Sell-Through Curves: The Weekly Signal That Tells You When to Markdown

Markdown by the calendar is wrong

Most retailers markdown on a schedule. Week 6 of the season, 25% off. Week 10, 40% off. Terminal clearance, 60% off. The calendar is predictable, the schedule is operationally clean, and the math is wrong for most of the assortment.

Sell-through curves vary by SKU, by store, by season. A SKU running 8% sell-through per week is going to land at 80%+ terminal sell-through with no help. A SKU running 3% per week is going to terminal-out at 30%, and the calendar markdown at week 6 is too late. The first SKU got an unnecessary margin cut; the second SKU needed earlier intervention.

Calendar markdowns are designed for operational simplicity, not margin optimization. The retailers that switched to curve-driven markdowns recovered 80-160 basis points of season margin, on average.

The weekly sell-through curve

Sell-through rate is simple: units sold divided by units received, expressed as a cumulative percentage of the season. Plotted weekly, it produces a curve. The slope of the curve in weeks 1-3 is highly predictive of terminal sell-through.

Empirically, across thousands of seasonal SKUs at mid-market apparel and specialty retailers, weekly ST trajectory in weeks 1-3 predicts terminal ST at week 12-14 with 75-85% accuracy. The retailers that ignore this signal spend the rest of the season reacting to lagged data and burning margin on calendar-based discounts.

A healthy 12-week seasonal curve looks like this: week 1 ST 6-9%, week 3 cumulative 22-28%, week 6 cumulative 50-58%, week 9 cumulative 72-80%, week 12 cumulative 88-94%. Terminal clearance handles the last 6-10%. Margin is preserved because most of the season sold at full or near-full price.

A troubled curve diverges early. Week 1 ST of 3-4%, week 3 cumulative 12-15%, week 6 cumulative 28-34%. The terminal projection is 55-65%, leaving 35-45% of buy stranded. By the time the calendar says "markdown" at week 6, you're already 4-5 weeks behind where intervention would have mattered most.

Early intervention math

A 15% markdown at week 3 on a troubled SKU often clears the inventory at 80-85% terminal ST. The same SKU, untouched until calendar markdown at week 6, requires a 30% cut to hit 70% terminal ST, leaving 30% for clearance at 50-60% off.

Run the numbers on a $20 retail SKU with $8 cost, 1,000 units bought:

• Curve-based path: 15% off starting week 3. 85% sells at average $17.50 (mix of full price weeks 1-2 and 15% off thereafter), 15% clears at 50% off ($10). Total revenue: $14,875 + $1,500 = $16,375. Margin: $8,375.
• Calendar-based path: Full price weeks 1-5, 25% off weeks 6-9, 40% off weeks 10-12, clearance at 60% off after. 28% full price ($20), 32% at $15, 25% at $12, 15% clears at $8. Total revenue: $5,600 + $4,800 + $3,000 + $1,200 = $14,600. Margin: $6,600.

The curve-based intervention preserved $1,775 of margin on a single 1,000-unit SKU. Multiply by 400-600 troubled SKUs per season at a mid-market retailer: $700K-$1.1M in margin recovery per season, $1.4M-$2.2M annually across two seasons.

The curve is a signal, not a rule

Weekly ST is one of multiple inputs, not the only input. A SKU running below curve in week 2 might be experiencing a fixable problem: blocked planogram placement, missing signage, an out-of-stock at a peer store creating localized inventory imbalance, or weather affecting the trade area.

The discipline is to investigate before discounting. Most "fixable" weekly ST shortfalls show up in 1-2 stores out of 50, not chain-wide. If 47 stores are tracking curve and 3 are not, the issue is operational, not demand-driven. Operational fixes don't cost margin. Markdown does.

The retailers running curve-driven markdowns at scale use a tiered framework:

• Tier 1: Investigate. Chain ST on-track, but 2-5 stores below curve. Fix the local issue: planogram, replenishment, signage, fixture.
• Tier 2: Localized markdown. Chain ST on-track, but a regional cluster of 8-15 stores running 25%+ below curve. Region-specific markdown rather than chain markdown.
• Tier 3: Chain markdown. Chain ST 20%+ below projected curve at week 3-4. Take the chain action, take it early, and use a smaller percentage than the calendar default.
• Tier 4: Reorder freeze + accelerated clearance. SKU is so far below curve that recovery is impossible. Freeze open POs, accelerate clearance markdown immediately, free shelf space for the next launch.

Store-level ST variance hides the signal

Chain ST is an average. A 28% chain ST at week 3 includes stores at 18% and stores at 38%. If you markdown based on the chain number, you discount the 38% stores unnecessarily and don't discount the 18% stores aggressively enough.

Store-level ST decomposition is where the real signal lives. Cluster stores by trade area demographics, climate, and competitive density. Run sell-through curves by cluster. Cluster A might be tracking curve at 32%, cluster B at 24%, cluster C at 16%. The markdown strategy should differ across clusters.

This requires data infrastructure that most retailers don't have natively. SKU-store sell-through at weekly cadence, joined to cluster assignment, with curve baselines for each cluster from prior-season data. Most BI tools can produce this report once, but they can't run it continuously across 5,000 SKUs and 100 stores without a data engineer pulling shifts to maintain it.

This is where signal-based monitoring infrastructure earns its keep. The SKU-store-week sell-through curve runs continuously. Cells diverging from curve trigger alerts. The merchandising team sees the divergence in week 3, not week 6. The intervention happens with margin intact.

The year-over-year improvement loop

Sell-through curves compound across seasons. Each season's curve data improves the baseline for the next season. SKUs are classified, store clusters are refined, intervention thresholds are tuned. Year 2 curve-based markdowns are 30-40% more accurate than year 1 because the model has actual data instead of category averages.

One mid-market apparel retailer ran the curve-based system for three seasons. Year 1 margin recovery was $1.6M. Year 2 was $2.4M. Year 3 was $3.1M. The same operational discipline applied with progressively better data produced compounding margin returns. Most importantly, terminal clearance percentage dropped from 18% of receipts to 9%, which freed shelf space for trend-right product earlier in subsequent seasons.

The retailers stuck on calendar markdowns are leaving this entire compounding curve on the table. The first season of switching is the smallest gain. The fourth season is when the math actually shows up.

Key takeaways

• Calendar markdowns are designed for operational simplicity, not margin. Curve-driven markdowns typically recover 80-160 basis points of season margin.
• Weekly sell-through trajectory in weeks 1-3 predicts terminal sell-through at week 12-14 with 75-85% accuracy. The signal is available early enough to matter.
• Early intervention with smaller discounts beats late intervention with larger discounts. A 15% cut at week 3 typically outperforms a 30% cut at week 6 by $1,500-$2,000 per 1,000-unit SKU.
• Weekly ST is a signal, not a rule. Most off-curve performance in weeks 1-3 has fixable operational causes: planogram, signage, replenishment, fixture. Investigate before discounting.
• Chain-level ST averages hide store-level variance. Cluster-based markdown strategies are needed when stores diverge 15-30% from the chain curve.
• Sell-through curve data compounds across seasons. Year 3 curve-based markdowns are typically 30-40% more accurate than year 1, with margin recovery growing season over season.
• One mid-market apparel retailer recovered $7.1M in margin over three seasons of curve-driven markdowns, with terminal clearance dropping from 18% to 9% of receipts.