First, split the product's sales into visits, add-to-basket and orders against like weeks
Rule out stock first. A product that sold out, even briefly, and has sold less since it came back belongs to the diagnosis for sales not recovering after a restock. A missing size or colour is a partial stock-out, so check every variant buyers used to choose. If the product slipped just after a sale that took its stock or pulled its orders forward, read the sale with the diagnosis for a sale or promotion underperforming as well. And if the product still sells but the baskets around it shrink, the diagnosis for a falling average order value is the better start.
Then split the fall. The KPI tree framework keeps visits and conversion on separate branches: the product's orders are its page visits, multiplied by the share that add it to the basket, multiplied by the share of those baskets that become orders. Google's e-commerce developer guide for Analytics documents item-level view, basket and purchase events that carry the item, and the product and merchandising analysis guide reads them as a product funnel. The item's add-to-cart rateShare of sessions in which a product was added to the cart. and its conversion rateShare of visits or clicks that convert. from view to order are the ratios to watch.
Read every number against the same weeks a year earlier, not last month. The guide to comparing periods aligns weekdays and moving holidays and separates growth from SeasonalityPredictable, repeating changes in demand tied to the calendar, such as months, holidays, weekdays or hours. with the help of the business's own records. Then split the visits by source.
| What fell | What held | Where to look first |
|---|---|---|
| Visits from shopping or catalogue ads | Organic and internal visits | The listing, then the campaign structure |
| Visits from organic search | Paid and internal visits | The product page's rankings |
| Visits from the store's own lists and banners | External visits | The homepage, menu and collection places |
| The add-to-basket rate | Visits from every source | The price elsewhere, then the reviews |
| The store's orders of the product | The brand's other channels, which sell more of it | The price elsewhere |
| Visits and conversion, with the category | Other categories | Demand for the product and its role |
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Paid shopping or catalogue impressions fell; the page did not change | Item status in Merchant Center diagnostics, the Google Ads products report and the Meta catalogue | The listing is limited, disapproved or matched to fewer searches | Correct the item at source; restore titles that lead with buyers' words |
| Paid spend on the product fell while the campaign's spend held | The account's record of changes to listing groups, labels and targets around the start of the fall | A structure, label or target change moved paid support elsewhere | Return the product to a group and target that fit its margin and history |
| Organic clicks to the page fell while other pages held | Search Console queries; an archived copy; today's results for its main query | Rankings lost after an edit, an address change or a results page shift | Restore what ranked, redirect a moved address or rework for the new intent |
| Basket additions fall, or the brand's other channels sell more of it while the store sells less | The price on the brand's marketplace listing, at stockists and at rivals | The same product is cheaper one click away | One headline price; compete on bundles, loyalty and service |
| Visits hold, the add-to-basket rate falls, no price change anywhere | Newest reviews, the rating over time, themes since a batch or specification change | Recent critical reviews lead the page or the rating fell | Answer publicly, fix the fault or the page, never remove reviews |
| Visits from the store's own lists fell while external visits held | Item list views and selections; homepage, menu and collection changes | The product lost its homepage, menu or collection places | Give back permanent places and restore them at every take-down |
| The product falls with its category, or a newer product takes its orders | Like weeks a year earlier; search demand; launches in the range and by rivals | Demand moved and the product's role changed | Review the line's role and set its support to match |
Fix, verify and prevent each cause
A listing that lost approval or searches
A product campaign has no keywords: the feed decides which searches an item is eligible for and its title which of them it wins, as the product feeds guide explains. Google's product data specification makes price, availability and the link required, and the guide records that an item whose price or availability disagrees with its landing page can be disapproved. Less visibly, a feed rule or supplier update can rewrite a title so it no longer leads with the words people search for, and the item stays approved while matching fewer searches.
Fix: follow the diagnosis for Shopping products not showing: find the item in Merchant Center diagnostics and the Google Ads products report, correct any disapproval or limitation at source, and restore a title that opens with the product type and the attributes people search for most. For Meta, the diagnosis for catalogue products not showing works outwards from the item's status through the product set and the feed.
Verify: the item shows approved and its impressions return in the products report.
Prevent: run the feed guide's weekly diagnostics in revenue order, because a disapproved best-seller matters more than many long-tail items, and test every title rule against the lead products before it goes live. The product feedA structured file of a store's products, prices and availability used to build shopping and catalogue ads. is a controlled document.
A campaign change that moved the paid support
A campaign with one target spends most on the products with the strongest history, as the guide to new, seasonal and clearance products in Shopping notes, and that changes when the structure does. A label refreshed from data can move a best-seller into a group with a tighter target, a new launch or clearance group can take its budget, and a core-products group can get a target that does not fit its margin. Google publishes no account of how Performance Max divides spend among products, so read the outcome.
Fix: set the product's weekly impressions and spend in the product report beside the dates of recent structure changes, and return it to a group and target that fit its margin and history.
Verify: its share of the campaign's impressions and spend returns and its paid visits follow.
Prevent: write down what every label value means, log each structure, label and target change with its date, and check the lead products the week after any change.
A product page that lost its rankings
Organic visits fall quietly, because nothing in the store looks different. The diagnosis for a rankings drop counts an edit that removed what ranked and a results page that now wants something else among its causes; a shortened description or a specification table turned into an image can remove the passages that matched searchers' queries. The guide to category and product pages sets out what a product page needs: a unique description, specifications in text, price and availability consistent with the feed, real reviews and identifiers.
Fix: set the live page beside an archived copy and restore what answered the query. If the address changed, give the old one what the guide gives a discontinued product with a direct successor: a permanent redirect and updated internal links. If the results now show a different kind of page, rework the page or target a query it still fits.
Verify: once the page is recrawled, impressions return for the lost queries.
Prevent: log every edit to a lead product's page with a date and a reason, and snapshot the results for its main queries.
The same product cheaper one click away
A shopper who finds the product cheaper elsewhere still buys it, just not from the store. The comparison of marketplaces with an own store keeps one headline price on every channel, since marketplaces watch for price parity, and lets the store compete on bundles, loyalty points, delivery thresholds and service. Stockists are handled as the comparison of wholesale with direct-to-consumer sets out, with any agreement about what a reseller charges checked by someone qualified, because the rules differ by country; nothing here is legal advice. A rival's promotion is answered as the guide to responding to competitors advises: value first, and any price response narrow, temporary and costed against the contribution margin.
Fix: align the brand's own listings to the store's headline price and give buyers a reason to buy in the store.
Verify: the store's share of the product's sales returns and its add-to-basket rate recovers.
Prevent: put the product's price on every channel the brand controls into the monthly read below.
Critical reviews at the top of the page
Reviews give a product page the social proofEvidence that other people have chosen, used or approved something, used to reduce a buyer's doubt. it cannot give itself, and the guide to reviews and social proof recommends showing the most recent first. A run of critical reviews after a batch or specification change therefore greets every visitor: visits hold while basket additions fall.
Fix: reply publicly and calmly; where a review points to a real fault, admit it and put the product right; where it shows buyers misread something, change the page, using the product page optimisation guide for what it should answer. Never remove reviews for being negative; the reviews guide limits moderation to abuse, spam and personal data.
Verify: newer reviews stop repeating the complaint and the add-to-basket rate recovers.
Prevent: read new reviews of the lead products weekly and route recurring themes to whoever owns the product and its supplier.
Lost places inside the store
Visits from inside the store depend on where the store puts the product. The guide to merchandising for campaigns ends each campaign by returning the homepage routes and banners to their usual state; if that state was written before the product became a best-seller, the take-down quietly removes it, and a collection re-sorted for a campaign can drop it down the list.
Fix: read the item list report, as the merchandising analysis guide describes, for the product's list views and selections before and after the change, then give it permanent places on the homepage, near the top of its collection and in the menu.
Verify: list views, selections and internal visits return to their earlier level.
Prevent: add the lead products' permanent places to every campaign's build checklist, so the take-down restores them.
Demand that moved
Sometimes nothing in the store changed: the season is turning, a newer product in the range is taking the orders, or a rival launched something buyers prefer. The product life cycle framework reads a line's stage from the business's own data, and the comparison of the product life cycle with diffusion of innovations names reading a seasonal dip as decline as its typical misuse.
Fix: compare the product with like weeks a year earlier and with its category; if the category holds while a newer product rises, the role has passed to it. Take the line to the review in the guide to choosing which products to put marketing behind, which looks for lines whose role has changed, and decide whether it stays the hero or hands the lead on.
Verify: the category holds and the product meets the measures of its new role.
Prevent: bring the lead products to every review of lines with their split and trend.
Prevention: a monthly read of the lead products, layer by layer
Every cause above leaves a trace on one layer before it shows in total sales. Each month, take the lead products from the review of lines and read the same split as the diagnosis, against like weeks a year earlier, then each layer.
| Layer | What to read each month | Owner |
|---|---|---|
| Listing | Approval and impressions in Merchant Center and the products report; catalogue status on Meta | Feed owner |
| Paid structure | Group, label and target; impressions and spend by week | Paid media owner |
| Organic | Search Console clicks and queries; edits logged since the last read | Search owner |
| Price | The price on the brand's own listings, at main stockists and at main rivals | Trading lead |
| Reviews | New reviews, the rating trend and recurring themes | Customer service and product |
| Store places | Item list views and selections; homepage, menu and collection positions | Merchandiser |
| Demand | The category trend, search demand and launches in the range or by rivals | Trading lead |
Keep the read short: one table per product, the change since last month and a decision for anything that moved. Log the store's own changes with dates in the same place, from feed rules and take-downs to page edits and repricing, because most causes above start with a change made for a good reason elsewhere. Read small numbers with care, and if a product's split keeps falling after a fix, run the diagnosis again rather than adding budget. Other store symptoms are collected in the e-commerce troubleshooting section.
A cookware shop finds why its best skillet slipped (illustrative)
Mardlevix Cookware traces its best skillet's two weak months
Mardlevix Cookware, an invented online shop selling cast-iron cookware, sells about 540 of its 26 cm skillet a month through the winter. In April and May it sells 330 and 310, while every size is in stock and the rest of the range sells within 4% of the same months a year earlier. Against the same weeks last year, product page views are down 38% and the add-to-basket rate holds at about 9%, so the shop splits the views by source.
Three things changed. Shopping impressions for the skillet fell by about half from 14 March, the day a new feed rule rewrote titles to lead with the collection name instead of "cast iron skillet 26 cm". Views from the homepage list fell to almost none after 2 April, when the spring campaign came down and the take-down restored a homepage layout written before the skillet became the best-seller. And on 20 March the brand's own marketplace listing was repriced from £45 to £41 after a fee change, and marketplace orders for the skillet rose from 60 to 150 a month.
The shop reverts the title rule so titles lead with product type, material and size, gives the skillet a permanent homepage route and the first position in its collection, sets £45 on both channels with a lid bundle only in the store, and adds the skillet to a monthly read of its 5 lead products. By July the store sells 510 skillets a month and the marketplace 70.
Common mistakes
- Blaming the product first. Most falls come from a layer of its route that changed.
- Reading against last month. A seasonal product sells less as its season ends; compare like weeks a year earlier.
- Skipping the variant check. A missing size or colour is a partial stock-out, not a demand problem.
- Adding budget before finding the layer. Money cannot serve a disapproved listing or restore a lost route.
- Matching a rival's price across the range. Keep any price response narrow, temporary and costed.
- Hiding critical reviews. The reviews guide limits moderation to abuse, spam and personal data, and the fault is still there.
- Taking a campaign down to an out-of-date homepage. The usual state should include today's best-sellers.
- Declaring decline too early. A decline label withdraws support and makes itself true.
