Data Retention

A vs B keeps the commerce data it collects for a limited time, then deletes it for good. How far back you can actually see that data in your reports depends on whether your organization has billing details on file. This page explains both numbers, what automatic deletion does, and why it is different from erasing one shopper on request.

The 730-day storage ceiling

A vs B stores your commerce and order data for up to 730 days (about 24 months). After that, it is deleted automatically. The clock on each record starts the moment it is recorded, so the oldest data always goes first.

Five kinds of data share this 730-day ceiling:

  • Tracking events: the page views, clicks, and custom events the snippet sends.
  • Metric events: the conversion and revenue signals your experiments measure.
  • Orders: the order records used for revenue, attribution, and audiences.
  • Order line items: the individual products on each order. These age out with their parent order, not on a clock of their own.
  • Profit records: the nightly profit figures computed from your orders and product costs.

One category is kept for a shorter, separate time. Error diagnostics (the technical logs behind experiment errors) roll off after 90 days, because they are short-lived troubleshooting data with no long-term value.

What billing details let you see

730 days is the outer limit A vs B stores data for. Whether your organization has billing details on file decides how far back your reports can actually reach:

Billing detailsHow far back your reports go
Not on file90 days
On file365 days

See Billing for more.

If your window is shorter than 730 days, older data is not deleted early. It simply stops appearing in your reports, hidden rather than gone. It comes back into view the moment you add billing details, or it hits the 730-day ceiling and is deleted for real, whichever happens first.

Adding billing details widens the window straight away

Add billing details and your reports can reach further back the moment it lands. Nothing was deleted while it was out of view, so there is no waiting period and nothing to re-collect.

Finished experiment summaries never expire

Your experiment list is the one exception to both numbers above. The result of every experiment you have run, including its lift and its verdict, stays on your experiment list for as long as your organization exists. This applies whether or not billing details are on file.

What automatic expiry does

Two different things happen to a record, and depending on whether billing details are on file, they can happen at two different times:

  1. It can drop out of your reports the moment it passes your window. A record on an account with no billing details on file stops appearing after 90 days, even though A vs B has not deleted it yet.
  2. It is permanently deleted once it passes the 730-day ceiling. This happens the same way whether or not billing details are on file, and it needs no action from you.

This second step is housekeeping on a timer. It runs on a schedule and is never triggered by (or waiting on) a shopper request.

Retention is not the same as erasure

It is easy to blur "this record is past my reporting window" together with "a shopper asked to be deleted." They are two separate mechanisms:

Automatic retention (this page)Erasure
Why it happensHousekeeping on a timer, or your reporting windowA shopper (or Shopify) asks to be forgotten
What it coversEvery record, once it is old enoughOne specific shopper's entire footprint
WhenUp to 730 days after each record, or sooner if your window is shorterRight away, on request
Who triggers itNobody: it is automaticYou, or a Shopify compliance webhook

Do you need to honor a "delete my data" request, under GDPR, CCPA, or a similar rule? Erasure is the mechanism to use. You do not wait for retention to catch up. Erasure removes that one shopper's orders, events, and profile right away, and leaves everyone else's data untouched.

Visitor profiles

Separately from the data above, A vs B keeps a small visitor profile for shoppers who have purchased. It stores a few facts: whether the shopper has bought before, how many orders they have placed, and their lifetime spend. A commerce audience (a named, reusable group of visitors defined by targeting rules) can use these facts to target a rule like "has purchased in the past."

Each visitor profile lives for 90 days and is refreshed only while the shopper stays active. In practice a profile ages out about 90 days after the shopper's last purchase activity. A shopper who stops buying quietly falls out of profile storage; a shopper who keeps buying keeps a current profile.

Inactive shoppers age out

The nightly job that maintains profiles only refreshes profiles for shoppers with recent order activity. A shopper with no recent activity is left alone, so their profile genuinely expires 90 days after their last purchase rather than being kept alive indefinitely.

Erasing a shopper also deletes their visitor profile at once: you do not wait for the 90-day clock to run down.

Changing your retention window

Whether billing details are on file sets your retention window, using the table above. There is no separate, self-serve setting for it: add billing details on the Billing tab and the window widens the moment they land.

Do you need something outside the standard 90 or 365 days, such as a longer window on an account without billing details? Get in touch, and we can set up a custom arrangement for your organization.

Was this helpful?