Store Statistics Pro
Recommendations built from your own completed orders, so you show products that genuinely sell together instead of products you guess might go well together.
Store Statistics is one of the two choices under How Products Are Chosen in the Setup step. With Rules (Default), you describe which products to pick from, using filters and sorting. With Store Statistics, the plugin reads your order history and finds the products that are really bought together.
It's a Pro feature. On a Free store it's in the list but locked.
What it does
For every pair of products that have appeared in the same paid order, the plugin asks one question: is this pair meaningful, or a coincidence? Pairs that pass become recommendations. Pairs that don't are ignored completely, not shown with lower priority.
That's the whole point of the feature. Simply counting "these were bought together twice" doesn't work on a real store, because two products that both sell well end up in the same orders all the time without being related at all.
Setting it up
In the Setup step, set How Products Are Chosen to Store Statistics. Two settings appear:
| Setting | Options | Default |
|---|---|---|
| Order History Window | Last 7 Days · Last 14 Days · Last 30 Days · Last 90 Days · Last 6 Months · Last 12 Months · All Time | Last 90 Days |
| Minimum Co-Purchases | Any · 5+ Orders · 10+ Orders · 20+ Orders · Custom | 10+ Orders |
Order History Window: "Which past orders to learn from." A longer window gives more data and smooths out seasonal swings. A shorter one reacts faster to what's selling right now. 90 days suits most stores.
Minimum Co-Purchases: "How often products must be bought together." This is your confidence setting. A pair is only used if the two products were bought together in at least this many orders. Raise it on a busy store; lower it on a quieter one. Custom lets you type any whole number.
Lower the minimum before you widen the window
On a store with modest sales, 10 orders together in 90 days is a high bar, and you may get very few pairs. Dropping to 5+ Orders usually produces results sooner than stretching the window to 12 months, and the results stay more current.
What counts as a co-purchase
Three rules, worth knowing because they explain results that might otherwise look odd:
- Only paid orders your store still counts. The plugin uses WooCommerce's own list of paid order statuses, minus any statuses your store excludes from its reports. Cancelled and refunded orders don't count.
- Variations count as their main product. A recommendation for "the blue medium shirt" is a recommendation for the shirt, so pairs are counted at the main-product level.
- Subscriptions count as their base product type, because the plugin reads the order's line items directly.
How a pair qualifies
A pair has to pass four tests, not one. They're the same thresholds WooCommerce's own Product Recommendations extension publishes, used here deliberately rather than invented:
| Test | Bar | What it prevents |
|---|---|---|
| Orders containing the product | at least 10 | Drawing conclusions from a handful of orders |
| How often the pair appears together | in at least 5% of the product's orders | Pairs that shared the odd order but aren't related |
| JLH significance | at least 0.2 | Popular products looking related just because they're popular |
| Similarity (normalised distance) | at least 0.8 | A rare product that happened to appear with a best-seller once |
The middle two tests are what make this different from counting. JLH measures how much more often product B appears with product A than it appears in general, so your best-seller doesn't become "frequently bought together" with everything in your catalogue. The similarity test catches the opposite mistake: a product sold twice ever, both times alongside your best-seller, looks strongly linked but has almost no evidence behind it.
Minimum Co-Purchases can only make this stricter. There's no way to lower the four tests from the builder.
When there isn't enough data yet
A new store has no order history, and the builder tells you so before you save, every time:
New store? "Store Statistics needs order history, so until enough data builds up it falls back to category & popularity."
That's exactly what happens. Until a product has pairs that pass the tests, a Store Statistics recommendation picks its products the Rules way instead, and:
- The shopper sees a normal block. There's no error, no empty box and no "not enough data" message.
- Your reports know. Its sales are recorded as coming from your rules, not from store statistics, so the Analytics split by source stays honest.
As orders come in, the recommendation switches to the mined pairs on its own. There's nothing to change.
When the mining runs
Never while a shopper is waiting. Reading through orders happens in the background, using WooCommerce's Action Scheduler:
- As orders are paid. When an order reaches a paid status, it's folded into the pairs.
- On a schedule. A consolidation pass runs every 24 hours. There's no setting for this in the admin yet.
A large store is processed in batches across several background runs, rather than in one long run that could time out.
Checking and rebuilding it
If you have shell access, two WP-CLI commands help:
# Which order-history windows have mined pairs
wp sprcm stats status
# Rebuild everything from scratch
wp sprcm stats rebuild
# Rebuild one window, without emptying the existing pairs first
wp sprcm stats rebuild --window=90d --keepstatus lists each window and whether it has any mined pairs yet. --keep matters on a live store: without it, the existing pairs are deleted first, so Store Statistics recommendations fall back to rules until the rebuild finishes. If no pairs qualify, the command tells you the store is below the thresholds. That's the normal "not enough data yet" situation, not a failure.
Full reference: WP-CLI.
Store Statistics vs Frequently Bought Together
These two get confused a lot:
| Frequently Bought Together | Store Statistics | |
|---|---|---|
| What it is | A sorting rule | A way of choosing products |
| Where it is | Setup step → Sorting | Setup step → How Products Are Chosen |
| What it does | Puts the products your filters already picked in order of how often they're bought with this one | Picks the products themselves, from the mined pairs |
| Tier | Free | Pro |
Both read the same mined pairs, and only Pro does the mining. So on a store without Pro, the Frequently Bought Together sort has no data to work with yet. It doesn't break anything: it steps aside, and the block is ordered by your other sorting rules. Once Pro is mining your orders, the sort starts to work on its own, on Free recommendations too.
Others Also Bought Pro is a third, related sort that reads the same pairs.
What's kept
Mined pairs are built from your orders, so they're never the only copy of anything. You can rebuild them at any time with the command above, and deleting them never touches your orders.
Next
- Setup step: the control in context.
- Analytics: checking whether it's working.
- WP-CLI: the statistics commands.