> For the complete documentation index, see [llms.txt](https://docs.elevateab.com/elevate-helpcenter/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.elevateab.com/elevate-helpcenter/getting-started/glossary.md). # Glossary Plain English definitions of every term used in Elevate A/B Testing, including the ones merchants most often confuse. *** Every term Elevate A/B Testing uses, in plain English. If a term in the app or the docs is unfamiliar, look it up here first. ### Terms | Term | What it means | | ------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **A/B test** | The general practice of comparing two or more versions to see which performs better. In Elevate, an individual A/B test is called an experiment. | | **Add to cart rate** | Visitors who added a product to their cart. Available as a goal metric. | | **Anti-Flicker** | A safeguard that prevents visitors briefly seeing the original version before a variation loads. | | **Assets** | Optional files attached to an experiment for reference: mockups, briefs, or notes. Images and documents up to 10 MB, videos up to 100 MB. | | **Audience targeting** | Rules that decide which visitors are eligible to enter an experiment. A visitor who does not match is never counted. | | **Average order value** | Net revenue divided by number of orders. Available as a goal metric. | | **Bayesian model** | The statistical method Elevate uses to evaluate results. It reports how likely each variation is to be the better one, rather than a pass or fail verdict. | | **Cart to checkout rate** | Carts that went on to start checkout. Available as a goal metric. | | **Checkout start rate** | Visitors who began checkout. Available as a goal metric. | | **Checkout success rate** | Visitors who completed their purchase. Available as a goal metric. | | **Control** | The unchanged version of your store, shown to one group of visitors. Labelled **A** in the app. Everything is measured against it. | | **Conversion rate** | Clicks on the tested element, divided by views. Available as a goal metric. | | **Custom audience** | A reusable targeting rule you build yourself from a dimension, a rule, and a value. | | **Custom metric** | A goal metric you define yourself, when none of the built in metrics measure what you care about. | | **Customer Authentication** | A targeting dimension for whether a visitor is signed in to a customer account. Not the same as Visitor Type. | | **Dimension** | The attribute a custom audience rule is built on, such as Country, Referring Domain, or Device Type. | | **Done** | The lifecycle state of an experiment that has been ended. Its results remain available. | | **Draft** | The lifecycle state of an experiment that has been created but not yet previewed or launched. | | **Duplicate** | A copy of a theme, page, or product that Elevate creates to serve a variation. Required for Theme, Page, and Product experiments only. | | **Experiment** | A controlled comparison between a control and one or more variations, run on live traffic to find which performs better. Elevate's standard term for a single test. | | **Goal metric** | The single metric Elevate uses to decide the winner. Every other metric is still calculated and shown in your report. | | **Headless** | A storefront built on a custom front end such as Hydrogen, Next.js, or Remix, rather than a Shopify theme. | | **Hypothesis** | What you expect to happen and why, recorded before you launch. An optional field on every experiment. | | **Liquid snippet** | A file added to a theme that delivers experiments on the storefront. Used on the manual install path. | | **Metafield** | Where Elevate stores experiment configuration on your Shopify store. Managed automatically. | | **Mutually exclusive groups** | Another name for traffic isolation, used by many testing tools. See Traffic isolation. | | **Paused** | The lifecycle state of an experiment that has been stopped temporarily. It is not collecting data and has not been ended. | | **Personalization** | A targeted experience delivered to a specific audience continuously, with no traffic split and no winner. Distinct from an experiment. | | **Preview** | The lifecycle state used to check variations before launching. Nothing is served to real visitors and no data is collected. | | **Product Costs** | Cost of goods data you add so Elevate can calculate profit alongside revenue. Required for the Profit per visitor metric. | | **Profit per visitor** | Gross profit after COGS, shipping, and fees, divided by unique visitors. Requires Product Costs. Available as a goal metric. | | **Redirect snippet** | The component that handles redirect based experiment types. Separate from the theme extension. | | **Revenue per visitor** | Net revenue divided by unique visitors, after discounts. Elevate's recommended goal metric for most experiments. | | **Running** | The lifecycle state of a live experiment that is serving variations and collecting data. | | **Split test** | Another name for an A/B test. See Experiment. | | **Statistical significance** | The point at which a result is unlikely to be down to chance. Elevate tells you when this has been reached. | | **Subscription percentage of orders** | Orders that included a subscription. Available as a goal metric. | | **Theme extension** | The Shopify app embed that lets Elevate change what a visitor sees on the page. Enabled once per theme. | | **Traffic allocation** | How the visitors entering an experiment are split between the control and each variation. Defaults to an even split. | | **Traffic isolation** | An optional setting that reserves a share of your total store traffic for one experiment, so those visitors enter no other experiment. Other testing tools call this mutually exclusive groups. Off by default. | | **Variant** | In Elevate, always use variation. Variant refers to a Shopify product variant, which is a different thing. See Variation. | | **Variation** | A changed version of your store, shown to one group of visitors and compared against the control. Labelled **B**, **C**, and so on. | | **Visitor** | A unique person who reached your store. Elevate counts visitors, not sessions, so someone who returns three times counts once. | | **Visitor Type** | A targeting dimension for whether someone has visited your store before. Not the same as Customer Authentication. | | **Web Pixel** | Shopify's tracking extension. Elevate uses it to record page views, add to carts, checkouts, and purchases. | | **Winner** | The variation Elevate identifies as most likely to perform better on your goal metric. | ### Terms that are easily confused #### Traffic allocation and traffic isolation These sit next to each other in the app and do different things. **Traffic allocation** splits the visitors who are already in the experiment between control and variations. An even split means half see A and half see B. **Traffic isolation** decides how much of your total store traffic is reserved for this experiment alone. Visitors in that reserved group enter no other experiment. Allocation is about who sees what. Isolation is about who else can reach them. If you have used other testing tools, traffic isolation is what they call **mutually exclusive groups**. It is the same idea: keeping experiments from overlapping so one cannot affect the results of another. Elevate implements it as a percentage of traffic you reserve, rather than as a named group you assign experiments to. #### Visitor Type and Customer Authentication **Visitor Type** is about familiarity. Has this person been to your store before? **Customer Authentication** is about identity. Is this person signed in to a customer account right now? A returning visitor is often logged out. A first time visitor can create an account and log in. Choosing the wrong one silently targets the wrong people. #### Experiment and Personalization An **experiment** splits traffic to answer a question, then ends. You do not know the answer in advance. A **personalization** delivers a chosen experience to a chosen audience, continuously. You already know what you want that audience to see. If you are trying to learn something, run an experiment. If you are trying to deliver something, build a personalization. #### Control and variation The **control** is your store as it is today. It is not a version you designed for the experiment, and it should not be changed while the experiment runs. A **variation** is what you are testing against it. An experiment can have several. #### Visitor and session Elevate counts **visitors**, not sessions. One person who comes back three times is one visitor, and stays in the same variation each time. This is why Elevate's numbers can differ from analytics tools that count sessions.