Competitive Pricing Bands: Setting the Right Price Floor and Ceiling

A SKU that has lost its anchor isn't a runaway — it's a band that has collapsed. The list price sits 7% under MAP. The competitive median has moved twice in a month and neither move was caught. The cart-price discount that was supposed to expire on Sunday is still running on Wednesday. Each of these is the same failure in different clothing: the merchant lost sight of the corridor the SKU is supposed to live inside, and the corridor quietly closed.

What a competitive pricing band actually is

A competitive pricing band is the corridor a SKU is allowed to live inside on any given week. Two rails define it:

1. The floor. The lowest price the merchant will accept this week. Built up from COGS plus shipping and a configured margin floor, then adjusted for vendor MAP and category expectations. Anything below the floor is a guardrail violation, not a candidate price.
2. The ceiling. The highest price the merchant can charge without losing the basket to a cheaper competitor. Approximated by the basket median of the headline competitors, then nudged by stock state and promo window.

The width of that corridor — the gap between floor and ceiling — is what tells the merchant whether the SKU has room to breathe (a 12% band) or whether the band has collapsed to nothing (a 2% band means a polite undercut is already unreachable). Bands are diagnostic first, prescriptive second: a wide band means a real recommendation is possible; a collapsed band means the merchant has a different problem — usually cost, MAP drift, or competitor crowding.

A band isn't a number. It's a corridor — a floor, a ceiling, and a width. Wide bands buy you room to undercut politely. Thin bands buy you a workflow problem, because any polite undercut has already crossed the margin floor.

Why bands collapse

Bands collapse for one of three reasons, and each one has its own diagnostic signature:

• Floor creep. COGS increases (freight, vendor surcharge) and the merchant doesn't propagate the new COGS into the band's floor calculation. The floor has been right; the inputs to it have been stale.
• MAP drift. A vendor announces a new MAP — usually a 4–8% downward adjustment to align with retail reality — and the merchant's PDP misses the change. The old floor is now below the new MAP, so anything above the MAP looks defensible but anything below it is a contractual violation.
• Ceiling drop. A competitor initiates an undercutting cycle and the merchant watches the basket median fall week over week. The ceiling collapses; the floor stays put; the corridor narrows.

None of these are loud. None of them appear on a single dashboard. All of them compound. That's why the band is worth defining as a first-class object — not a calculation buried inside a recommendation engine.

How to set a band's boundaries per SKU

Three inputs are non-negotiable; below them, the math is the same for every category:

1. Floor. floor = max(COGS + shipping, vendor MAP) × (1 + marginFloorPct). The max with MAP matters: a floor that ignores MAP is a future contract violation. The marginFloorPct is per-category, not per-merchant — an apparel floor and a beauty floor live in different worlds.
2. Ceiling. ceiling = competitor_basket_median × (1 + premiumPct). The premiumPct is small (typically 1–4%) — anything larger reads as gouging and the basket will move.
3. Promo adjustment. While a declared promo window is active, the effective floor drops by the promo depth. After the window, it restores automatically. An undeclared promo drag is the most common band collapse we observe — see the next section.

Per-SKU overrides exist (a SKU can have a tighter floor for a clearance event, a wider ceiling because it has reach the competitors don't) but the rule is: defaults first, overrides second, and every override expires on a date.

The three inputs that reset weekly

Bands decay. What was a healthy 9% corridor last Tuesday can be a 3% corridor this Tuesday. The three signals worth recomputing every week:

• Competitor basket median. Scraped on a cadence that matches the category — top-decile SKUs polled hourly, mid-tail twice daily, long-tail weekly. A weekly manual check misses the burst.
• Cart-vs-list drift. If the cart price disagrees with the list price outside of a declared promo window, the band is being silently eroded by a misconfigured coupon or a stale cart rule.
• FX-adjusted median for international listings. A USD-anchored price on a EUR listing wipes out the entire European margin band the moment the FX rate moves 5%.

Operating without losing the merchant

A bands dashboard that produces 200 collapsed-band reports a week is not an asset — it's a queue. Four habits keep it shipping productively:

• Triage by corridor width. Ignore SKUs with bands wider than 8% — they're fine. Investigate SKUs with bands narrower than 4% — collapsed or near-collapsed. Ship weekly digests with an exception lane for SKUs at zero width.
• Promo windows are explicit, never implicit. Every band respects a declared promo event; the band recalculates at window start and restores at window end. A discount that "just stays" is a band-killer.
• Reorder reviews the band against current data, not memo. Buyer's memos anchor on the prior selling price; bands anchor on current competitor median, current MAP, current COGS. Every reorder should generate a new band, not inherit the prior one.
• Realized vs recommended gets reviewed monthly. A persistent gap between the band's ceiling and the price that actually shipped is a workflow problem (CMS override, buyer's memo, manual promo), not a model problem. Find the override.

Closing the loop with the rest of the cluster

Competitive pricing bands are the diagnostic that sits between observation and action. The other posts in this sequence cover the upstream input (the monitor that feeds the ceiling) and the downstream behavior (the margin-aware recommendation that defends the floor).

For the inputs to a band's ceiling — what to scrape and how often — see our competitor monitoring guide. For the diagnostic queries that catch band-collapse patterns — stale MAP, forgotten promos, cost-plus anchoring — see 7 pricing mistakes. For an end-to-end worked example of bands applied to recommendation output, try the pricing report tool, or compare the tier with scheduled monitoring on the pricing page.