Forecast Accuracy
Forecast accuracy measures how close your revenue forecast came to actual sales. Here's how to calculate MAPE and the benchmarks to aim for.
Forecast Accuracy
How close a revenue forecast was to actual results, usually measured as MAPE — the average percentage error across periods.
Forecast accuracy is the gap between what you predicted you'd sell and what you actually sold, expressed as a percentage error. The standard metric is Mean Absolute Percentage Error (MAPE): the average of the absolute deviations between forecast and actual, divided by actual.
Low MAPE means your planning is dependable — you can commit to inventory orders, paid-media budgets, and hiring with confidence. High MAPE means you're either over-buying stock that ties up cash or under-buying and stocking out during peak demand. Most online stores treat sub-10% monthly MAPE as the working threshold for a healthy forecast.
Forecast accuracy sits inside the broader practice of revenue intelligence — the discipline of turning historical sales data into reliable forward-looking numbers. If your forecasts are consistently wrong, every downstream decision (ad spend pacing, restock timing, warehouse staffing) inherits that error.
The point isn't to hit a perfect forecast — that's impossible with real consumer demand. The point is to know how wrong you usually are, in which direction, and to shrink that error over time. A team that knows its MAPE is 8% and bias is +2% can plan around it. A team that doesn't track accuracy is gambling.
MAPE = (1/n) × Σ |Actual - Forecast| / Actual × 100
n
Number of periods
How many forecasted periods you're averaging across (e.g. 12 months).
Actual
Actual revenue
The real revenue booked in that period.
Forecast
Forecasted revenue
What you predicted at the start of the period.
A skincare brand forecasted €120k for March but booked €132k actual revenue. They calculate the absolute percentage error for that month.
Forecast: €120,000
Actual: €132,000
Periods (n): 1
→ 9.1% MAPE
A 9.1% error for a single month is within the acceptable band for a mid-size online store. Repeated over 12 months at this level, the brand would hit roughly the industry-average annual MAPE.
MAPE is the most common accuracy metric, but it has a known weakness: it punishes under-forecasts more than over-forecasts and breaks down when actuals approach zero. For new SKUs or seasonal launches with small baselines, supplement MAPE with absolute error in euros so a 50% miss on a €2k week doesn't dominate the average.
Monthly revenue forecast accuracy benchmarks for online retail
| Store maturity | Excellent MAPE | Good MAPE | Needs work |
|---|---|---|---|
| Established store, stable SKUs (3+ yrs) | <5% | 5–10% | >15% |
| Growing store, expanding catalog (1–3 yrs) | <8% | 8–15% | >20% |
| New store or new vertical (<1 yr) | <15% | 15–25% | >30% |
| Peak season (BFCM, holiday) | <10% | 10–18% | >25% |
Track accuracy at the same cadence you forecast. If you commit to monthly numbers, log MAPE every month and review the rolling 3-month average. Watch the bias too — a consistent +5% means you're systematically optimistic, which is a different problem than random noise and usually traces back to one input (a paid-channel ROAS assumption, a returning-customer rate).
Forecast Accuracy FAQ
For an established Shopify store with stable SKUs, monthly MAPE under 10% is considered good and under 5% is excellent. Newer stores or those with rapidly changing catalogs typically run 10–20% MAPE and improve as they accumulate historical data.
MAPE measures the size of your average error regardless of direction. Bias measures whether you systematically over- or under-forecast. You want both low MAPE (small errors) and near-zero bias (errors that cancel out).
Store-level forecasts are almost always more accurate because individual SKU demand is noisy. Forecast revenue at the store or category level for planning, and only drop to SKU level for inventory decisions on your top 20% of products.
At minimum 13 months so you can capture year-over-year seasonality. 24+ months is much better because it lets you separate trend from seasonality. Stores under a year old should expect higher MAPE until they accumulate a full seasonal cycle.
MAPE divides by actual revenue, so when actuals are small (a launch week doing €500), even a €250 miss becomes 50% error. Use absolute euro error or weighted MAPE for launches, then switch to standard MAPE once volume stabilizes.
Monthly at minimum. Each month, log actual versus forecast, calculate that month's APE, and update your rolling 3-month and 12-month MAPE. Quarterly, review which inputs (channel ROAS, conversion rate, AOV) drove the largest errors.
It matters more, not less. Fast growth amplifies the cost of bad forecasts — you'll either stock out and lose momentum or over-buy and tie up cash you need for ads. Even at 50% YoY growth, MAPE under 15% is achievable with disciplined tracking.
In order of impact: paid-traffic assumptions (ROAS or CPC shifts), conversion-rate changes from site updates or seasonal traffic mix, return-rate misestimates in apparel and footwear, and promotional cannibalization where a discount pulls forward demand from later periods.
Forecast net revenue (after returns and refunds) for any vertical with returns above 5%. For apparel and footwear, gross-revenue forecasts can look accurate while net revenue misses by 20%+ because return rates aren't modeled. Track both, but plan on net.
Forecast accuracy is the feedback loop that makes revenue intelligence trustworthy. Without measured accuracy, your forecasts are opinions. With it, you know which inputs are reliable, where your blind spots are, and how much buffer to build into operational plans.
See Metricuno on your data
Bring your stack — Google Analytics, Stripe, a CRM, anything — and we'll walk through the metric tree that turns your funnel into one number.