Buyer Insights

Buyer Insights

The shop can record how buyers browse it, and reports on what it finds.

Where to find it

Architect Panel → Commercial:

• Shop Insights — the five reports
• Carts — the option that turns recording on

The five reports

• Product search impressions — which products appeared in search results.
• Product views — which were actually opened.
• Most common search keywords — what people typed.
• Most viewed categories.
• Most selected filters, as percentages.

Each is shown twice: all time, and the past thirty days. The comparison is where the value is — a keyword that is high all-time and absent recently is a product going out of fashion, or a listing that has broken.

Impressions against views is the useful pair

A product appearing in results constantly and rarely opened has a problem with its image, its title or its price. That gap is the single most actionable thing on the page, and it is invisible in sales data because the product never sold.

Search keywords are a shopping list

People search for what they want, including things you do not stock. Keywords with no good result are either a product to add, a synonym to handle, or a category name to change.

This is the cheapest product research available to a shop, and most shops never look at it.

Filters tell you what to keep

The filter report is percentage-based, so it says directly which filters buyers use. Anything at a couple of per cent is clutter; the top two deserve prominent positions.

Recording is optional

Insights are a cart option, off unless enabled. That is a real decision rather than a formality — you are recording browsing behaviour, and browsing behaviour is personal data.

Say so, and mean it

If you enable insights, your privacy notice should say what is recorded and why, and your cookie handling should match. Recording quietly because a switch was available is exactly the practice regulators look for.

Act on it or turn it off

Data collected and never read is pure liability — a personal-data holding with no benefit to set against it. Either put the report in somebody’s monthly routine, or switch it off.

Worked example

A shop reviews insights monthly. The impressions-to-views gap identified four products with poor images, which were reshot. Search keywords produced two products the shop did not stock and now does, and a term customers used that the catalogue did not — which was added as a category synonym.

Recommendations

• Compare thirty days against all time, not either alone.
• Work the impressions-to-views gap first.
• Read search keywords as product research.
• Say what you record, and read it monthly or stop.

Wishlists and Reviews

Two customer-facing features that produce information about your catalogue as a side effect.

Where to find it

Architect Panel → Commercial:

• Carts — the cart itself and its options

Architect Panel → Data:

• Datastores — then View Data on the cart table, and look for the row actions

Lists and wishlists

A customer can keep named lists, each holding products with additional information. So a buyer can maintain "kitchen refit" and "birthday ideas" separately rather than one undifferentiated saved list.

For trade customers this is a re-order list, which is a genuine convenience and a reason to sign in.

Lists are demand signals

A product saved often and bought rarely is priced wrong, out of stock too often, or has a delivery cost that kills it at checkout. That is a different signal from views, and a stronger one — saving something is a deliberate act.

Reviews

A review carries a rating, a title and content, against the master product or a specific variant. Ratings are cached for display, so listings can show them without recalculating.

Reviews are a cart option

Off unless enabled, and worth thinking about before switching on. Reviews raise conversion and they also publish criticism of your products on your own site, permanently.

For most shops that trade is worth it. For some — bespoke work, professional services, anything where a single unhappy customer is unrepresentative — it is not.

Moderate, but publish the bad ones

Moderation should remove abuse, personal data and things that are not reviews. It should not remove criticism.

A product with nothing but five-star reviews is not believed. The occasional three-star review is what makes the rest credible, and buyers actively look for them.

Reply to the worst ones

A bad review with a considered reply reads better than a bad review alone, and the reply is aimed at the next reader rather than the reviewer.

Reviews are personal data

Written by identifiable people, published indefinitely. Have an answer for somebody asking for theirs to be removed, and be careful that review display does not reveal more about the reviewer than they expected.

Watch what reviews say about a product

Three reviews mentioning the same problem is a product fault, not a customer problem. Reviews and refund reasons together are the clearest quality signal a shop has.

Worked example

A shop enables reviews with light moderation, removing only abuse, and replies to anything below three stars. A pattern in the reviews of one line matched its refund reasons exactly — a fitting that failed — and the product was withdrawn before it became a recall.

Recommendations

• Publish critical reviews — they make the rest credible.
• Reply to poor reviews for the next reader.
• Read lists as demand for products that do not convert.
• Read reviews with refund reasons to spot faults early.