Every retail brand I've worked with over the past 17 years has had some version of the same problem: the OTB process was built to answer the question 'how much can we buy?' when the real question it should be answering is 'how much should we buy, and when?' That distinction — subtle as it sounds — is the difference between a planning function that protects margin and one that quietly erodes it.
The Root Cause: OTB Built on Last Year's Receipts
The most common OTB failure mode I encounter is a model that anchors to prior-year receipt flow rather than forward-looking sell-through velocity. The logic feels intuitive: last year's receipts funded last year's sales, so this year's receipts should fund this year's plan. But this reasoning ignores the most important variable in the equation — how fast the product is actually selling.
When sell-through velocity slows — due to a trend shift, a competitor's promotion, a marketing miss, or simply a bad buy — a receipt-anchored OTB model keeps the buying engine running at full speed. Receipts arrive on schedule. Inventory builds. And by the time the Planning team surfaces the problem, the markdown liability is already baked in.
"The OTB model should be a live instrument, not a spreadsheet you reconcile once a month. By the time you reconcile, the damage is done."
Peng Huang
The Fix: Velocity-Adjusted Receipt Pacing
The solution is to rebuild the OTB model around a single governing principle: receipts should be paced to sell-through, not to plan. This means the model needs to ingest weekly sell-through data at the class or style level, compare it against the original plan assumption, and automatically adjust the forward receipt schedule to reflect the new reality.
In practice, this looks like a dynamic receipt-pacing layer that sits on top of the traditional OTB. For each major category, the model calculates a sell-through index (actual ÷ planned) on a rolling 4-week basis. If the index falls below a defined threshold — say, 0.85 — the model flags the category for receipt deferral and surfaces the markdown exposure if no action is taken.
- 01Ingest weekly sell-through at class or style level — not just total department
- 02Calculate a rolling 4-week sell-through index (actual ÷ planned)
- 03Set threshold triggers for automatic receipt deferral flags (e.g., index < 0.85)
- 04Model the markdown liability of inaction alongside the receipt deferral option
- 05Surface the reconciled view to merchants and planners in a shared dashboard
The Organizational Challenge
The technical fix is the easy part. The harder problem is organizational. In most retail structures, the merchant owns the buy and the planner owns the OTB. When the OTB model flags a receipt deferral, the merchant interprets it as a challenge to their buy decision. The conversation becomes adversarial instead of collaborative.
The brands that solve this problem don't just build better models — they redesign the operating rhythm around the model. Weekly OTB reviews become a shared forum where merchants and planners look at the same data and make joint decisions. The model doesn't make the call; it surfaces the trade-off. The team makes the call.
"The best OTB process I've ever seen wasn't the most sophisticated model. It was the one where the merchant and the planner were looking at the same screen and speaking the same language."
Peng Huang
What Good Looks Like
A well-functioning OTB process has three characteristics: it is forward-looking (driven by sell-through velocity, not prior-year receipts), it is shared (merchants and planners operate from a single source of truth), and it is actionable (it surfaces specific decisions, not just data). When all three are present, OTB stops being a constraint and starts being a competitive advantage.
In one recent engagement, rebuilding the OTB model around these principles reduced markdown liability by 31% in a single season — not by buying less, but by buying smarter. The total receipt budget was nearly identical. The difference was timing, velocity sensitivity, and a shared operating rhythm that caught the problems six weeks earlier than the old model would have.