Reading Demand Signals for Inventory and Planning
Rank as a leading indicator for stock decisions, where first-party data is thin, and the specific ways this goes wrong at the moments it matters most.
For a seller, rank tracking of your own products adds little — you have the actual sales figures. Its value in planning is in the products you do not sell.
Where it helps
Category demand trend. A fixed basket's median rank over quarters shows whether the category is growing, contracting or flat, ahead of any published data.
Competitor supply problems. Their outage is your opportunity, and it is visible within a day.
Emerging products. A new entrant gaining traction indicates demand you may not be serving.
Seasonal timing. Multi-year basket data shows when a category's season actually starts, which is frequently earlier than the trade assumes.
Substitute behaviour. When a leading product goes out of stock, watching which competitors' ranks improve shows what buyers substitute toward. This is genuinely hard to learn any other way and it is available free to anyone tracking a basket.
Where it does not
Absolute demand. Estimation error is too wide for stock quantities.
Your own demand, which you already know better.
Lead time decisions, which need magnitude rather than direction.
Anything requiring a number to order against.
The substitution study
Worth describing because it is the most useful and least used application here.
Identify an outage in a leading product from your availability capture.
Observe which competitors' ranks improve during the outage window, relative to the basket.
The magnitude of their improvement indicates how much of the displaced demand they captured.
Repeat across several outages to build a picture of the substitution structure in the category.
This tells you who you actually compete with, which is frequently not who you assumed, and it is derived from natural experiments the market runs for free.
Seasonality for planning
Two or more years of basket data gives a seasonal profile for the category.
Compute the basket median by week across years. The shape is the season.
The onset date matters more than the peak for ordering decisions, and rank data shows it earlier than sales data shows it to you, because it aggregates the whole category.
Compare your own sales seasonality to the category's. A product peaking later than its category is losing the early demand to competitors, which is an actionable finding.
The confounders, again
Catalogue growth pushes basket medians down over years, which looks like category decline. Use a fixed basket and note that it ages.
Basket composition drift. Products leave the market; the basket shrinks or must be topped up, and topping it up changes what you are measuring.
Marketplace changes — category restructures, ranking changes — produce discontinuities across years.
Multi-year comparisons need care and should be presented with these caveats visible.
What to build
A fixed basket of twenty to fifty products in each category you care about.
Daily rank, price and availability.
A quarterly review of the basket median, the seasonal profile and any structural changes.
An outage log, so substitution studies are possible retrospectively.
This is a small, cheap, durable dataset that becomes more valuable every year and cannot be bought once you have missed the years.
The honest framing for planning
Rank data gives direction, timing and structure. It does not give quantities.
Used for the first three alongside first-party data for quantities, it is a genuine planning input. Used to generate order quantities, it produces numbers with an error range wider than the decision can absorb, and the first stockout or overstock will correctly be blamed on it.
The substitution study, step by step
A concrete procedure for the most useful natural experiment available in this data.
From the outage log, select an outage of a significant product lasting at least several days.
Define the candidate substitute set — products a buyer might switch to.
Compute each candidate's rank relative to the basket before, during and after the outage.
Identify which candidates improved during the outage beyond their normal range and returned afterwards.
The magnitude of improvement approximates the share of displaced demand each captured.
Repeat across five or more outages to distinguish a pattern from a coincidence.
The output is a substitution map for the category: who competes with whom, weighted. That is a genuinely valuable competitive artefact and it costs only the outage log and an afternoon.