Skip to content
upliftt

Your site is leaking sales.

Insert the code now and start generating conversions.

FIRE MY POWER

Most visitors leave without buying. Upliftt finds where they give up, changes one thing there, and tests it against a control. You keep only what wins.

One snippet to install. Nothing goes live without your approval.

Plug a leak

An example store with 1,000 visitors. Switch a fix on and watch where the visitors go.

27.2purchases per 1,000 visitors

No fixes switched on yet.

Checkout: 45% leave hereUpliftt never changes checkout, so this one stays yours.

Example numbers for an invented store. Real lifts come from tests on your own store, and some tests find nothing.

From one script tag to a proven fix.

  1. Step 1

    Paste

    Add one script tag to your theme, or add it through Google Tag Manager. The dashboard tells you the moment it is live.

    Any platformTag Manager
  2. Step 2

    Find

    Upliftt opens your key pages on desktop and mobile, reads two weeks of shopper behaviour, and ranks fixes by their expected impact.

    DesktopMobileRanked
  3. Step 3

    Preview

    Open any fix on your own store before a single shopper sees it. It runs in your browser only, and nothing is tracked while you look.

    Only you see it
  4. Step 4

    Prove

    Each approved change runs as an A/B test against the original, on the same traffic. You roll out a winner, or keep what you had.

    Live controlYour call

An audit that shows its evidence.

Most website changes barely move sales, and the cosmetic ones average zero. So the audit looks only for problems research has measured, finds them on your pages and in your visitors' numbers, and names the study behind each one.

  • Store wideNo buy now, pay later option at checkoutOffer Afterpay or Zip, then measure what it does to sales.Controlled experiment
  • Product pageNothing says when the order will arriveAdd your own dispatch times below the add to cart button.Usability research
  • Product pageMost products have no reviewsAsk every buyer for a review: the first five per product matter most.Observational study

Findings are ordered by strength of evidence, controlled experiments first. Nothing cosmetic is suggested: in thousands of published tests, colour, button and wording changes averaged zero.

Four kinds of change, nothing else.

Small on purpose. Each one is easy to preview, easy to test and easy to undo, and none of them runs code on your store.

  • Replace text

    Rewrite what shoppers read at the moments that matter.

    HeadlinesButton labelsDescriptions
  • Restyle

    Make the thing you want clicked easier to see.

    ColourSizeSpacing
  • Hide

    Take away what gets in the way of buying.

    BannersPopupsClutter
  • Insert

    Add the reassurance shoppers go looking for.

    Trust badgesShipping notesUrgency notes

Kept only if it wins.

Half your visitors see the change and half see the original, on the same days. The bar is high on purpose: results get checked every day, and a looser rule calls winners that are not there.

  • A winner needs 99% probability, 14 days and 500 visitors per version.
  • Before you launch, Upliftt estimates how many weeks the test needs at that page's traffic.
  • A change that is clearly hurting sales is paused on its own.
Example test result
Add to cart button labelRunning, day 9 of 14
Original2.61% convert
With the change3.02% convert

96.2% probability the change beats the original

Not a winner yetA winner needs 99%

Careful with your store.

A tool that changes a live store has to be boring about safety. These are rules in the code, not promises.

Checkout is off limits

Checkout, payment, account and admin pages are never changed. On an order confirmation page, the snippet records the purchase and nothing else.

Four kinds of change

Replace text, restyle, hide or insert. A change never runs code on your store, and anything inserted is cleaned first.

One kill switch

A single switch turns the snippet off for the whole site: no tracking, no tests, no changes. It takes effect within a minute.

Guardrails that act

A test whose variant is clearly worse, or whose change throws script errors, is paused automatically and the reason is recorded.

Behaviour only

Page views, clicks, scroll depth, cart and purchase events. No session recordings, form fields are never read, and search terms are removed from recorded addresses.

Consent aware

Where your site asks for consent, tracking and tests wait for it. Visitors who decline are left out.

What a better conversion rate is worth.

Put in what your store makes in a month and move the slider. The sum is yours to check.

$
+1%+30%

That is about$60,000 more a year.

A model, not a forecast: the same visitors and the same order value, with more of them buying. What a test finds on your store decides the real number.

The method is published.

Anyone can claim a lift. You should be able to check how a recommendation was ranked and how a winner was called.

The evidence

Every check comes from published research, controlled experiments first, and every finding names its source and what it lets you expect.

The measurement

Each check is measured on your pages and your own visitors' numbers. The AI only writes wording, from your own policy pages, and every line is matched back to them.

The decision rule

The probability that the change beats the original, worked out exactly. 99% to win, at least 14 days, 500 visitors per version, and a check that traffic really split as set.

Paste it once. You are set up.

One asynchronous script, under 15 KB compressed. It never blocks the page from rendering and sits beside the analytics you already run. If anything goes wrong, it does nothing rather than break the page.

<script async src="…/snippet.js"
  data-key="cro_…"
  data-config="…"></script>

Nothing to connect and no app to approve. Any store that accepts a script tag: Shopify, WooCommerce, BigCommerce, Magento, Squarespace and custom built stores, directly or through Google Tag Manager.

Questions people ask first.

How is this different from an A/B testing tool?

A testing tool gives you the machinery and leaves the ideas to you. Here the audit decides what to change and writes the change, and the test comes last, to check the change earned its place.

Can it change my store without me?

No. It recommends. You preview a change, you launch its test, and you decide whether the winner is rolled out.

What if a change makes things worse?

A test that is clearly hurting sales, or throwing script errors, is paused automatically. You can also stop any test from its page, and one kill switch turns everything off for the whole site.

How much traffic do I need?

Enough for a test to reach a result. Before you launch, the recommendation tells you roughly how many weeks a test would need at that page's traffic. Quiet pages can take months, and Upliftt says so rather than pretend.

Does the snippet slow my pages down?

It loads asynchronously, is under 15 KB compressed and never blocks rendering. While a tested element is being changed it is hidden for a moment so shoppers do not see it flicker, and never for more than a second.

What does it record about shoppers?

What they do on your store: page views, clicks, scroll depth, add to cart, checkout and purchase events. No session recordings, what shoppers type into form fields is never read, and search terms are taken out of the page addresses that are recorded. Where your site requires consent, nothing is recorded without it.

Find the first leak on your own store.

Add your site, install the snippet, and run your first audit on your own pages and your own numbers.

Get started
Upliftt · AI conversion optimisation for ecommerce