A supermarket interior with stocked displays and open aisles.

Stockouts: why retail media results need availability context

An ad cannot convert demand into a sale when the advertised product is unavailable.

A person beside illuminated product displays in a futuristic retail illustration.

That sounds obvious, yet campaign reports often place impressions, clicks and sales beside one another without showing whether the product could be bought in the relevant store, channel and hour. The result can make strong media look weak, weak media look efficient, or substitution look like brand growth.

Availability belongs in two places: campaign eligibility before delivery and measurement context after delivery.

Availability is not one binary field

"In stock" can mean several things:

  • inventory exists somewhere in the network;
  • inventory exists in the shopper's selected store;
  • inventory exists but is not sellable;
  • the product is on the shelf but not recorded accurately;
  • the product is orderable but cannot meet the promised fulfillment window;
  • a substitute is available but the advertised variant is not.

Google Merchant Center requires availability data to match the landing page and checkout experience.[1] Its local inventory specification connects product identifiers, store codes and store-level availability.[2] Retail media needs the same basic discipline: the availability signal must match the unit at which the shopper can act.

A national "available" flag is not enough for a store-level campaign.

How stockouts distort media results

They suppress conversion

A shopper can receive the message, show interest and fail to purchase because the item is absent. If availability is omitted, the report assigns the failure to media or creative.

They change the eligible audience

A campaign that targets shoppers near stores with no inventory is spending against impossible conversions. Delivery should be suppressed or redirected when current availability falls below the campaign rule.

They create substitution

Shoppers may switch brand, pack size, flavor, retailer or purchase time. Research on consumer response to stockouts shows that the outcome varies with commitment to the unavailable option and the difficulty of choosing an alternative.[3]

A substitute sale may help the retailer but hurt the advertised brand. A delayed sale may look like no response inside a short attribution window.

They bias store comparisons

Treatment stores with better availability can outperform control stores even if the media has no effect. The reverse can also happen. Store matching and experiments need availability balance or adjustment.

They can corrupt audience models

One model-risk scenario occurs when nonpurchase becomes a negative label without an availability feature or censoring rule. In that case, a shopper who could not buy during a stockout may be labeled low propensity or lapsed, and the model may learn supply failure as weak demand.

They distort basket patterns

A missing item changes what appears with the rest of the basket. Complement attachment can fall, substitute relationships can spike, and mission-completion analysis can point to the wrong gap. That is why basket influence analysis and category-affinity interpretation need availability checks.

The availability data a campaign needs

The minimum useful table links:

  • timestamp or defined availability interval;
  • store or fulfillment location;
  • product identifier at the advertised level;
  • sellable availability state;
  • quantity or confidence where available;
  • price and promotion state;
  • source system and update time.

The exact fields depend on the retailer. The principle is to align availability with the campaign's exposure and outcome grain.

If media is delivered hourly by store and SKU, a weekly regional availability rate is too coarse. If measurement is weekly by category, second-by-second inventory may add cost without changing the decision.

A worked example: a four-week beverage campaign

Consider a hypothetical four-week beverage campaign across 100 stores, using in-store screens and onsite placements.

The topline report shows weaker conversion in the second week. Without availability, the team might change creative or reduce spend.

The store-SKU view shows that a distribution delay reduced sellable availability in 28 treatment stores. Exposed shoppers bought a competing flavor and, in some cases, another brand. In this illustration, the "expected conversion pattern" means the predeclared range predicted from matched pre-campaign weeks and control stores, not a target invented after results arrived.

The correct actions are different:

  1. Exclude or separately report periods where the promoted SKU was unavailable.
  2. Measure substitution and total category effect, not only same-SKU sales.
  3. Suppress future media where availability fails the delivery rule.
  4. Investigate whether the campaign accelerated depletion in high-response stores.
  5. Do not claim the stable-store result for the entire network.

Availability does not become an excuse to remove inconvenient data. The report should show the full result and the availability-conditioned analysis.

Delivery rules should fail closed

Before launch, define when media should stop or change.

Examples include:

  • suppress the SKU when local availability is stale or below threshold;
  • switch to an approved substitute or category message when the original item is unavailable;
  • remove a pickup promise when the fulfillment window cannot be met;
  • cap spend in stores with repeated inventory-feed failures;
  • prevent dynamic creative from displaying an unavailable pack or price.

The rule must specify the fallback. Silent delivery against stale inventory wastes spend and damages shopper trust.

How to include availability in measurement

The IAB/MRC guidelines emphasize transparent data sources, methods and outcome definitions.[4] An availability-aware report should include:

Availability rate. Sellable eligible store-SKU-time cells / all eligible store-SKU-time cells with a known state. Report unknown cells separately rather than treating them as available. State whether cells are equally weighted or weighted by planned opportunity.

Media delivered while available. Actual impressions delivered in sellable eligible cells / all actual impressions delivered in eligible cells. A planned-delivery version may be shown separately, but should not replace actual delivery.

Lost delivery or suppressed delivery. Media not served because the product failed the rule.

Conversion conditional on availability. Useful for diagnosis, but not a replacement for the full campaign result.

Substitution. Same-brand variant, competitor, category and retailer effects where measurable.

Depletion. Whether media response increased the chance of a later stockout.

Incrementality. The treatment-control effect with availability handled according to the predeclared design and estimand.

Availability can enter the analysis in four different ways. Pre-treatment availability supports balance and eligibility. Time-varying availability can describe whether the shopper could act at exposure. Campaign-caused depletion is part of the treatment pathway. Post-result exclusions condition on something that may have been changed by treatment and can create selection bias.[5]

Do not simply delete stockout stores after seeing the result. Define the estimand first. An intention-to-treat estimate asks what happened under the assigned campaign, including operational failures and depletion. A separately labeled availability-conditioned analysis asks what happened during sellable periods, but it does not replace the full result. Predefine both, report sensitivity analyses and connect the method to the wider holdout and incrementality design.

Responsibilities across teams

Availability-aware media needs shared ownership.

The retailer inventory team maintains the operational signal. The media platform handles delivery suppression and logging. The brand or supplier manages replenishment commitments where applicable. The measurement team defines the analysis rule. The campaign lead decides what happens when the product cannot support spend. Observed pathing or zone-traffic evidence can explain whether shoppers reached the relevant zone, while closed-loop measurement keeps delivery, response, attributed outcomes and incrementality separate.

Common mistakes

Using distribution as availability. A SKU authorized for a store can still be absent.

Using today's inventory for yesterday's exposure. Measurement needs historical availability aligned to media time.

Treating all stockouts as external. Successful media can accelerate depletion.

Removing stockout periods after results arrive. The exclusion can create selection bias.

Ignoring substitution. A shopper may buy another variant or brand rather than abandon the category.

Serving media on stale data. A feed with unclear freshness is not a reliable eligibility control.

Buyer checklist

Before approving the plan or report, ask:

  • What does "available" mean operationally?
  • Is the signal aligned to store, SKU, channel and time?
  • How fresh is it, and what happens when it is stale?
  • Does delivery suppress or change when the item is unavailable?
  • Is the fallback creative approved?
  • How are substitutions and delayed purchases measured?
  • Could media itself have caused depletion?
  • Was the availability method defined before results were seen?
  • Are full and availability-conditioned results both shown?
  • Who is responsible for the corrective action?

FAQ

Should impressions during a stockout be removed from reporting?

Not automatically. Report them as delivered media and disclose the availability failure. Any exclusion from effectiveness analysis should follow a predefined method.

Is an out-of-stock always a lost sale?

No. The shopper may substitute, delay, change channel or leave the retailer. The outcome depends on commitment and available alternatives.

Can media cause stockouts?

Yes. A successful campaign can accelerate demand beyond replenishment. That operational effect is part of the result.

What availability level should trigger suppression?

The retailer should define a rule based on data quality, expected demand, fulfillment promise and campaign risk. There is no universal threshold.

Who should own availability in a retail media report?

Measurement should report it, but inventory, platform and campaign owners must agree on the source, delivery rule and response.

Bottom line

Availability is both a media eligibility rule and a measurement variable.

Align it to the store, product, channel and time of exposure. Suppress delivery when the shopper cannot act. Report substitution and depletion. Keep the full campaign result visible, and predefine how availability enters the causal analysis.

Without that context, a sales line cannot tell the team whether media failed, supply failed or shoppers simply chose another path.

Sources

[1] Google Merchant Center, Availability attribute.

[2] Google Merchant Center, Local inventory data specification.

[3] Gavan J. Fitzsimons, Consumer Response to Stockouts.

[4] IAB and Media Rating Council, Retail Media Measurement Guidelines.

[5] Hernán and Robins, Causal Inference: What If, guidance on post-treatment variables and selection bias.

Ready to see how this works in practice?

Footprints AI helps brands and retailers measure what matters. See our customer success stories or get in touch to discuss your retail media strategy.

More Stories

By clicking “Accept All”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.