Appearance

Glossary ​

Key terms used throughout Dalton and this documentation.

A ​

A/B test ​

A method of comparing two versions (A and B) of a page to see which performs better. Dalton splits traffic between them, evenly by default, for a planned window, then compares the results.

Action required ​

A finished test waiting for you to deploy or conclude it. See the status table.

Anti-flicker ​

Hiding only the elements a test changes until the variant is applied, so visitors don't see the original flash first. See Performance & SEO.

Audience ​

Which visitors a test runs for: device, traffic source, new or returning, country, referrer or query parameter, combined with AND or OR. Set next to the targeting group. See Audiences.

Average order value ​

Revenue divided by orders. Revenue per visitor is conversion rate times average order value.

B ​

Baseline ​

Your original page without any Dalton modifications. Every experiment runs its variants against the baseline at the same time, so they are always measured over the same traffic. Also called the original.

Brand Intelligence ​

What Dalton knows about your brand: your design system, writing guidelines, competitors, customer reviews and the rules the agent has learned. Dalton fills most of it from a scan of your site, and the agents use it when building variants. See Brand Intelligence.

C ​

Chance to beat ​

How likely the variant is to be really better than the original, as a percentage. Dalton calls a winner at 90% by default; each account can choose 80, 85, 90 or 95%. It is not the same number as statistical significance. See Choosing your bar.

Chrome extension ​

The Dalton extension for desktop Chrome. Without it, some pages don't render well in the editor. See Installation.

Conclude ​

Close a finished test without deploying it. Its finding is filed in Learnings.

How strictly Dalton waits for your cookie banner: Standard (tests run for everyone, cookies only after consent) or Consent-first (nothing runs until the visitor accepts). See Consent.

Conversion ​

When a visitor completes a desired action (purchase, signup, form submission, click). This is what Dalton optimizes for.

Conversion goal ​

The action Dalton measures every test against, usually a purchase. Set once for the account under Settings → Conversions.

Conversion rate ​

The percentage of visitors who complete your conversion goal. Calculated as: (conversions / visitors) x 100.

Core Web Vitals ​

Google's page experience metrics: LCP (loading), INP (responsiveness) and CLS (layout shift). See Performance & SEO.

CRO Agent ​

Dalton's agent that scans your store, reports conversion problems and builds a test for each. See CRO Agent.

CSP ​

Content Security Policy: a header that limits which domains a page may load from. A strict one needs Dalton's two domains allowed. See CSP and firewalls.

D ​

Dalton Script and Dalton Pixel ​

The script goes in the <head> of every page and runs the tests. The pixel reports purchases where the script can't see them, such as on Shopify checkout. See Installation.

Data layer ​

window.dataLayer, where Dalton pushes a dalton_experiment event for every visitor in a test, so GA4, GTM or your own tools can split results by variant. See Analytics integration.

Deploy ​

Serve a winning variant to visitors after the test ends, to everyone or to a segment. See Deploy to a segment.

E ​

Early stop ​

Ending a test before its window. Dalton only flags a clear loser, after at least 7 days; winners are never called early. See How experiments run.

Editor agent ​

The AI chat in the editor that builds and changes variants, and sets audiences, from a plain description. See Editor agent.

Experiment ​

A test running on your page with one or more variants. Dalton automatically distributes traffic and measures results.

F ​

Field data ​

Performance measured from real visitors, such as the Chrome UX Report. Compare with lab data.

Flicker ​

A visitor briefly sees the original page before it changes into the variant.

L ​

Lab data ​

Performance measured in a simulated visit, such as Lighthouse.

Landing pages ​

A beta feature that gives each Meta ad its own version of a landing page, matched to the ad. See Landing pages for ads.

Learning ​

A conclusion Dalton draws across several concluded experiments, describing what does and does not work on your store. See Learnings.

Lift ​

How much better a variant did than the original, in percent, with a 90% range around it. See Reading your results.

Locale path ​

The part of a URL that marks a market or language, such as /fr-BE/ or /de/. Use it to target a market. See One test across markets.

M ​

MCP server ​

Connects Claude, ChatGPT or Cursor to your Dalton account, so you can ask about results or draft tests from your own AI tool. See MCP server.

Multi-page experiment ​

An experiment whose targeting group covers many pages, such as every product page. A change applies to all of them. See Testing across many pages.

N ​

Not enough data ​

A test that ran its full window without every variant reaching 350 visitors and 50 conversions. See Troubleshooting.

P ​

Personalisation ​

Showing different visitors different versions: a test for one audience, copy per market or campaign, or a winner deployed only where it won. See Personalise by audience.

Planned window ​

The number of days an experiment is scheduled to run, fixed before launch from the traffic its targeted pages get. Always whole weeks. See How experiments run.

Preview ​

Opens your live site with a variant forced on in your own browser, on desktop or on your phone. Previews don't count in results.

Promising ​

A variant ahead of the original, with a chance to beat between 80% and your bar. Dalton may offer an extra week.

Prompt ​

What you type to Dalton's agents to describe a change, a scan or an audience. See Editor agent.

R ​

Range ​

The 90% range around a lift: where the true effect most likely sits. Read the range, not the single number. See Reading your results.

Referrer ​

The site a visitor came from. Usable as an audience rule, matched on the root domain.

Render-blocking ​

Anything that delays the first paint of the page. The Dalton script is not render-blocking.

Revenue per visitor ​

Revenue divided by visitors: conversion rate times average order value. Catches a variant that sells more but cheaper. See The table.

S ​

Secondary metric ​

An extra event or page view shown in results without being optimised for, such as newsletter signups. See Conversions & goals.

Segment ​

A group of visitors by device, channel or country, used when deploying a winner only where it won. See Deploy to a segment.

Session ​

See Visit. Dalton shows visits as sessions or visitors.

Smart deploy ​

A deploy option that serves a winner only to the segments where it did not clearly lose, or a non-winner only where it clearly won. See Deploy to a segment.

Statistical significance ​

How unlikely a gap this big would be if there were no real difference. Tools like Convert and GA4 report it; Dalton reports chance to beat instead. Roughly, 95% chance to beat in Dalton matches 90% significance in those tools, and 97.5% matches 95%. See Choosing your bar.

T ​

Targeting group ​

A reusable set of URL rules describing which pages an experiment runs on. See Targeting.

U ​

URL split test ​

A test that sends part of the visitors from the original URL to a different page you built, and compares them. See URL split tests.

UTM parameters ​

Tags on a link, such as utm_campaign=spring-sale, that say where a visit came from. Usable as an audience rule or to change copy per campaign. See Personalise by audience.

V ​

Variant ​

A modified version of your page being tested against the baseline. Each experiment can have multiple variants.

Visit ​

One browser over 24 hours. Dalton's unit of measurement: a visitor keeps their variant for the visit, and results count visits, shown as visitors or sessions. See What is counted.