No-Code Experiments

Visual experiments let you build and launch A/B test variants on your website through a visual editor, with no code deploy per test. You open the page you want to test, click the element you want to change, edit it directly or describe the change to Mixpanel Agent, then launch.

No-code experiments are in closed beta. Access is enabled per project. Contact your account team to request to join.

Beta Scope

	Supported in beta
Browser for authoring	Chrome only
Install method	Web snippet in your page’s `<head>`
Site type	Best suited to static and marketing sites

Complex single-page apps are better served by a code experiment. Mutations are applied to the rendered DOM, so the more dynamic the page, the less predictable the result. The visual editor re-applies mutations when elements re-render, but a heavily client-routed app will still behave less reliably than a static page. Authoring is Chrome-only. The experiment itself runs for all of your end users in any browser, because the variant is delivered by the Mixpanel JavaScript SDK rather than by the extension.

Visual vs Code Experiments

Mixpanel has two experiment types. You choose one at creation time.

	Visual Experiment	Code Experiment
Variants built in	The visual editor, on your live site	Your codebase
Code change per test	None	Yes
Who can run it	Anyone with the right permissions	Requires an engineer
Best for	Copy, styling, and layout changes	Backend logic, new components, anything data-driven

Visual experiments run on Mixpanel Feature Flags, and the visual editor creates and configures the flag for you. You do not write any flag evaluation code, and you do not need an engineer to implement the flag. The one-time snippet install is the only code change the feature requires.

The experiment type can’t be changed after the experiment is created. If you need to convert a test, create a new experiment.

Setup

Setup is a one-time job. After it’s done, anyone with permission can create and launch experiments without touching code again.

Add the snippet to your site

Place this in the <head> of every page you want to test. It loads the beta build of the Mixpanel JavaScript SDK and turns on flags and visual experiments.

<script>
var MIXPANEL_CUSTOM_LIB_URL = 'https://cdn.mxpnl.com/libs/mixpanel-2.83.0-ve-alpha-1.min.js';
(function (f, b) { if (!b.__SV) { var e, g, i, h; window.mixpanel = b; b._i = []; b.init = function (e, f, c) { function g(a, d) { var b = d.split("."); 2 == b.length && ((a = a[b[0]]), (d = b[1])); a[d] = function () { a.push([d].concat(Array.prototype.slice.call(arguments, 0))); }; } var a = b; "undefined" !== typeof c ? (a = b[c] = []) : (c = "mixpanel"); a.people = a.people || []; a.toString = function (a) { var d = "mixpanel"; "mixpanel" !== c && (d += "." + c); a || (d += " (stub)"); return d; }; a.people.toString = function () { return a.toString(1) + ".people (stub)"; }; i = "disable time_event track track_pageview track_links track_forms track_with_groups add_group set_group remove_group register register_once alias unregister identify name_tag set_config reset opt_in_tracking opt_out_tracking has_opted_in_tracking has_opted_out_tracking clear_opt_in_out_tracking start_batch_senders people.set people.set_once people.unset people.increment people.append people.union people.track_charge people.clear_charges people.delete_user people.remove".split(" "); for (h = 0; h < i.length; h++) g(a, i[h]); var j = "set set_once union unset remove delete".split(" "); a.get_group = function () { function b(c) { d[c] = function () { call2_args = arguments; call2 = [c].concat(Array.prototype.slice.call(call2_args, 0)); a.push([e, call2]); }; } for (var d = {}, e = ["get_group"].concat(Array.prototype.slice.call(arguments, 0)), c = 0; c < j.length; c++) b(j[c]); return d; }; b._i.push([e, f, c]); }; b.__SV = 1.2; e = f.createElement("script"); e.type = "text/javascript"; e.async = !0; e.src = "undefined" !== typeof MIXPANEL_CUSTOM_LIB_URL ? MIXPANEL_CUSTOM_LIB_URL : "file:" === f.location.protocol && "//cdn.mxpnl.com/libs/mixpanel-2-latest.min.js".match(/^\/\//) ? "https://cdn.mxpnl.com/libs/mixpanel-2-latest.min.js" : "//cdn.mxpnl.com/libs/mixpanel-2-latest.min.js"; g = f.getElementsByTagName("script")[0]; g.parentNode.insertBefore(e, g); } })(document, window.mixpanel || []);

mixpanel.init("YOUR_PROJECT_TOKEN", { flags: true, visual_experiments: true });
</script>

Both flags and visual_experiments are required. Feature Flags is the mechanism that delivers variants to your users, so visual experiments will not run without it.

During beta, the snippet is the only supported install path, and it must point at the pinned build above rather than mixpanel-2-latest. Installing through npm with the visual_experiments init option is coming at general availability.

If you already load mixpanel-js on the page, replace that snippet rather than adding a second one, and keep your existing init options alongside the two new ones.

Install the Chrome extension

Install the latest Mixpanel extension from the Chrome Web Store.

Uninstall or disable any older Mixpanel browser extension first in order for the visual editor to work properly.

After installing, refresh your Mixpanel tab. Mixpanel checks for the extension when the page loads, so without a refresh it will keep reporting the extension as missing.

Create the experiment

From the Experiments report, create a new experiment and choose Visual Experiment. Enter the URL of the page you want to test in Visual Editor URL, then configure variants, success metrics, and duration the same way you would for any experiment. See Setup & Launch Your Experiment for those settings.

Design your variants

Click Edit in Editor to open your site with the visual editor overlaid on it. Design each variant, save, and launch.

Designing Variants

The editor opens your live site in a new tab with a toolbar for switching between variants and adding new ones. Select an element to edit it, or open Mixpanel Agent and describe the change you want.

What You Can Change

Change	Notes
Text	Replaces the element’s text. Markup is not parsed.
Colors	Fill and gradient, through a color picker
Other CSS	Any CSS property, set as an inline style
CSS classes	Add or remove classes
Attributes	Including `src`, which is how you swap an image
Hide	Removes the element from the page for that variant

Editing raw HTML is not currently supported. Moving, reordering, inserting, and duplicating elements are not available yet.

Editing With Mixpanel Agent

Mixpanel Agent can apply several changes at once from a single description, such as restyling a set of buttons and swapping the copy in different sections. Agent proposes the changes, you approve them, and they land in the changes list like any other edit. Agent can only author visual changes from inside the editor; the dashboard chat hands off to the editor instead.

Reviewing and Removing Changes

Every edit is recorded in a changes list in the editor. You can review each change and delete it individually, which is also how you undo a change: e.g., to unhide an element, delete the hide change.

Targeting a URL

A variant applies to a page when the page’s origin and path match the URL saved for that variant.

Query strings and hash fragments are currently ignored, so /cart?utm_source=email matches /cart
A trailing slash is ignored, so /cart/ matches /cart
Matching is case-sensitive, so /Cart does not match /cart
Wildcards, prefixes, and regular expressions are not yet supported

To test several pages, you currently need to create a separate experiment for each one.

Overlapping Experiments

When you enter a URL that another draft or running visual experiment already targets, Mixpanel warns you in case of potential conflicts. Overlapping visual experiments can interfere with each other and make results hard to attribute. The warning does not block you from saving or launching, so it’s on you to decide whether the overlap is acceptable.

Permissions

Launching a visual experiment requires the Launch Visual Experiments permission, which is separate from the ability to create or edit one. This lets you open variant design up to a wider group while keeping control over what reaches your users.

Role	Can launch visual experiments
Owner	Yes
Admin	Yes
Analyst	Yes
Consumer	No
Billing Admin	No

Grant it to a custom role under Organization Settings, in the Feature Flags and Experimentation group of project permissions. Without it, a user can design variants and save them, but the launch action stays disabled.

Limits and Best Practices

Variants you create from inside the editor start at a 0% traffic split and are served to nobody until you set their split on the experiment page in Mixpanel. Check your variant splits before you launch.

Editing variants on a running experiment changes what live users see immediately and can skew your results
An experiment can’t launch until at least one variant has an edit
Don’t edit the flag variant values by hand for a visual experiment; doing so breaks the editor
Edits to elements inside dynamic feeds can be inherited by items that render later
Visual changes are capped at 32KB per variant
One editor session per user at a time. Opening the editor again ends the previous session
Editor sessions expire after an hour
Concurrent edits to the same variant overwrite each other, so coordinate before multiple people edit one experiment

Troubleshooting

The Editor Reports the Extension Is Missing

Refresh the Mixpanel tab. Detection runs on page load. If it still fails, confirm you’re on the latest extension and that older Mixpanel extensions are disabled.

Changes Don’t Appear on Your Site

Confirm the snippet loads the pinned beta build and passes both flags: true and visual_experiments: true. The editor reports an error when it detects that visual experiments are turned off in the SDK. Also check that the URL you’re viewing matches the saved URL, keeping case sensitivity in mind.

Launching and Reading Results

Launching and analysis work the same as any other experiment. Results appear in the Experiments report. See Monitor Your Experiment and Interpret Your Results.

When a Visitor Counts as Exposed

Assignment and exposure are separate, and only exposure puts a visitor in your results. The SDK fetches flags on any page that loads the snippet. That fetch returns a variant only for experiments whose targeting rules and traffic allocation the visitor qualifies for, and qualifying is what assigns them to a variant. A visitor outside the targeting or the rollout receives no variant at all. Exposure is tracked later, when the SDK reads an assigned variant in order to apply its changes. That only happens on a page whose URL matches the one saved for the experiment, and the SDK sends an $experiment_started event at that point, once per experiment per page load. So a visitor can browse your site without ever entering your results, either because they didn’t qualify for the experiment or because they qualified but never reached a matching page. Neither case appears in your experiment results or counts toward your Monthly Experiment Users. Exposure events from a visual experiment have the same shape as those from a code experiment, so the Experiments report reads them identically.

INTRO

DATA IN

ANALYSIS

MIXPANEL AI

ADMIN

DATA OUT

SUPPORT

No-Code Experiments

Beta Scope

Visual vs Code Experiments

Setup

Designing Variants

What You Can Change

Editing With Mixpanel Agent

Reviewing and Removing Changes

Targeting a URL

Overlapping Experiments

Permissions

Limits and Best Practices

Troubleshooting

The Editor Reports the Extension Is Missing

Changes Don’t Appear on Your Site

Launching and Reading Results

When a Visitor Counts as Exposed

​Beta Scope

​Visual vs Code Experiments

​Setup

​Designing Variants

​What You Can Change

​Editing With Mixpanel Agent

​Reviewing and Removing Changes

​Targeting a URL

​Overlapping Experiments

​Permissions

​Limits and Best Practices

​Troubleshooting

​The Editor Reports the Extension Is Missing

​Changes Don’t Appear on Your Site

​Launching and Reading Results

​When a Visitor Counts as Exposed

Beta Scope

Visual vs Code Experiments

Setup

Designing Variants

What You Can Change

Editing With Mixpanel Agent

Reviewing and Removing Changes

Targeting a URL

Overlapping Experiments

Permissions

Limits and Best Practices

Troubleshooting

The Editor Reports the Extension Is Missing

Changes Don’t Appear on Your Site

Launching and Reading Results

When a Visitor Counts as Exposed