Remove a company-store product when it no longer serves a needed role, has had a fair chance to be selected, and has no unresolved service obligations. Low sales alone are not enough. First check whether employees could find it, afford it within their allowance, choose a suitable variant and receive it where they work. Then choose one of four actions: retain, repair, pause or retire.
This is a review method for a live store, not a rule to delete the bottom quarter of a sales report. A quiet non-apparel alternative may make the program usable for people who do not want clothing. A popular jacket may deserve a pause if orders repeatedly require replacements. The decision is about useful choice and dependable fulfillment, not simply the longest bar on a chart.
Publisher perspective: Brand Sauce publishes this guide for company-store buyers. The review framework and examples below are our editorial recommendations, not industry benchmarks or a claim that every store platform provides these reports. Confirm the data and controls available in your own program.
1. Give the review a specific decision
Begin with the reason a product is under review: suspected redundancy, repeated delivery trouble, weak selection, a discontinued blank or a seasonal campaign ending. Do not start with an arbitrary target catalog size. A smaller assortment is useful only if it still covers the people and occasions it exists to serve.
Write the product's current role in one sentence. For example: “A non-apparel welcome gift that can ship to every location in this campaign.” Compare that role with the alternatives, not just with the best-selling hoodie. Our company-store curation guide explains how to establish those roles before launch; this review determines what the evidence now says about keeping them.
| Decision | Use it when | Required next step |
|---|---|---|
| Retain | The item fills a needed role and service is acceptable | Record why it stays and when to review again |
| Repair | Information, placement or a fixable ordering issue blocks a fair test | Assign one change, an owner and a new observation window |
| Pause | Availability or suitability is unresolved | Stop new commitments and resolve open orders separately |
| Retire | The role is redundant, no longer needed or better served by an approved replacement | Close the listing without abandoning prior recipients or stock |
Safety concerns, recalls and confirmed product failures should enter the appropriate response process immediately. They do not need to wait for a merchandising review or a minimum number of orders.
2. Test whether the item had a fair opportunity
Use a review window that covers a meaningful campaign or buying cycle. Thirty days may tell you little about an annual recognition gift. A winter layer introduced after the cold-weather campaign should not be compared uncritically with a bottle available all year.
For each item, record the launch date, eligible audience, campaign placement, allowance rules and periods when useful variants were available. “Listed for three months” is not the same as “selectable for three months.” An apparel style with only one size left may technically remain live while being irrelevant to most shoppers.
- Discovery: Was the item visible in the collection employees were actually invited to use?
- Eligibility: Could that audience select it under the campaign rules?
- Affordability: Did the allowance cover the required total, or did it introduce an employee payment?
- Variant coverage: Were relevant sizes, fits and colors genuinely orderable?
- Destination: Could it be fulfilled to the intended recipients?
- Timing: Did the displayed lead time fit the occasion?
Keep unknowns visible. If you lack historical availability records, do not reconstruct precise “available days” from memory. Start recording them prospectively and mark the current conclusion provisional. A question such as “Was your preferred size unavailable?” can supplement the record, but optional feedback is not a census of everyone who visited.
3. Keep units, people and events separate
A manager buying 40 shirts is not 40 employees independently choosing that shirt. Likewise, repeated product-page visits are not necessarily different people. Separate recipient-choice campaigns from manager bulk orders, mandatory uniform issues and replacements before interpreting product popularity.
For stores using GA4, Google's Ecommerce purchases report documentation explains that item views and purchases depend on submitted ecommerce events; ecommerce collection is not automatic in every implementation. A zero in a report can therefore require an instrumentation check before a merchandising conclusion.
Google also distinguishes event-scoped and item-scoped metrics. One purchase event can contain multiple units. For this review, label every number: units ordered, distinct recipients choosing, product views or orders containing the item. Do not divide one by another and call it employee preference without explaining the denominator.
Build a modest evidence sheet rather than waiting for a perfect dashboard:
| Field | What to record | What it prevents |
|---|---|---|
| Identity | Style, variant and decoration version | Merging unlike products under one familiar name |
| Opportunity | Campaign, eligible group, placement and known availability gaps | Calling an inaccessible item unpopular |
| Selection | Recipient choices separately from bulk and replacement units | Confusing purchasing volume with broad demand |
| Delivery | Fulfilled orders, cancellations and unresolved commitments | Counting checkout as a completed experience |
| Service | Fit, decoration and product issues with specific reasons | Hiding costly problems behind high volume |
| Coverage | Needs or audiences that would lose an option | Removing the only practical alternative |
Use aggregate counts for the assortment meeting. It generally does not need a list of which employee chose which size. If the underlying system cannot distinguish free choice from assigned items, say so and avoid preference claims until the records can support them.
4. Read the evidence before ranking products
Illustrative scenario only—not client data or a recommended performance threshold: imagine a campaign in which 120 employees are each allowed one item. The finalized record shows 96 recipients chose something: 54 selected a bottle, 30 selected a sweatshirt and 12 selected a tote. The other 24 have no completed selection.
Among completed selections, those shares are 56.25%, 31.25% and 12.5%. Among all eligible employees, they are 45%, 25% and 10%. Neither view proves satisfaction, and neither explains why 24 people did not choose. Keep the count and denominator beside the percentage.
Now add context: the sweatshirt was unavailable in several sizes during part of the campaign. The tote was the only bag and served people who already owned suitable drinkware. Eight bottle orders produced unresolved lid-related support reports. Those reports are a reason to investigate the exact product and issue; they are not, by themselves, eight confirmed manufacturing defects.
The defensible outcome is not “delete the tote because it came last.” Retain the tote's role while gathering usefulness feedback; repair sweatshirt availability before evaluating demand; investigate the bottle issue despite its strong selection count. If a second near-identical bottle is proposed, ask which unmet need it would serve before adding it.
For small counts, one or two choices can move a percentage substantially. Treat a modest pilot as directional evidence, not a statistically established winner. Explain what further observation would change the decision.
5. Run a bounded repair, not endless probation
A repair should name a plausible cause and one observable improvement. Replace missing dimensions, clarify a fit chart, correct a broken allowance rule or give a buried item a comparable placement. Do not change the product, price, campaign audience and homepage position together, then attribute the outcome to one of them.
Define the next observation period before making the change. Compare similar campaigns where possible, and record differences in season, audience and allowance. This is an operational check, not a controlled experiment unless you have actually designed one.
Some repairs belong with the supplier rather than the product page. Recurring decoration trouble may require a different placement or a new finished sample. Repeated blank shortages may justify a replacement style. Evaluate that candidate against the original role and use the branded apparel sample approval process before opening replacement orders.
End the repair with a decision. If the same unresolved constraint persists through the agreed review, pause or replace the item rather than repeatedly moving its deadline. Retain a low-volume option deliberately when its coverage value is clear; do not pretend it passed a popularity test it never needed to pass.
6. Separate assortment decisions from inventory decisions
For owned stock, sell-through helps describe movement. Shopify's product analytics documentation defines its product sell-through calculation using units sold and units remaining in inventory. That is an inventory measure, not a complete employee-choice score.
An on-demand assortment may have no employer-owned finished inventory to clear. Do not manufacture a sell-through ranking by treating every zero-stock listing as a success. Review selection, coverage, availability and service instead. Blank-supplier inventory and production capacity still matter even when your company holds no decorated stock.
Bulk can remain appropriate for a confirmed event or mandatory uniform issue. A product with little discretionary demand might be exactly right for that assignment. Hybrid programs can retain a small justified reserve while handling less predictable selections individually. Compare those purposes separately rather than forcing all orders into a single bestseller list.
Retiring a listing also does not determine what happens to physical leftovers. Count remaining goods, identify confirmed uses and assign a disposition owner. The company swag inventory audit guide covers that distinct decision. Do not quietly distribute unwanted stock to make the assortment review look successful.
7. Close the listing without breaking the program
Before retiring an item, check active carts where your platform exposes them, pending approvals, placed orders, reserved stock and campaign communications. Confirm the provider's actual controls; hiding a product is not necessarily the same as preventing an already-created order from proceeding.
- Record the decision: item/version, reason, evidence window, owner and effective date.
- Resolve open commitments: fulfill as agreed or obtain approval for a documented alternative. Never treat retirement as permission for silent substitution.
- Protect coverage: identify which recipients or use cases lose an option and approve the replacement where needed.
- Update the journey: remove stale collection tiles and campaign links; test direct product access and the intended replacement path.
- Preserve support history: retain the old style, decoration and order references needed to answer recipient questions.
- Review the result: check whether the gap was filled and whether employees can complete the intended selection.
A useful closing note is concrete: “Retire duplicate style B after outstanding orders close; retain style A for the same role; confirm the replacement covers the required sizes; review support and availability after the next campaign.” That is more actionable than “refresh the catalog.”
Company-store assortment review FAQs
How often should we remove products?
Review after meaningful campaigns and when availability, quality or program needs change. Choose a recurring cadence your team can maintain, but do not make deletion a quota. A review can reasonably conclude that the assortment should stay unchanged.
Should every product meet a minimum order count?
Not necessarily. Required uniforms, specialty fits and non-apparel alternatives may serve smaller audiences. Establish their role and coverage value first. If you use an internal volume threshold, document the exceptions and the observation window.
What if we have order exports but no browsing analytics?
Start with campaign scope, order types, availability notes and support reasons. Inspect the actual shopping journey. You can identify duplicates and repeated fulfillment problems without page-view data; you cannot reliably diagnose discovery from orders alone.
Can strong sales justify keeping a troublesome item?
Demand is relevant, but it does not cancel service or suitability problems. Investigate the cause, set a repair deadline and pause new commitments when appropriate. The goal is an assortment employees can successfully receive and use, not a report with attractive order totals.
Primary documentation checked October 4, 2026. All example counts and review rules are illustrative editorial guidance; no platform integration, product availability or performance outcome is implied.
