### Why don't I see every option?

Otter only shows the breakdowns it actually has data for. If no one in your test arrived on a tablet, the tablet option won't appear; if you never set UTM tags, the UTM filters stay hidden. The list grows automatically as that kind of visitor shows up.

### How do I apply a segment?

On the results page, open the Filter control and pick the values you want — say Device Type: Mobile. The page updates instantly. Add more filters to narrow further, and use the clear (✕) button to remove a filter and go back to everyone.

### Is it safe to pick a winner from a segment?

This is the most important rule: usually no. When you slice results into a small group and go looking for a standout, you'll often find one purely by chance — the smaller the group, the more likely the "win" is a fluke. So read the overall result first and make your ship decision on that. Treat anything exciting you spot in a segment as an idea to confirm with a fresh, dedicated test — not a conclusion. The Experiment health card shows a reminder whenever a segment filter is on.

### Does the export and chart follow my segment?

Yes. While a segment filter is on, the chart, the headline numbers, and any CSV or PDF you export all reflect just that group. Want the full picture back? Clear your filters before reading or exporting.

### Does breaking results down change my test?

No. Segments are an analysis lens only — they never change who sees which version, how conversions are counted, or your score. Slice away with total confidence; you can't break anything.

### What do 'Identified' and 'Anonymous' mean?

Identified visitors are ones your site has told us about by calling the SDK's identify() — for example, logged-in users. Anonymous visitors haven't given any identity signal yet. Splitting by this lets you compare, say, how a change lands for signed-in customers versus first-time strangers.

### Can I break results down by my own data?

Yes. If you send custom properties with the SDK (like a plan tier or account type), those show up as a custom dimension you can filter by. It's the same idea as the built-in breakdowns, but for whatever matters to your business.

---

Canonical page: https://www.otterab.com/docs/analyzing-results/segments