The words behind a personal shopper for every visitor.
One page per term: what it means, why it matters for a store, an example, and what the agent does with it.
- Written for the people who run the store, not for engineers
- Every number carries its source and its period
- Each term links to the product page that does the job
Short definitions, one page each
Grouped by where the term lives: the agent, the site, retention, measurement and channels.
The agent
Autonomous sales agent
An autonomous sales agent is software that does what your best salesperson would, for every visitor at once. It learns your products, your business and each shopper, shows the right products, answers questions and follows up until they buy, without a person running it.
Read the definition →The agentAI personal shopper
An AI personal shopper is what a visitor experiences when a sales agent runs the store: search results that fit her, products picked for her on every page, a chat that answers from the live catalog, and messages that match what she looked at.
Read the definition →The agentNext-basket prediction
Next-basket prediction is a model that estimates what a shopper will buy next, and when, from her past orders and recent behavior. The agent uses it to pick the products on a page, the items in a cart reminder and the timing of a message.
Read the definition →On-site
Anonymous visitor
An anonymous visitor is someone on your site or in your app who has not logged in, signed up or given an email. Most visitors are anonymous for most of their visits, and most tools cannot act on them.
Read the definition →On-sitePersonalized site search
Personalized site search ranks results differently for each shopper. The same query returns the same products, but in an order that reflects her size, her brands and what she looked at, and it suggests products from the first keystroke.
Read the definition →On-siteZero-results search
A zero-results search is a query that returns nothing: a typo, a synonym the catalog does not use, a product name in another language or an item you do not carry. Each one is a shopper who told you what she wanted and got a blank page.
Read the definition →On-siteMerchandising rules (boost and bury)
Merchandising rules are instructions a team sets on top of the agent's ranking: boost a product line to the top, bury items that are low in stock, pin a campaign to a slot. The agent personalizes within the rules; the team keeps the last word.
Read the definition →Retention
Abandoned cart
An abandoned cart is a cart a shopper filled and left without paying. Abandoned-cart recovery is the message, usually email or push, that reminds her of the items, and it is the single most profitable automated message most stores send.
Read the definition →RetentionBrowse abandonment
Browse abandonment is when a shopper views products and leaves without adding anything to the cart. It happens many times more often than cart abandonment, and most of the shoppers who do it are anonymous, so most tools never reach them.
Read the definition →RetentionRFM segmentation
RFM segmentation groups customers by recency (when they last bought), frequency (how often) and monetary value (how much). It turns a customer list into states such as new, active, at risk and lost, so each state gets a different message.
Read the definition →RetentionDynamic coupon
A dynamic coupon is a discount code generated for one shopper, with its own value, products and expiry, used once. It replaces the fixed public code that leaks to coupon sites and ends up in the hands of people who would have paid full price.
Read the definition →RetentionLoyalty points
Loyalty points are credit a customer earns per order and spends as a voucher on a later one. A points program gives repeat customers a reason to come back to the store instead of to the ad that brought them the first time.
Read the definition →RetentionNPS (net promoter score)
NPS, the net promoter score, comes from one question: how likely are you to recommend us, from 0 to 10. Customers who answer 9 or 10 are promoters, 0 to 6 detractors; the score is the share of promoters minus the share of detractors.
Read the definition →Measurement
Predicted customer lifetime value
Predicted customer lifetime value is an estimate of what a customer will spend with you over a future period, made from her behavior and orders so far. It lets a store put its effort, and its ad budget, where the future value is, not where the last click was.
Read the definition →MeasurementHoldout group
A holdout group is a share of visitors or customers who do not get the agent, the message or the campaign, chosen at random. Comparing them with everyone else shows what the action added, in orders the store can count, instead of what a report claims.
Read the definition →MeasurementIncrementality
Incrementality is the part of a result that happened because of an action and would not have happened without it. An incremental order is one the reminder, the recommendation or the ad caused, as opposed to one it merely touched on the way to the checkout.
Read the definition →MeasurementAttribution window
An attribution window is the period after a click or a view during which a platform claims an order as its own. Meta's default is seven days after a click and one day after a view; a store's own analytics may use a different window, so the same order is counted in several places.
Read the definition →MeasurementServer-side tracking
Server-side tracking sends events such as page views, carts and orders from the store's server to the analytics and ad platforms, instead of from the shopper's browser. It records what the browser pixel misses when a blocker, a consent choice or a slow page stops the pixel from firing.
Read the definition →MeasurementIdentity resolution
Identity resolution is the work of recognizing that the visits, devices, emails and orders of one person belong to one person, and joining them into a single profile. Without it a store sees the same customer as three strangers.
Read the definition →The numbers, once a month
What the stores on Releva saw last month: Viber against email, email benchmarks, Meta ads, the season. Shares and ratios you can compare with your store, one email a month, no sales pitch.
Thanks. The next numbers reach you at the start of the month.
One email a month. Unsubscribe in any of them. Privacy policy
See which of last week's visitors you missed
Thirty minutes, with your store and ad accounts open. Then 30 days free. A holdout group decides: if the agent doesn't add orders in 30 days, you don't pay.
Book 30 minutes
You'll talk to one of us.