E-commerce glossary — no jargon.

A step that changes something in a supported platform, e.g. adds a tag, updates a field or adds a contact to a list. Available actions depend on the integration. An action doesn't have to mean sending a message.

• Automation actions and integrations →

Average order value: revenue divided by the number of orders for the same period. For example, PLN 10,000 from 100 orders gives an AOV of PLN 100. How returns and taxes are treated depends on the report's definition.

• How to analyze product sales →
• Does a free shipping threshold pay off? →

The way credit for a conversion is assigned to touchpoints, such as ad clicks or visits from messages. The rules depend on the model and the measurement window. Available models vary from tool to tool.

• Attribution models in Flowpeek →
• Why reports show different results →

In GA4, the share of sessions that didn't meet the engagement criteria. It is the complement of engagement rate, so the two add up to 100%. Other tools may define a bounce differently.

The cost of acquiring a new customer. You calculate it by dividing the acquisition costs you include by the number of new customers. Before comparing, agree on the period and which marketing and sales costs are in scope.

• ROAS vs. profit: how to judge a campaign →
• How much to spend on store advertising →

The path from the cart to the purchase confirmation. How the number of steps affects results depends on the store and needs to be tested. Measure it step by step in the funnel.

• The conversion funnel step by step →

A group of people who share a trait or event in a given period, e.g. customers who made their first purchase in September. Cohort analysis lets you compare their repeat purchases, retention and value over time.

A question asked of the data during a scenario, e.g. whether the visitor made a purchase in the last 24 hours. A YES or NO answer sends the flow down the matching branch.

• The abandoned cart scenario →

A mechanism that passes consent status to Google tags. The technical mechanism on its own doesn't replace deciding how consent should be properly collected and handled.

A label on a contact in a marketing platform, e.g. high_intent_before_cart. It supports segmentation and can start a further scenario in that tool. In Flowpeek, adding or removing a tag requires a configured action, a linked contact and current permissions. A tag doesn't expire on its own just because the prediction score has changed.

• Tagging by intent →

Conversion rate: the share of users or sessions that end with a specific action. For example: sessions with a purchase divided by all sessions. Define the event and the denominator before comparing reports.

• How to calculate your store's conversion rate →
• How to read a conversion funnel →

A model that uses the available data to estimate each touchpoint's contribution to a conversion. In GA4 it draws on paths that ended with a key event as well as paths that didn't. The credit assigned depends on the model and the data.

Pausing a scenario for a chosen amount of time. For example, after an add to cart you can wait 3 hours and then check the purchase condition before running the action.

• Waits and conditions in a scenario →

In GA4, the share of engaged sessions. Google's documentation counts sessions that last longer than 10 seconds, include a key event, or have at least two page or screen views. Meeting one condition is enough. Engagement rate and bounce rate add up to 100%.

Data a company collects through its direct relationship with its users or customers, e.g. from its own store. Using your own domain doesn't by itself guarantee complete measurement, and it doesn't change any choices about consent.

• Preparing and signing off a Flowpeek setup →

The sequence of steps leading to a chosen goal, e.g. visit, product, cart, order form and purchase. The numbers at each step and the transitions between them help you find where to dig deeper. A drop-off on its own doesn't explain the cause.

• Diagnosing the mobile funnel →
• Conversion funnel report →

A map showing how recorded interactions, such as clicks and scrolling, are distributed across a page. In Flowpeek, heatmaps and session recordings with AI analysis are in development (beta). A heatmap doesn't track eye movement; it helps you choose which part of a page to examine more closely.

• How to read heatmaps →
• Heatmaps in Flowpeek — coming soon (beta) →

A customer's value over time, expressed e.g. as the revenue or margin from your relationship with them. It can describe observed history or a forecast. When comparing LTV with CAC, check the period used and how the value was calculated.

Attribution that splits credit for a conversion across more than one recorded touchpoint. The split follows the model's rules and the available history. It isn't a standalone measure of a campaign's incremental effect.

• Flowpeek's six attribution models →

A machine learning model's assessment of purchase intent signals. Flowpredict shows a relative score of 0–100 used to compare levels of interest. A score of 71 means points, not a 71% chance of purchase. On the next recalculation the score may go up, go down or stay the same. You can combine a chosen threshold with behavior conditions in an automation.

• How to read the Flowpredict score →
• Flowpredict in practice →

Marketing aimed at people who have already interacted with the store. The group can be defined from the behavior available, e.g. the products viewed, the cart or the intent score. Flowpeek can pass a label for a recognized contact to a supported platform. Further communication and any sync with an ad system depend on the configuration and permissions in those tools.

• Remarketing by intent →

The percentage of customers who come back to make another purchase. The result depends on the period, the customer group and how a return is defined.

Attributed revenue divided by ad spend. Revenue of PLN 12,000 on spend of PLN 3,000 gives a ROAS of 4. The metric doesn't account for all of the store's costs and isn't a measure of profit.

• Why ROAS doesn't mean profit →
• Marketing cost analysis →

Connected steps: a trigger, conditions, an optional wait and actions. In Flowpeek you prepare a draft, review it and publish it. You can then go through the run history in the dashboard.

• Abandoned cart automation →

A points-based score stored on a contact in a marketing platform. An automation can increase or decrease it. It's a separate value from the relative Flowpredict score of 0–100.

• Prediction vs. contact points →

A set of people or visits selected by specific traits, e.g. intent score, traffic source or behavior. A high-score segment in the ML report is for analysis; you set the automation threshold separately. A group in a report doesn't automatically mean a contact list you can send messages to.

• From intent to a contact group →

Sending events to platforms from your server instead of the browser. It doesn't recover events that were never recorded, and it isn't a way to bypass consent.

• How to prepare and sign off a setup →

A single visit covering a series of actions in the store, grouped according to the analytics tool's rules. The inactivity timeout and the rules for starting a session can differ between systems.

• Dashboard reading basics →

The signal that starts a scenario. It can be, for example, adding a product to the cart, reaching an engagement threshold or no visit for a chosen period of time.

• How to build automations →

Link parameters, such as utm_source, utm_medium and utm_campaign, that tell your analytics the source, medium and campaign. Consistent tagging helps you compare acquisition channels.

• UTM links for your store, step by step →
• Your store's traffic sources →

Core Web Vitals describe three aspects of how a page performs: LCP — rendering of the largest content element, INP — responsiveness to interaction, CLS — layout stability. They're used to assess user experience.