The biggest door
is not the best door.
Not for this drop.
Allocation by conversion, not store rank — so a limited release sells out where it was always going to sell.
Illustrative · not a client store
The AI that
Drop sell-out rate, brand-partner sell-through and full-price ratio — on one board, narrated by Iris.
Where it clears. How many.
What the contract owes you.
Three decisions taken before a launch window opens — and all three are usually taken on store rank, gut feel and a spreadsheet.
Allocation by conversion, not store rank — so a limited release sells out where it was always going to sell.
Iris knows which doors sell out and which sit.
Pre-registration, traffic and prior drops turned into a buy quantity before the launch window opens.
Iris reads the queue before it forms.
Sell-through claims and co-op spend checked against contract terms — automatically, every cycle.
Iris reads the contract so nobody else has to.
Every door
on one board.
Rules fire. Models predict.
Iris explains.
| Rule | Threshold | Now | State |
|---|---|---|---|
| Gross margin · drop season | ≥ 43% | 46.2% | clear |
| Member mix of sales | ≥ 57% | 61% | clear |
| Drop sell-out rate | ≥ 85% | 89.3% | clear |
| Release demand accuracy | ≥ 89% | 93.1% | clear |
| Model | Predicts | Accuracy |
|---|---|---|
| Release demand | drop sell-out | 93.6% |
| Allocation fit | conversion by door | 92.0% |
| Re-order cycle | replacement timing | 89.8% |
This is the sector where all four rules read clear and the business still has a problem. Every threshold is green; the allocation model is the layer that can see two doors holding 41% of the units on 18% of the conversion. Iris then converts that into the only sentence that matters on a Friday — move 140 pairs, and the drop clears inside forty-eight hours.
Pick one. It opens
the real screen.
Drop sell-out rate, brand-partner sell-through and full-price ratio.
Point of Sale
On a drop Friday the till is the launch. A queue, a per-customer limit, a membership check and a stock pool shared with the app — all of it resolved in the seconds a customer is at the counter.
The member is recognised, not asked. With 61% of sales going to members, membership is the default path through the till rather than a question at the end of it.
Launch rules are enforced at the counter. Per-customer limits and release timing apply as rules, so a launch is governed rather than policed.
Iris knows what completes the kit. The upsell is the technical layer that belongs with what is already in the basket, not a generic accessory prompt.
Balanced Scorecard
Drop-season margin, member mix, sell-out rate and release demand accuracy on one live board — with Iris naming the doors that are holding units they will not convert.
Margin is read against full-price ratio. 46.2% at 82% full price is a different business from the same margin bought with markdown, and the board shows both together.
Every rule green is not the same as nothing wrong. This board is entirely clear while a drop is mis-allocated — which is exactly why the model layer exists underneath it.
The fourth lens audits the AI. Release demand accuracy sits beside margin, because a buy quantity is only worth what the model behind it is worth.
Numerus · CFO
Trial balance, P&L, balance sheet and cash flow on a ledger that ties out to the penny — with brand-partner rebates carried as the receivable they are rather than as a hoped-for credit note.
A rebate is a receivable, not a surprise. Sell-through claims and co-op spend are accrued against contract terms as they are earned, not discovered at the end of a cycle.
Drop season is a working-capital shape. A limited release is paid for long before it trades and cleared in days, and the ledger shows that rhythm rather than a monthly average of it.
Ask Iris for any number. Any door, any drop, any partner, any period — answered in the conversation rather than a week after close.
Nexus · Supply Chain
A drop is a supply chain problem with a publicised start time. The quantity is fixed before the window opens, and after that the only lever left is where the units are standing.
The buy quantity is modelled, not negotiated. Pre-registration, traffic and prior drops give a number before the launch window opens, which is the last moment it can still change.
Allocation is the second buy. Once the quantity is locked, putting 900 units in the wrong 92 doors costs exactly as much as buying the wrong quantity.
Inter-door movement is costed, not assumed. Iris prices a 140-pair reallocation against leaving it and marking down, so the move is chosen on numbers.
Motus · Operations
Launch execution across 92 doors, run to a standard rather than to whoever opened the store that morning — because a 97% launch SLA is an operating discipline, not a description.
A launch is a checklist with a clock on it. Floor set, queue plan, stock staged and till rules live — verified before the window, not reported after it.
Launch SLA is measured per door. A group number of 97% hides which three doors were late, and the three doors are the only actionable part of it.
Iris watches the exception, not the routine. The doors that opened to standard stay quiet; the ones that did not are named.
Manus · Workforce
The roster built against the drop calendar — because a launch morning at a streetwear door and an ordinary Tuesday at the same door are not the same shift.
Cover follows the drop, not the week. The doors that will clear their allocation in hours are the doors that need the staffing, and they are known before Friday.
Technical selling is a skill on the roster. Fit and product knowledge is what converts a compression base layer, and the schedule can treat it as a named capability.
Hours, payroll and compliance on one record. Who worked, where, at what rate — without a second spreadsheet running alongside the season.
Every solution,
on the real screen.
Grouped by what they act on. Pick one and it runs.
How Iris enforces the rules.
Thresholds that fire on the shift the breach happens (the “Zen Rules” layer) — the fit guidance, the club order and the partner contract. Three solutions run on this layer.
16.8% against a 14% target, and 58% of those returns cite fit. Compression base layer is the worst line on the floor at 31% — and it is the one silhouette with no size guidance at all.
Technical size & fit guidance
| Silhouette | Returned | Fit cited | Guidance | State |
|---|---|---|---|---|
| Compression base layer | 31% | 78% | no | no guidance |
| Technical shell | 19% | 54% | no | no guidance |
| Studio legging | 12% | 41% | yes | guided |
| Club kit | 8% | 22% | yes | guided |
| Other lines | 15% | — | partial | partial |
The evidence is already in your own data — silhouettes with fit guidance return at 10.1%, those without at 24.6%. Compression is the worst line you sell and has no guidance at all. This is not a modelling problem; it is two silhouettes missing a rule.
Team & club bulk orders
A club order is forty people, one deadline and a size list that arrives incomplete. It is not a large retail sale — it is a small manufacturing job that happens to be taken at a counter.
The order is a roster, not a basket — names, sizes, numbers and personalisation held per member rather than as a quantity by size.
Missing sizes are the default state of a club order, so the queue tracks what is outstanding instead of waiting for a complete list to start.
Personalisation and decoration are stages with their own lead times, and the promised date is quoted against them rather than against stock availability.
Club stock is ring-fenced from the retail pool, because a drop Friday must not consume the kit a team has already paid a deposit on.
Re-orders through the season are the point of the relationship, and the roster is what makes the second order take minutes rather than weeks.
Brand-partner rebate reconciliation
Sell-through claims and co-op spend checked against contract terms — automatically, every cycle. In a multi-brand sportswear business this is real money that is routinely left on the table because checking it by hand costs more than the error.
Contract terms are held as rules — tiers, thresholds, qualifying lines and claim windows — rather than as a PDF somebody remembers.
Sell-through is claimed from the same ledger that recorded the sale, so the number in the claim and the number in the accounts are one number.
Co-op marketing spend is matched to the activity it was granted for, which is the part most often unclaimed because nobody can evidence it later.
Every cycle is reconciled automatically, so a shortfall is a query raised inside the window rather than a write-off discovered after it.
The rebate is carried on the balance sheet as it is earned, which is what makes drop-season margin at 46.2% a real number rather than a provisional one.
How Iris predicts the drop.
What the buy should have been (the “Zen Models” layer) — modelled early enough to act on, with the cause named and the move costed. Three solutions run on this layer.
Drop & launch allocation
| Door | Allocated | Conv. | Will clear | State |
|---|---|---|---|---|
| Westfield White City | 180 | 0.9% | 62% | over-allocated |
| Trafford Centre | 160 | 1.1% | 71% | over-allocated |
| Shoreditch | 40 | 4.2% | 100% | under-fed |
| Camden Lock | 40 | 3.8% | 100% | under-fed |
| Other 88 doors | 480 | 1.8% | 88% | sized |
The drop is allocated by store size, not by who buys this product. Westfield and Trafford hold 41% of the units on 18% of the conversion, while Shoreditch and Camden will clear everything they are given in hours. Moving 140 pairs to the four streetwear doors lifts sell-out to 89% inside 48 hours.
Limited-release demand modelling
Pre-registration, traffic and prior drops turned into a buy quantity before the launch window opens. A limited release has one number to get right and no second chance to get it right.
Demand is modelled at 93.6% accuracy from signals that exist before the launch — registrations, traffic and the shape of comparable prior drops.
Under-buying a release and over-buying it fail differently. One costs the margin you never made; the other costs the margin you had and marked down.
The queue is a forecast input, not a marketing photograph — pre-registration is the earliest honest read on how many people intend to buy.
Scarcity is a deliberate setting rather than an accident of buying, which is the difference between a sell-out and a stock-out.
Every drop scores itself against its own forecast, which is why release demand accuracy is reported on the board at 93.1% and improving.
Training-cycle re-order prediction
Performance product wears out on a schedule set by how hard it is used. A runner covering sixty miles a week replaces shoes on a calendar; the retailer who knows that calendar owns the re-order.
Replacement timing is modelled per customer at 89.8% accuracy, from what they bought and how they train — not from a fixed number of months.
Re-order rate at 34% is the number this model exists to move, and it is reported on the board beside member mix rather than buried in CRM.
Member mix at 61% is what makes the model possible at all — a known customer has a training history; an anonymous basket does not.
The prompt lands in the window before the shoe is worn out, which is the only window in which the sale is still available to you.
Predicted re-orders feed the buy as base demand, so replacement volume is planned rather than left to arrive on its own.
Ask Iris · the drop
Not a screenshot — the actual agent, reading the same drop board the rules and models write to.
Twelve core POS solutions,
in every sector.
The sector solutions sit on top of these. They are not an upsell and they are not configured per customer — every Zentallio retail deployment ships with all twelve.
Implementation,
run by Iris.
You configure a sector playbook, not a custom project. Iris applies it herself — agentically, from day one.
Onboarding is agentic. Iris connects the door list, the drop calendar and the brand-partner contracts directly — no manual data mapping.
Prior drops are imported, not retyped. Historic release performance by door carries in, which is what makes allocation useful on the first Friday rather than the tenth.
Live in weeks, learning from day one. Go-live is a configuration, not a project plan — and the next drop date is already published.
Support starts
with Iris.
The first line of support is agentic — Iris resolves most of it herself. Our engineers pick up from there.
Layer 1 — Iris, 24/7. Configuration questions, anomalies and routine issues resolved directly, instantly — including through a launch window.
Layer 2 — our engineers. Anything Iris can't close escalates automatically to a Zentallio engineer.
No blank tickets. Every escalation arrives with Iris's own diagnosis — engineers start from an answer.
The drop, cleared.
where they convert
every rule is green
named as a rule
inside the window
Where a sell-out
quietly becomes a markdown.
A drop allocated by store rank, so the two biggest doors take 41% of the units and convert worst on the product.
A 16.8% return rate treated as a returns problem, when 58% of it cites fit and two silhouettes simply have no guidance.
Rebate money left unclaimed because reconciling the contract by hand costs more than the shortfall it would find.
Fit guidance published as a rule per silhouette, club stock ring-fenced from retail, and contract terms enforced every cycle.
Release demand at 93.6%, conversion by door at 92.0%, replacement timing at 89.8% — so the cause is named, not guessed.
“Why, and what do I do?” — move 140 pairs, and the drop clears inside forty-eight hours.
The same spine,
a different floor.
Sportswear & Activewear is sector four of nine in Fashion Retail. The layers, the six products and Iris herself are the same ones running across ten Food & Beverage sectors.
on your next drop.