Skip to content
All articles
Retail and E-commerce

Customer identity and the 360 view · how retail recognizes a person

In retail the same person shows up as an email, a phone number, a store code and a device. This screen shows which signals the customer was recognized by, the merge history and the button to undo a merge, the trust score calculated from active risk signals, the household so you do not send the same campaign twice to one address, and the preferences on file.

Anyone selling at retail rarely has a single record per person. There is the email from the site, the phone the person gave in store, the customer code from the platform, the social login and the device that browsed for two weeks before buying. Without pulling those together, you see five short stories instead of one customer. This part of the vertical solves exactly that, and it makes a point of showing its own reasoning: you see which signals the person was recognized by, not just the result.

Where to find it

  1. Open Retail from the menu, on the Orders tab.
  2. Click the number of an order that has a customer linked to it.
  3. Below the list, the customer panel appears, with the sections How this customer was recognized, Merges for this customer, Customer trust, Household and Preferences.
  4. Clicking the same number again closes the order and the panel.

If the order has no customer linked, the panel does not appear. That happens when the platform sent the order without identifying the buyer, the classic case of a walk-in counter sale.

How this customer was recognized

The table lists every identity signal with four columns: Signal, Value, First seen and Last seen. The Signal says what kind of identifier it is, for example the store customer code, the email, the phone or the device. The dates tell the story: a signal first seen eight months ago and last seen yesterday is a long standing, active customer; a signal seen only once is usually a visit that never became a relationship.

The rule the system follows is written on the screen itself: a strong identifier, such as the store code, the loyalty number or the social login, relinks earlier browsing to the person on its own. Email, phone and device do not do that on their own, because families share contact details and tablets, and joining two people by mistake is the expensive error in this sector. In those cases, merging stays a human decision.

Merges and the Undo merge button

When two records are unified, the Merges for this customer section records When it happened, how many Events moved changed owner and the status. The number of events moved is the honest measure of how big the damage is if the merge was wrong: it is exactly what goes back where it belongs when you undo it. The Undo merge button returns the identifiers to the record that had been absorbed and restores it. The merge was designed to be reversible from the start, and not as a one way door.

Example: Marina Alencar bought on the site as marina@example.com and, months later, left the phone number (555) 888-1234 in store, which is also her husband's number. The two records were unified by mistake. The panel shows the merge from March 14 with 212 events moved. One click on Undo merge separates the two people again, and the 212 events go back to whoever generated them.

Customer trust

The panel shows a number from 0 to 100 with the scale written next to it, 100 = trustworthy. It starts at 100 and subtracts the weight of every active risk signal for that customer. A signal that was reviewed and dismissed stops counting. Next to the number comes the classification: a score up to 30 shows High, up to 60 shows Medium, above that shows Low. Pay attention to the scale, because reading 92 as high risk is the classic way to block your best customer.

The full queue of signals lives on the Risk tab, with the columns Signal, Subject, Score and Status. There, Score is the weight of the signal, that is, how much it subtracts from the trust of whoever is linked to it. The vertical flags and organizes the review; it does not block a payment or cancel an order.

Household and preferences

The Household section shows the household name or address and the other people who live there. It serves a practical purpose: not sending the same campaign twice to the same address. That is the difference between marketing and nuisance. When there is nobody else in the household, the screen says No other people in this household. The Preferences section lists what was recorded for that person, in key and value pairs, for example a size or a preferred category, and only appears when there is something to show.

Limits and common questions

  • Undo merge is the only write on this screen. Everything else is read only.
  • No signal showed up? The message No identity signal recorded means the platform has not sent identifiers for that customer yet.
  • The Merges section only appears when a merge exists. A customer who was never unified does not show the section.
  • Trust is recalculated from the current signals. Dismissing a signal in review makes trust go back up.
  • The Retail vertical has to be on. Without it, the screen does not even open.

Open this article inside the system

Read it and want to see it working?

The account is free and the whole manual is available inside the system, with an assistant that answers from this very content.

Create free account
Customer identity and the 360 view · how retail recognizes a person · Sellio