Work like a growth team before you hire one.

Watch every visit, error and slow page on Free, up to 100,000 visitors a month. Build pages and test them on Starter, with your first 60 days free and no card to start.

Start freeBook a demo

The Mtrix dashboard for Halden (haldentea.com) on its Executive overview board for the last 30 days: product revenue $31,860, up 9.4% (+$2,740) on the previous 30 days; conversion rate 2.19%, 684 orders, $46.58 average order value and 31,240 unique sessions. The Getting started guide is open over it with eight steps in two groups, Start here and Grow. Four tick themselves one after another — Measure your live site, Rebuild the page your ads point to, Make a variant of that page, Run your first experiment — until it reads 4 of 8, with Publish it open next.

Your first thousand shoppers
will show you what to fix.

Put one script on your store and everything starts on the next visit: sessions, errors, page speed and revenue. Long before a test could tell you anything, you can watch the people who got stuck and fix what stopped them.

Three screens from the Mtrix app on Halden, a small tea store. First, Settings › Install: the Script tag tab shows the one-script snippet loading https://cdn.mtrix.io/sdk/v1/mtrix.js, and the line "Waiting for your first event…" turns green as "First event received", with a button to watch the first session. Second, the Sessions list for the last 7 days (7,380 sessions, 6,120 unique visitors, average duration 1:48, 2.6 pages per session, 41 rage clicks), filtered by the Rage clicks flag down to 34 sessions; an iPhone Safari session from Instagram in the United Kingdom, 2 pages and 4:12 long with the Errors and Rage clicks flags and no revenue, is opened. Third, that session's replay: the shopper scrolls the Brew Kit page past the photo to How to brew and back, goes to the cart, types FIRSTCUP and taps Apply three times; the seek bar marks the rage click, and the Console tab shows the error at 2:40, "null is not an object (evaluating 'input.value')", with its stack trace pointing at cart-discount.js line 42 and a Seek to 2:40 button. The session's Errors (1) group lists the same error.

Make the better page yourself,
then put it up against the old one.

Select part of the page and describe the change. Mtrix AI draws a few designs in your store’s style, and the one you like can go into a test as a new variant, next to the page as it is. No developer, and no page-builder app on top of your theme.

The Mtrix builder open on Halden's Brew Kit page. The hero section is selected on the canvas. The AI panel opens and the prompt "Put the three brewing steps above the photo. People on phones scroll past the photo looking for them." is sent; the panel shows "Applying changes…" and then a "3 Design Variants" card with a "Preview in Canvas" button. On the canvas, the AI Preview banner switches between Original, Design 1, Design 2 and Design 3, and Design 2 is approved. The experiments popover then shows "Brew guide first" with two arms, Control and Variant 1, with Variant 1 on the canvas.

Test the changes shoppers notice.
Your traffic can answer those.

With a few hundred shoppers a day, a new button colour can run for months without an answer. A new page layout that changes how many people buy can get one inside your first 60 days. The report tells you every day which of those you're looking at.

A 60-day timeline of Halden's two A/B tests, each drawn as a row of days coloured by the report's state. Brew guide first, on /pages/brew-kit: still gathering data for the first 6 days, readable from day 7, and on day 32 the report reads "Variant 1 wins, roll out Variant 1": Variant 1 converts at 3.3% against 2.4%, a 100% chance of being the best, and the peeking-proof verdict says it is beating control. Add to cart colour, on /products/low-tide: readable from day 7 and on day 60 still "Readable, but not decided yet", with Variant 1 at 3.3% against 3.2%, a 70% chance of being the best, and a verdict of not enough to call yet. Under the timeline, the report's Statistics view for the selected test on the selected day.

Day 32
  • Test something big enough to see. A different layout, a different first photo, a kit instead of a single jar. Small tweaks need far more shoppers before anything can be called.
  • Check it every day. Each variant shows its chance of being the best and the risk if you pick it, next to the control it's measured against.
  • A winner has to earn it. Every variant needs at least 100 visitors, 25 conversions and 7 days, and a clear lead in conversion rate, before anything is called.

Free while you watch.
$700 a month once you test.

Free is never invoiced and never runs out. Starter adds experiments, the builder and unlimited replays, and your first 60 days of it are free, with no card to start. Neither plan counts seats.

Free

$0

Watch the store you have.

Monthly visitors
Up to 100,000
Events / month
1,000,000
Session replay
1,000 a month
Seats
Unlimited
Experiments + Builder
Locked
Support
Docs and email

On Halden's store, this replaces

  • Google Analytics Replaced by Mtrix
  • Hotjar Replaced by Mtrix

Starter

$700 /mo

Test it and build it.

Monthly visitors
Up to 100,000
Events / month
5,000,000
Session replay
Unlimited
Seats
Unlimited
Experiments + Builder
Included
Support
Email + in-app

On Halden's store, this replaces

  • Shoplift Replaced by Mtrix
  • Replo Replaced by Mtrix
  • Klaviyo Native integration
  • Judge.me Keeps working
  • Seats are unlimited on every plan. Your co-founder, a freelance designer and the agency running your ads each get their own login and role.
  • A busy week never switches anything off. If a launch sends more visitors than your plan covers, collection carries on. On Free you get an email, not a bill.
  • Growing past 100,000 visitors? Growth and Scale cover more traffic and more stores on the same script, data and tests. See every plan →

Questions founders ask first

  • Is the Free plan actually free?

    Yes. Free is never invoiced and needs no card. It covers analytics, session replay, error tracking and page speed for one store, up to 100,000 visitors and 1,000 replays a month. Experiments and the builder are the two things it leaves out.

  • Do I need a developer to install it?

    No. The script goes into your theme in a couple of minutes, and the install screen tells you when the first event arrives. If someone else built your site, you can email them the snippet from the same screen.

  • Does it work with Shopify?

    Yes. Analytics, replay, errors, page speed and experiments all run on a Shopify store through the same script, and on any other platform. The builder is different: pages you make in it are hosted by Mtrix.

  • How long will a test take at my traffic?

    It depends on how many shoppers see the page and how big the change is. The report tells you where you stand every day: still gathering data, readable but not decided, or a winner. Bigger changes on busier pages get an answer soonest.

  • Can our freelancers and agency have logins?

    Yes. Seats are unlimited on every plan, Free included. Invite anyone by email and give them a role: Admin, Project Manager, Member, Editor or Tester.

  • What happens when we outgrow Starter?

    You move up from Settings › Billing and the change is immediate. Growth covers 300,000 visitors and three stores, and nothing is reinstalled: the same script, data and tests carry on.

Start with the store
you already have.

Free watches it from the first visit. Starter is there the day you want to test something.

Waiting for your first event… First event received
Start freeBook a demo
Build with Mtrix