Sena knows where a cut item sends its shoppers and what an unlisted item would take.
Assortment optimization decides which items earn a place in the range, how many, and in which outlets. Three properties of the data behind that decision push the range toward what already sold.
The ranking runs on the sales history of items the range carried. An item left off the range for three years contributes zero rows to the model that decides whether to list it.
A delisting removes a volume line. Whether those shoppers moved to another item in the range or left the category entirely is the difference between a saving and a loss, and the sales file records the same thing either way.
Item performance arrives from the retailers who report. Outlets outside that set carry different ranges, and their shoppers rank the same items differently.
Sena is the decision AI with access to real-world data. It records what shoppers bought across every item in the category, including the ones this range leaves off, and asks those shoppers what they would take if an item disappeared, so a cut is costed before it is taken.
Receipts cover the whole category, so an unlisted item arrives with the volume it already takes elsewhere.
Purchase and stated substitution together show whether a delisting keeps volume in the category.
Market and outlet type sit on every figure, so one ranking opens into the channels underneath.
Assortment optimization decides the item set a category carries. It answers four questions at once, and the four pull against each other: how wide the range runs, how deep each segment goes, which items get cut, and which outlets carry which version.
Assortment planning sizes and phases the buy against an open-to-buy budget, working in units and weeks of cover.
Assortment optimization decides which items earn a place in the range at all.
| Decision | What it returns | Where it goes wrong |
|---|---|---|
| Width | How many segments the range covers. | Widened on rival listings, over shopper demand. |
| Depth | How many items inside each segment. | Deepened where the reporting account sells. |
| Cuts | Which items leave. | Costed as a saving, with the trip unpriced. |
| Localisation | Which outlets carry which version. | Set by store size, over who shops there. |
Assortment optimization in retail carries a constraint the manufacturer side avoids: the shelf is finite, so every listing displaces another. The decision is comparative at all times. Three reads settle it.
Whether a listing brings volume the range would otherwise miss or moves volume between its own items.
Which items a shopper accepts when a first choice is absent, and at what point the trip is abandoned.
Which shopper groups an item brings in that the rest of the range holds zero access to.
The list rests on four collected inputs, gathered market by market.
| Input | What it answers |
|---|---|
| Receipts | What shoppers bought across the whole category, item by item, dated. |
| Store captures | Which items each outlet carried, and which it left off. |
| Geo-verified photos | Which items faced the shopper, placed and timed. |
| Stated preference | What a shopper takes when a first choice is absent. |
The team's own numbers join separately. Sales history, listing files, planograms, and margin records connect through 250+ integrations, so internal ranking meets external purchase.
Three questions live beyond the sales and listing file, and each changes the range.
Real people share what they buy and prefer, under explicit consent. Sena captures it directly at the source, so every figure traces back to where it came from whenever a number comes under question.
Purchase across the whole category and stated substitution reach Sena from the same shoppers, each figure tagged to its market and its week.
Receipts separate a listing that brings new volume from one that moves volume inside the range.
A basket with one category item and little else marks a shopper at risk of leaving.
Outlet-level purchase shows which range version each location needs.
Assortment optimization built on listed-item history can rank what the range already carries and struggles to price a cut. Sena captures real-world signals from the range photographed in a real outlet through to the receipt showing what the shopper took instead.
Purchase is recorded across the whole category, so an item outside this range arrives with the volume it already holds.
Reads which items an outlet carried, off images captured in real stores, so the range on paper meets the range in the market.
Signal arrives from the consumer network under explicit consent. What a shopper takes when a first choice is absent comes from that shopper.
Sales history, listing files, planograms, and margin records connect over 250+ integrations, so internal ranking meets external purchase.
Market and capture week sit on every figure, so a delisting decision opens onto the baskets it was argued from.
Listing files, range reviews, and planograms from prior cycles load in, as far back as the category has been scanned.
Four teams read the same range, and each one needs a different cut of it before they can act.
The range. Needs each cut costed against the trip it risks.
The listing decision. Needs incrementality separated from cannibalization.
The range review. Needs evidence the retailer accepts as neutral.
The consequence. Needs the range fixed before the buy is sized.
The same category-wide count, read against whatever each industry treats as a substitute.
Item ranking per outlet type, with each cut costed against the trip.
Own-label against branded lines, with the substituting shopper named.
Pack and format spread, read against the drinking occasion.
Tier coverage read against where a shopper trades up or leaves.
One mechanism, applied per market and per category. Each step is documented, which is what carries a delisting through a range review.
Receipts return what shoppers bought across the category, resolved to the outlet that sold it.
Sena prices each candidate cut against where its shoppers go, so a saving separates from a lost trip.
Stated substitution covers the items the range has yet to carry, so the ranking reaches past its own listings.
Most assortment optimization runs on the sales history of listed items in reporting outlets, which ranks what the range carries and holds zero rows for what it left off. Sena covers the category around it.
| Capability area | Typical setup | Sena |
|---|---|---|
| The dataset | Sales history for listed items. | Purchase across every item in the category. |
| A delisting | Volume removed from the line. | Where the volume goes, or the trip that leaves with it. |
| An unlisted item | Absent. | The volume it already takes elsewhere in the market. |
| Incrementality | Inferred from the total. | Read from the basket, item against item. |
| Substitution | Modelled from price distance. | Stated by shoppers, with the abandonment point. |
| Localization | Set by store size and format. | Set on who shops at each outlet, across 190+ countries. |
| Evidence in a review | A ranked list. | Open any rank onto the purchases behind it. |
Three range decisions where purchase from outside the file changes the answer.
Split the candidate cuts into the ones that keep volume in the category and the ones that lose the trip.
Read the volume a candidate item already takes across the market before it enters the range.
Capture what each outlet carries and who shops it, then set the range version per location.
A settled list sets what the category carries. The four decisions below are what a category team runs next.
Every category decision Sena answers, in one place
The space each item receives
Allocates the space each carried item receives at the shelf.
Explore shelf space optimization 02The balance across items
Balances volume across the items the listing decision carried.
Explore product mix optimization 03What the category contains
Maps what shoppers substitute, which is what a listing cut risks.
Explore product category analysis 04The measurement, every cycle
Measures what the listed range returned, cycle after cycle.
Explore category analyticsThe walkthrough takes one category in one market, prices a proposed set of cuts against where each item's shoppers go, and sizes one unlisted candidate while the team watches.
Name the category and the markets it sells in, and we will price the cuts on the table.