Can't tell why your sales dropped? Sledge reads your store data and ranks the leaks by what fixing them is worth. Free for 14 days →

// Level 04 The Fixes · 4.8

Illustrated golden aurora over a narrow violet canyon, with a river winding out of the waterfall below.

Product discovery

Updated August 2026 7 min read
EVERY WAY A SHOPPER REACHES A PRODUCT NAVIGATION browsing, by category SEARCH 15 to 30% of visitors RECOMMENDATION shown, not sought DIRECT LINK an ad, one product only THE PRODUCT PAGE NO ROUTE AT ALL the shopper leaves, and the exit is identical to disinterest in every report you have A DIRECT-LINK STORE HAS EXACTLY ONE ROUTE, AND IT LEADS TO ONE PRODUCT.

The bottom bar is the one to notice. Nothing in a standard report distinguishes it from a shopper who simply was not interested.

The short answer

There are four routes to a product page: navigation, search, a recommendation, and a direct link from an ad. Most dropshipping stores have one working properly, usually the direct link, which reaches exactly one product per campaign. Everything else in the catalogue depends on the other three, and when they fail the result is not a slow product, it is an unreachable one. That distinction is invisible in every standard report, because an exit caused by not finding something looks exactly like an exit caused by not wanting it.

The four routes

Every product page view in your store arrived through one of four doors. Listing them out is more useful than it sounds, because most stores discover they have been relying on one.

RouteReachesFails whenCovered in
NavigationEvery visitorCategories use your words, not theirsThis chapter
SearchThe 15 to 30% who use itSynonyms and misspellings return nothing4.9 and 4.12
FiltersShoppers with a constraintA valid combination empties the page4.10
RecommendationWhoever is shown oneIt only ever shows the top sellers4.4
Direct linkOne product per campaignThe ad stops running2.2

Now look at the last row honestly. A store that grew on paid traffic to a single landing page has one route, it costs money every time it is used, and it stops the moment the campaign stops. That is not a discovery system. It is a rental.

Why this failure is silent

A shopper who cannot find a product does not complain, does not search twice, and does not tell you. They leave. In your reports that produces a session, a short duration, and an exit.

Which is identical to what a shopper who looked, considered, and decided against it produces.

The two are different problems with different fixes, and every standard report collapses them into the same row. Chapter 3.4 covers why that is structural rather than an oversight: the distinguishing information is an event, and events are not retained unless something retains them.

An unreachable product and an unwanted product produce the same number. Only one of them is fixed by changing the product. The Dropshipping Playbook

Search reaches the minority who use it. Filters reach the ones with a constraint. Navigation reaches everyone, which makes it the first thing to fix and the thing most often broken in a way the owner cannot see.

Three failures, in the order they occur:

Naming: the category is called the wrong thing

Supplier feeds arrive with supplier vocabulary, and it survives into the menu because it is already there and it is technically accurate. A shopper looking for a "phone stand" does not open "Mobile Accessories > Desktop Mounting Solutions".

This is invisible from the inside, because you know what the words mean. It is instantly visible to the stranger in chapter 3.2's fifteen-minute test.

Depth: too many decisions before a product

One level is comfortable at almost any catalogue size. Two is fine when the branches differ meaningfully. Three or more is where chapter 3.5 puts the depth at which nobody can find a product.

Depth usually accumulates rather than being designed. Each new subcategory was reasonable on its own.

Coverage: products that belong to no collection

An import that missed a tag, a manual product added in a hurry, a collection rule that stopped matching after a variant change. The product exists, has a URL, and is in nothing. The only routes left are search and luck.

Check this directly: how many products in your catalogue belong to zero collections? On most stores past a few hundred products, the answer is not zero.

A discovery audit, in one sitting

Half of this is checkable with no tools at all. The other half needs the discovery layer recorded, and if it is not, note it as unknown rather than passing it, per chapter 3.7.

Fix order, and why

  1. Coverage first. A product in no collection cannot be found by browsing at all, and fixing it is data entry rather than design.
  2. Naming second. Free, and it improves every subsequent route including search.
  3. Depth third. Flattening a menu is disruptive, so do it once the names are right.
  4. Search fourth, per chapter 4.9, because it serves a minority but a high-intent one.
  5. Filters fifth, per chapter 4.10.
  6. Sort order last, per chapter 4.11, because it only matters once the right products are reachable at all.

The whole sequence is one change at a time with the date written down. Discovery work is particularly easy to do all at once and particularly hard to attribute afterwards, because every change affects every other route.

Common questions

How do I know if a product is unreachable rather than unwanted?
Check whether it has any views at all. A product with traffic and no sales is unwanted, or priced wrong, or badly presented. A product with almost no views is unreachable, and no amount of rewriting its description changes that. Chapter 3.5 covers the catalogue size at which this starts happening in bulk.
Should I fix navigation or search first?
Navigation, if a stranger cannot find a named product in thirty seconds without using search. Search, if they can browse fine but a meaningful share of searches return nothing. If both are broken, navigation first: it serves every visitor, while search serves the 15 to 30% who use it.
My store only sells one product. Does this apply?
Less of it, but not none. A single-product store still has a discovery problem if the product page is the only page, because there is nothing to show a visitor who was not convinced the first time. Variants, bundles and complements are what give a one-product store somewhere to go.
How many categories should my navigation have?
Few enough to read without a submenu on a phone, and named in the words your customers use rather than your supplier's. Community store audits keep flagging navigation that is too complicated to find a product in as a top conversion problem, and the fix is almost always fewer, clearer categories rather than more precise ones.
Illustrated violet ridges receding to a yellow sun, over a field of pale flowers.

The app behind these fixes

Most of Level 04 is a setting in Sledge rather than a project.

See it on the Shopify App Store