---
title: "Audience targeting"
description: "Restrict a test to specific URLs, countries, devices, or visitor types. Visitors outside the rules see the unchanged page and never count toward results."
lastUpdated: 2026-05-15
canonical: https://culsin.com/docs/simple-split-testing/audience-targeting/
source: https://culsin.com/docs/simple-split-testing/audience-targeting/
---
By default, a test runs against every visitor on every page that contains its markup. Audience rules let you narrow that.

A visitor only enters a test if they pass **all** of its audience rules. Visitors who fail any rule see the unchanged page (`data-split-variant` elements stay in place, classes and styles are not applied) and are not counted in the test's results.

---

## Supported rules

| Rule | Type | Values |
|---|---|---|
| **URL include** | List of strings | Each is a case-insensitive substring of `pathname + search` |
| **URL exclude** | List of strings | Same matching, but any match rejects the visitor |
| **Country** | List of ISO-2 codes | Country is determined server-side from the visitor's IP — not from the browser, so it can't be spoofed by client code |
| **Device** | `desktop` / `mobile` / `tablet` | Classified server-side from the User-Agent |
| **Visitor type** | `new` / `returning` / `logged_in` | New = no prior `_shopify_y` or `splt_usr_id` cookie. Returning = either cookie present. Logged in = Shopify customer session active |

All rules combine with AND. Each list within a rule (URL includes, URL excludes, countries) combines with OR.

---

## URL matching is substring, not glob

This is the part that most often surprises people: the matcher is a plain case-insensitive substring check, not a wildcard or regex match.

| You want to match | Use |
|---|---|
| Every product page | `/products/` |
| Every collection page | `/collections/` |
| One specific collection | `/collections/winter-sale` |
| Visitors arriving from a UTM source | `utm_source=newsletter` |
| Just the cart page | `/cart` (also matches `/cart/anything` — accept that, or use exclude rules to narrow) |

**Do not use `*`.** A pattern like `/products/*` will only match a URL that literally contains the string `/products/*`, which essentially never happens. Drop the asterisk.

If you need anchored or exact matching, the easiest way is to write the URL with its trailing context (`?` or end-of-path) — but remember the haystack is `pathname + search`, lowercased. There is no way to anchor to start-of-path; if `/cart` matches both `/cart` and `/cart-recovery`, exclude the unwanted one with an URL exclude rule.

---

## Country targeting

Country is determined server-side from the visitor's IP. It works without an IP geolocation library on your side and cannot be spoofed by client code.

Country codes are ISO 3166-1 alpha-2 (`DE`, `US`, `SE`, `GB`). Case is normalised before comparison.

If the country cannot be resolved (rare — mainly bots and some VPNs), the visitor is rejected whenever country targeting is set.

---

## Audience rules run on every request

A returning visitor is **re-evaluated against the audience rules on every page view**, not just on first entry. If they hit a page that doesn't match the URL rules, they get no variant assignment for that page view — even if they previously received one on a different page.

This means:

- If your test's markup appears on multiple URL patterns but only one is in the include list, the variant only applies on the included URL.
- A visitor who entered the test in country `US` and then routes through a VPN to `DE` will be rejected if the test is country-locked to `US`.

If you only want the variant to render on the included URLs, only put the test's markup on those pages. The sticky-assignment cookie speeds up subsequent page loads, but the audience rules are still checked on every request.

---

## What "rejected by audience" means for results

A visitor who fails any rule is **not** logged as a participant for that test. They are not in any variant's denominator, they do not affect conversion rates, and they do not contribute to sample size calculations.

---

## Common patterns

**Run a test only on the homepage.**
URL include: leave empty if `/` is the only path; otherwise add `/` (which matches everything).
URL exclude: `/products/`, `/collections/`, `/pages/`, `/cart`, `/account`, `/blogs/`.

**Test only US visitors on product pages.**
URL include: `/products/`
Countries: `US`

**Test only new mobile visitors on a campaign landing page.**
URL include: `/pages/spring-sale`
Device: `mobile`
Visitor type: `new`

**Avoid testing during checkout flow pages.**
URL exclude: `/cart`, `/checkout`

---

## Preview bypasses audience rules

Audience rules are bypassed for [preview links](/docs/simple-split-testing/preview-links) so you can verify a variant from any URL, country, device, or visitor type.