What we repeatedly see in ecommerce trading reports is a conversion-rate decline blamed on traffic or merchandising even when the products receiving demand could not be purchased. Sessions remain in the denominator, product views remain in the funnel, and revenue disappears—but stock availability is reviewed in a separate operations dashboard.
Availability-adjusted conversion brings those views together. It does not erase stockouts from performance. It shows how much demand faced a genuinely purchasable assortment, how shoppers substituted when the preferred option was unavailable, and where inventory constraints—not weak intent—limited commercial output.
Table of contents
- Keyword decision and search intent
- Why the standard conversion rate can mislead
- Define purchasable exposure
- Build the availability scorecard
- Estimate constrained demand carefully
- Measure substitution and recovery
- Use cohorts that can mature
- Turn the analysis into decisions
- Common analytical mistakes
- EcomToolkit point of view
Keyword decision and search intent
- Primary keyword: ecommerce inventory analytics
- Secondary keywords: availability-adjusted conversion rate, stockout analytics, lost sales from out of stock, ecommerce product availability KPI
- Search intent: informational-commercial
- Funnel stage: mid-funnel
- Page type: measurement framework and decision guide
- Why this angle can win: inventory articles often focus on replenishment, while conversion guides rarely correct for whether the demanded SKU and variant were actually purchasable.
Why the standard conversion rate can mislead
A 2025 peer-reviewed study on grocery stockouts and seller performance in Amazon marketplaces reported strong negative relationships between stockout rates and sales performance across the countries examined. The exact effect should not be transplanted into an unrelated store, but the mechanism is clear: availability constrains what demand can become.
The standard ecommerce conversion formula is:
orders / sessions
It remains useful, but it mixes at least three states:
- the shopper saw the intended product and could buy it;
- the shopper saw the product but the desired variant was unavailable;
- the shopper never saw an appropriate product because the assortment or ranking did not expose one.
When the mix changes, the headline conversion rate moves even if customer intent and page quality do not.
Do not replace standard conversion. Add an availability lens so the trading team can distinguish traffic quality, experience friction, merchandising gaps, and supply constraint.
Define purchasable exposure
Availability is not a daily product flag. It is a customer-level state at the moment of demand.
Capture:
| Field | Purpose |
|---|---|
| product and variant ID | identifies the demanded unit |
| availability at impression | tells whether the card represented a buyable option |
| availability at PDP view | captures the decision state |
| selected variant availability | distinguishes default from intended choice |
| sellable quantity band | separates low-stock risk without exposing exact units |
| location or fulfilment eligibility | explains geographic constraint |
| backorder/preorder state | separates delayed purchase from true stockout |
| restock expectation | supports waitlist and promise analysis |
| availability reason | out of stock, restricted, discontinued, data error |
| price and promotion state | controls for concurrent demand changes |
A product can be globally in stock and unavailable to a specific destination. A size can be unavailable while the parent product remains buyable. A bundle can fail because one component is missing. Model availability at the lowest commercially relevant grain.
Define an eligible demand event such as a product-detail view, variant selection, internal-search click, or product-card impression. The best denominator depends on the decision:
- impressions support ranking and assortment questions;
- PDP views support product-level conversion;
- variant selections expose size or colour demand;
- add-to-cart attempts expose the strongest blocked intent.
Build the availability scorecard
| Metric | Formula | Interpretation |
|---|---|---|
| Buyable impression rate | buyable product impressions / all product impressions | how much discovery inventory could satisfy |
| PDP availability rate | buyable PDP views / eligible PDP views | product-level exposure quality |
| Variant availability rate | available selections / all variant selections | depth of size/colour coverage |
| Stockout attempt rate | blocked add-to-cart attempts / all attempts | high-intent lost opportunity |
| Availability-adjusted CVR | orders / sessions with meaningful buyable exposure | experience among serviceable demand |
| Standard CVR | orders / all sessions | complete commercial outcome |
| Substitution rate | buyers of alternative / shoppers blocked on preferred item | recovery quality |
| Waitlist capture rate | back-in-stock signups / eligible stockout views | retained future demand |
| Restock recovery rate | purchasers after restock / reachable waitlist cohort | realised delayed demand |
Report standard and adjusted conversion together. The gap is diagnostic, not a result to celebrate. A high adjusted rate with a low standard rate can mean the site converts serviceable demand well while failing to offer enough serviceable inventory.
Weight availability by demand. “90% of SKUs in stock” can look strong even when the missing 10% receives half the customer interest. Use impressions, selections, forecast demand, or recent sales to create demand-weighted availability.
Estimate constrained demand carefully
“Lost sales” is not directly observed. A shopper blocked on one item may substitute, delay, buy elsewhere, or decide not to buy at all. Multiplying out-of-stock views by the normal product conversion rate will usually overstate certainty.
Use a range:
| Method | Strength | Limitation |
|---|---|---|
| Historical same-SKU baseline | intuitive and product-specific | seasonality and campaigns confound |
| Matched available periods | controls for product traits | matching can remain imperfect |
| Variant sibling comparison | useful for size/colour gaps | preferences differ |
| Geographic availability comparison | natural operational variation | market mix differs |
| Waitlist-to-purchase observation | measures delayed recovery | captures only identifiable interest |
| Controlled holdout or phased restock | stronger causal evidence | operationally difficult |
Label the result as constrained-demand estimate, not booked revenue. Publish low, central, and high cases with the assumptions. Margin, returns, discount depth, and fulfilment cost belong in the value calculation; gross demand is not contribution.
For a cash-aware companion model, use the cohort payback and inventory cash framework.
Measure substitution and recovery
Stockout analytics is incomplete if it ends at the missing product.
Create a stockout journey:
- customer expresses demand for unavailable product or variant;
- site exposes an alternative, waitlist, preorder, or store-availability route;
- customer views or selects the recovery option;
- customer purchases now, purchases after restock, or exits;
- later return and repeat behaviour mature.
Measure substitution at product-family, category, price-band, size, and margin similarity. A substitution that protects revenue but sends the shopper to a lower-margin, higher-return item may not be commercially equivalent.
Recommendation quality also matters. Do not count an alternative as “shown” merely because a carousel request fired. Track whether it entered the viewport and whether the substitute was actually available in the relevant variant.
Back-in-stock programmes need cohort logic. Preserve signup date, product, variant, consent state, restock date, message date, and purchase window. The back-in-stock analytics guide provides a fuller recovery model.
Use cohorts that can mature
Inventory outcomes arrive at different speeds. A blocked add-to-cart is immediate; a restock purchase may occur weeks later; return and repeat effects take longer.
Set windows before analysis:
- same-session substitution;
- seven-day cross-product purchase;
- 14- or 30-day restock recovery;
- category-specific return maturity;
- 60- or 90-day repeat behaviour where appropriate.
Do not compare a fully matured available cohort with a recent stockout cohort. Mark incomplete periods and restate them as data matures.
Customer identity is also partial. Anonymous shoppers may return on another device or channel. Report observed recovery as a lower-bound measure unless identity resolution is robust and privacy-compliant.
Segment by new versus returning customer. Loyal customers may wait; new customers may leave. Segment by acquisition cost as well: paying to send high-intent traffic toward unavailable hero products can turn a supply problem into a media-efficiency problem.
Turn the analysis into decisions
Merchandising
- demote products with no buyable variants where appropriate;
- expose relevant available alternatives;
- separate discontinued items from temporary stockouts;
- preserve SEO value without creating a dead shopping route;
- show accurate restock or preorder information.
Buying and planning
- rank stockouts by demand-weighted contribution opportunity;
- identify chronic size and colour gaps;
- compare forecast error with campaign and ranking changes;
- account for supplier lead-time reliability.
Marketing
- suppress or redirect ads for unavailable destinations or variants;
- align campaign landing pages with serviceable inventory;
- use margin and availability together when allocating spend.
Product and engineering
- reduce inventory-data lag;
- monitor cache invalidation and oversell risk;
- ensure search, recommendations, PDP, cart, and checkout share a consistent state;
- log why an item became unavailable.
The highest-priority action is often not “buy more.” It may be correcting a feed, changing rank, improving substitution, or stopping paid traffic until supply recovers.
Common analytical mistakes
Using end-of-day stock
A daily snapshot cannot prove what the shopper saw. Capture availability at the demand event.
Counting parent products
A fashion PDP can be “in stock” while the most demanded sizes are unavailable. Use variant grain.
Removing stockout sessions from the business result
Adjusted conversion is diagnostic. Standard conversion remains the outcome.
Claiming every blocked view was a lost order
Demand is not purchase certainty. Use ranges and observed recovery.
Ignoring margin and returns
Substitution revenue can be lower-quality revenue. Mature the cohort.
Mixing data errors with real shortages
An item may be unavailable because the feed is late, the location rule is wrong, or the inventory is truly zero. These require different owners.
EcomToolkit point of view
Conversion rate should not absolve inventory, and inventory should not excuse the customer experience. The useful analysis keeps both truths visible: how the store converted demand it could serve and how much demand it made impossible to serve.
EcomToolkit recommends making demand-weighted availability a first-class trading KPI, measured at impression, PDP, variant, and cart attempt. Pair it with the assortment productivity framework to balance demand capture with cash discipline, or contact EcomToolkit for an analytics audit that reconciles storefront behaviour with stock reality.
