Capability / Product data & reporting
Product data and reporting, done properly once
Two kinds of tedious work that AI handles well when it is given rules: filling and fixing a product catalogue, and assembling the regular performance report.
Product data
An online catalogue is only as good as its data. Missing sizes, inconsistent colour names, thin descriptions and empty attributes hurt on-site search and filters, and they are a common reason shopping feeds reject products. In most small brands the data arrives in a supplier spreadsheet in a different shape every time, and someone re-keys it.
What a product data build does
- Reads supplier sheets, spec PDFs and your own notes, and maps them to your catalogue fields.
- Normalises values to your rules: “Navy” not “navy blue”, “One size” not “OS”, metric measurements in a set format.
- Drafts titles and descriptions in your house style from the facts it has, and marks anything it couldn’t confirm.
- Checks for gaps: missing images, missing alt text, missing required feed attributes.
- Stages everything as a batch for approval before it is written to the store.
What stays with a person
Pricing, product claims (especially anything about health, sustainability or origin), and the final approval of each batch. The build never publishes straight to a live store without a person approving the batch, unless you deliberately set a narrow rule otherwise.
Illustrative example
Made-up scenario, not a client. A small Melbourne skincare brand stocks 400 SKUs across its own range and two distributed brands. Distributor data arrives as spreadsheets with different column names each season. A product data build maps both formats, fills ingredients and size fields, drafts descriptions in the brand’s voice, and flags any ingredient claim that isn’t in the source sheet for the owner to check.
Reporting
The weekly or monthly report is where agency and in-house time disappears quietly. Someone logs into the ad platforms, analytics, the email tool and the store; exports; pastes into a deck or a spreadsheet; writes the commentary; formats it; sends it. The commentary is the valuable part, and it is the part that gets rushed.
What a reporting build does
- Pulls the agreed metrics from each platform on a schedule, through official connections where the platform offers them.
- Assembles them into your existing report format, whether a document, slides or a spreadsheet, so clients and owners see the familiar layout.
- Drafts plain-English notes on what changed and the likely reasons, clearly marked as a draft.
- Flags anomalies, such as a tracking drop, a spend spike or a campaign that stopped delivering, for a person to look at.
- Waits for the account lead or strategist to edit and approve before anything is sent.
Why the notes stay human-approved
An AI can see that conversions dropped 30% week on week. It may not know that the client paused their best-selling product, or that the tracking tag broke on Tuesday. The draft speeds up the writing; the strategist’s knowledge makes it correct.
Data access
Reporting builds use read-only access wherever the platform allows it, through accounts you own and can revoke. We list every connection in the handover document.