All insights
Retail Strategy & Innovation· By RetailSense Editorial 6 min read

Retail Product Trials: How to Read Sales During Stockouts

Interpret product-trial sales when stockouts interrupt selling opportunity. Separate shelf observations, inventory records and unknown demand.

Share

A trial item sells out and the buyer wants to reorder. Another store reports disappointing sales of the same item. Before treating either result as a product verdict, ask what customers actually had an opportunity to buy.

Stockouts interrupt that opportunity. Inventory records, shelf checks and transactions can help describe what happened, but they do not automatically establish continuous availability or the demand that went unobserved.

This article develops the availability question in our retail assortment-testing guide. The aim is a better buying decision from the evidence the trial collected, with the missing information kept visible.

Separate the four things the records describe

Begin with the exact item or variant, its intended selling location and the offer being tested. A different size, colour or pack is not evidence that this offer was available.

Keep four observations distinct:

  • Recorded sales: The transactions recorded for the trial item at their stated times.
  • System inventory: The quantity the inventory system reported, subject to the accuracy of those records.
  • Physical stock: Units located somewhere in the store or its stockholding areas.
  • Customer-facing availability: Whether the intended item could be found and bought through the trial’s selling arrangement.

Positive system stock does not settle the last question. Physical units may need to be found, replenished or made available through the approved process. Equally, no recorded sale does not prove the shelf was empty.

Raman, DeHoratius and Ton’s 2001 retail-operations research distinguishes inaccurate inventory records from misplaced stock. The publisher’s accessible abstract supports that qualitative distinction; its study results are not current benchmarks for this trial.

An empty intended location alongside positive recorded stock calls for investigation. It does not identify the cause by itself.

Work through a fictional shelf-check example

Imagine a shop trialling a travel pouch in one specified size and colour. The intended location is observed stocked at opening. A later check finds it empty, while the system still reports units. Transactions were recorded between the checks.

Those facts establish the opening observation, the later empty-location observation, the recorded quantity at its stated time and the transactions. They do not establish when the location became empty or whether it remained stocked continuously beforehand.

The pouch may have needed replenishment, stock may have been misplaced, or records may need correction. These are questions to investigate through the retailer’s approved processes, not conclusions supplied by the discrepancy.

The buyer also cannot tell how many customers would have purchased during any unavailable interval. Nor can sales of another pouch identify which customers substituted for the trial item.

The immediate task is to understand the selling conditions, restore the intended arrangement where appropriate and decide what further evidence is useful. The example is fictional and contains no reported retailer result.

Treat a check as a check, not a duration

Build a short sequence of relevant events. For each, retain its time, source and what it establishes: a shelf observation, recorded transaction, stock record or documented replenishment.

Then identify the gaps. Seeing stock at opening and an empty location later does not reveal the exact time availability changed. A recorded transaction supports a recorded sale at that time, not continuous availability throughout the surrounding period.

Other suitable evidence may clarify an interval. If it does not, leave that interval uncertain. Do not fill the gap with a precise number of available hours merely because a spreadsheet needs a denominator.

Keep observations relevant to the decision. More frequent checks may improve the evidence but consume staff attention. Choose a workable arrangement around the trial question and operating conditions rather than inventing a standard checking frequency.

Calculate a descriptive rate only when the period is supported

A rate can help describe observed selling, provided the numerator and denominator concern the same documented interval.

For example, an explicitly defined measure could be units recorded sold during a documented available interval, divided by that interval’s duration. State the time unit, included trading periods, sales-unit basis and evidence supporting availability.

Before using it, check:

  • Are the included sales confined to the same interval as the available time?
  • Are partial days and different dayparts visible?
  • Does the availability evidence establish the interval, rather than just isolated snapshots?
  • Have uncertain periods remained separate?

Do not divide whole-trial sales by a denominator covering only selected available periods. If availability cannot be established adequately, show recorded sales and the observations without reporting a falsely precise rate.

Even a correctly matched rate describes those observed periods. Multiplying it across a stockout does not automatically estimate the sales missed there.

Vulcano, van Ryzin and Ratliff’s 2012 operations-research paper explains that unavailable-product demand and substitution are not directly revealed by sales transactions. Its formal estimation method uses explicit assumptions; it is not equivalent to scaling a simple in-stock rate.

Compare the conditions behind the rate

Available periods may differ in more than duration. Ask what customers encountered in each period:

  • The price and any promotion.
  • Daypart, trading hours and customer traffic where known.
  • The intended selling location and staff execution.
  • The availability of relevant alternatives.
  • Seasonal or other material changes affecting the comparison.

NIST’s experimental-design guidance explains that factors other than the main factor of interest can affect measured results. Deliberate comparison design addresses those influences; selecting available periods does not, by itself, control them.

A rate from a busy promoted period and one from quieter ordinary trading describe different conditions. Normalizing by time does not remove that difference. Similar stores and before-and-after figures are not automatically controlled experiments or causal estimates either.

State the comparison’s limits before using it to support another commitment.

Keep substitution a question to investigate

Review relevant alternatives and their availability alongside the trial item. Another product selling while the candidate is unavailable may be consistent with substitution, but the totals do not establish each customer’s alternative choice.

The reverse matters too. A candidate available while a familiar alternative is missing may receive purchases under conditions that would not continue when both are stocked.

Do not count the alternative’s sales as recovered demand for the candidate, or use them to manufacture an estimate of lost sales. Bring the findings back to the range question: what offer was being tested, and which conditions remain unresolved?

Choose the response that addresses the gap

A buying action and a demand conclusion are separate decisions. A bounded replenishment can provide further selling opportunity without declaring a permanent range winner.

Match the response to the finding:

  • Stock records and the intended location disagree: Investigate the discrepancy and correct execution through the approved process before issuing a product verdict.
  • A documented sellout followed limited selling opportunity: Consider whether separately authorized replenishment could answer the trial question, using actual supplier terms, storage and remaining-stock constraints.
  • Outage timing or execution is uncertain: Improve the availability evidence and retest rather than report a precise demand rate.
  • Supply cannot sustain the intended offer: Assess that operating constraint separately from whether customers wanted the product.

Further stock increases cash and remaining-stock exposure. Further checking uses staff time. Explain what the next commitment should reveal and why that information is worth pursuing.

Close the review with a plain statement: what sold, what availability was established, what remains unknown and which action addresses it. That is a stronger basis for a range decision than either “it sold out” or “sales were weak” on its own.

Sources