A/B Test

A test comparing the performance of two or more alternatives of a single page element, page, or funnel, showing both alternatives to visitors in equal (“50/50”) proportions.


A test or target state. When a test or target is archived, all effects of a test or target are removed from the site, and the site reverts back to its pre-test state. Archived tests and targets cannot be modified or reused.

See Test and Target States


The (usually unintentional) use of a page or location by multiple tests. Optimize helps you avoid collisions and running the risk of invalidating the results of the competing tests. Within a project, you can run multiple tests on a page or location using segmentation to ensure visitors are safely routed to relevant variations without risking content collisions.

See Avoiding Collisions


The baseline (unchanged) comparison for all new or modified content variations within a test.

Conversion Event

A conversion event is what happens when a key performance indicator (KPI) is achieved as a result of one or more visitor actions–for example clicking a button, filling out a form, subscribing to a newsletter, or making a purchase.

Conversion Rate

The percentage of visitors who triggered a conversion event.


 An individual HTML component in a web page’s markup. An element must have an ID (id=”example”) for it to be used in a test.

Entry Page

The first page requested and shown in a visit. A visit has only one entry page.

Exit Page

The last page viewed in a visit.


Attrition that happens as site visitors go through a scenario, a series of defined steps such as a purchase or a registration on a web site. Because the number of people participating in each step is usually smaller than the step before, a graph of the declining participation resembles a funnel.

Landing Page

The first page a visitor sees when navigating to a site. Often, the goal of the landing page is to get visitors to trigger a conversion event.


The percent increase in conversion rate when comparing one test alternative to another. Lift is calculated by subtracting the control’s conversion rate from the alternative’s conversion rate, dividing the result by the control rate. Then multiplying that result by 100. For example:
Experiment conversion rate:  4.5%
Control conversion rate:   4.0%
Lift:  ((4.5 – 4) / 4 = .125) * 100 = 12.5%


A test or target state. A Live test or target is visible to all visitors in the audience segment (if applied). Optimize is actively changing webpage content depending on the design of the test or target.

See Test and Target States


The test alternative that resulted in the highest conversion rate in a given test.

Page View

A page that is displayed by a browser. Also, “page view” often indicates page files that are delivered to a browser, whether or not they are displayed on the screen–for example, a redirect page.


A way of organizing related tests, targets and conversion events. Tests and targets are always associated with a project.


A test (but not target) state. When a test is Published, the chosen variation appears on the website. A test is typically published after it has been live for a while and a variation “winner” (the best-performing or optimal variation) was determined.

See Test and Target States


A web domain, site, or page that contains a link to one of your site pages that was used by a visitor to get to your site.

Referring Domain

A web domain that contains a link to one of your site pages, used by a visitor to get to your site. For example,


A “slice” of site visitors that share one or more preferences and characteristics. You segment audiences to help better target web content, maximizing its relevance to those who see it.


All user activity on a site within a specific time frame. Sessions simply represent all hits by the same user, organized in one group. Synonymous with visit.

Split Test

A split test displays two or more different web pages to visitors in unequal ratios. Split test traffic is divided among each page in the split test using a percentage you define: for example, a split test with two pages could have view percentages of 50/50, 20/80, or any increment of 5% totaling 100%. Using Optimize, you can perform both A/B testing and split testing.


A test or target state where it is visible only to someone that has the preview link. You can view, edit, and share the appearance of the variations before the test is made Live.

See Test and Target States


Personalized content and experiences delivered to specific segments of visitors.


The display multiple variations of content to certain ratios of visitors–for example, showing one headline variation to half of visitors, and a different headline to the other half. You can choose what percentage of all visitors will participate in a test, and control what ratio of participants see each variation.


Setting the percent of total visitors that will participate in a test or target. You can throttle what percent of total visitors are included in this test, and also what percent of tested visitors will see each variation. Throttling allows any percentage (0.1% to 100.0%) of a page’s traffic to be exposed to testing or targeting. Restricting traffic too much could drastically limit results and make it more difficult to declare an optimal variation.

Unique Visitors

Number of unique individuals who visited your site during a defined time period, as identified by a persistent cookie. If someone visits more than once during the time period, they are counted only as one unique visitor. Unique visitors may not perfectly match the number of unique individuals visiting the site, because someone may visit a site from more than one computer and have a different cookie at each computer, or people may share the same computer to access the same web site.


Any visitor to a page on your website.


A design choice that you want to show visitors as part of a test–for example, a page headline or a web form. You


All the activity, of one visitor’s browser to a web site, within certain time constraints. A visit is a series of page views, beginning when a visitor’s browser requests the first page from the server, and ending when the visitor leaves the site or remains idle beyond the idle-time limit.