What Is a Good MAPE for Demand Forecasting? A Benchmark Guide for Small Brands
If you're measuring your forecast accuracy and wondering whether your number is good, bad, or somewhere in between — you're already ahead of most brands your size.
Most founder-led consumer brands don't track MAPE at all. They feel the misses (a stockout here, excess inventory there) but don't have a single number that tells them how accurate their planning actually is. If you're calculating MAPE and looking for a benchmark, the fact that you're doing it at all is meaningful.
Now for the number you came here for: for a founder-led brand doing around $10M in revenue, a MAPE of 25–35% is normal and acceptable. Below 25% is strong. Above 40% consistently means something structural is wrong with how you're forecasting.
But that benchmark only makes sense in context. Here's what MAPE actually measures, why the right target changes at different revenue stages, and what to do when your number is high.
What MAPE Is (And What It's Not)
MAPE stands for Mean Absolute Percentage Error. It measures the average size of your forecast errors, expressed as a percentage of actual demand.
The formula:
MAPE = (1/n) × Σ |Actual − Forecast| / Actual × 100
In plain language: for each period you're measuring, calculate how far off your forecast was as a percentage of what actually sold. Average those percentages across all periods. That's your MAPE.
Example:
MAPE = (17.6 + 16.7 + 28.2) ÷ 3 = 20.8%
A few things MAPE doesn't tell you:
It doesn't tell you which direction you're off. A 25% MAPE could mean you're consistently over-forecasting, consistently under-forecasting, or randomly wrong in both directions. The direction matters — which is why MAPE is always used alongside bias. More on that below.
It treats a miss on a $500 SKU the same as a miss on a $50,000 SKU. If you're measuring MAPE across your whole portfolio, a slow-moving SKU with a 60% error is weighted equally to your hero SKU with a 15% error. For most decisions, you care much more about accuracy on high-velocity SKUs.
It can be gamed. A forecast that's always slightly under actual will show a lower MAPE than one that occasionally swings wide. If you're optimizing for the number rather than the decision, you can hit a good MAPE while still making bad inventory calls.
MAPE is the right primary metric. Just don't use it alone.
The Benchmark by Revenue Stage
Here's where most MAPE content falls short: enterprise benchmarks don't apply to small brands. A $2B CPG company with three years of SKU-level POS data, a team of demand planners, and statistical forecasting software should be hitting 10–15% MAPE on core SKUs. A founder with a spreadsheet and 18 months of Shopify data should not be measured against that standard.
Under $5M / Early stage
At this stage, you may not be tracking MAPE at all — and that's fine. Your SKU count is probably low enough that you can feel your inventory position without a formal accuracy metric. Focus on building clean unit-level sales history before you measure how accurately you're forecasting it.
If you are measuring: anything under 40% is reasonable. Demand patterns are still establishing themselves, you likely have limited history, and your channels may be changing (adding retail, shifting mix). Wide swings are normal.
$5M–$15M / Growth stage
This is where MAPE becomes genuinely useful. You have enough history to forecast statistically, enough SKUs to need a consistent method, and enough riding on inventory decisions to care about the accuracy.
Target: 25–35% MAPE on core SKUs.
This is the benchmark Lauren Pitts — who has planned demand for brands at this stage as well as $1B+ categories at Unilever — consistently cites for brands at this revenue level. It reflects the reality that you're working with limited data, variable lead times, promotional noise, and potentially a retail transition that makes historical patterns less reliable.
Below 25% at this stage is exceptional and usually means either your demand is very stable (a subscription model, for instance) or you're measuring over a period without meaningful volatility.
Above 40% consistently means something specific is wrong — either the forecasting method, the promotional assumptions, or the data inputs. Worth diagnosing rather than accepting.
$15M–$30M / Scaling stage
By this point you should have 2+ years of SKU-level history, visibility into retail sell-through (if you're buying syndicated data), and ideally a dedicated operator running your S&OP process.
Target: 15–25% MAPE on core SKUs.
The improvement from the $10M benchmark comes from better inputs, not just better math. More data, cleaner retailer signals, a longer promotional calendar to plan against, and a monthly process that catches assumptions before they cause expensive misses.
If you're at $20M+ and still running 35%+ MAPE, the problem is usually one of three things: promotional lift is being incorporated inconsistently, retailer inventory data isn't feeding back into the plan, or the forecast is being adjusted to match a revenue target rather than a demand signal.
MAPE Benchmarks at a Glance
These are portfolio-level ranges. For individual SKUs, expect more variance. A new product in its first two quarters will often run 40–60% MAPE — that's normal, not a failure. A mature hero SKU with three years of stable velocity should be tracking below 20%.
MAPE Alone Is Not Enough: Track Bias Too
This is the thing most small brands miss when they start measuring forecast accuracy.
MAPE tells you how big your errors are. Bias tells you which direction they go.
Forecast bias is calculated as:
Bias = Σ (Forecast − Actual) / Σ Actual × 100
A positive bias means you're consistently over-forecasting (you think you'll sell more than you do). A negative bias means you're consistently under-forecasting (demand keeps coming in above your plan).
Both are problems, but they have different consequences:
Positive bias (over-forecasting): You produce too much. Cash gets tied up in inventory. Storage costs accumulate. If the product has a shelf life, you're running a clock you might not win.
Negative bias (under-forecasting): You produce too little. You stock out. You miss sales. At retail, you risk losing shelf space or damaging a buyer relationship.
A brand with a 25% MAPE and near-zero bias is in a healthy position — errors are roughly symmetrical and cancel each other out over time. A brand with a 20% MAPE and a consistent +15% bias looks better on the headline metric but is systematically overproducing every cycle. The bias number is the one to fix.
Target bias range: +/- 10%. Beyond that, you have a directional problem that's costing you either cash or sales.
Why Your MAPE Is High (The Most Common Causes)
If your MAPE is running above your target range, it's almost always traceable to one of these:
Promotional lift isn't being modeled separately.
A 20%-off promotion, a Prime Day event, or a retailer display program will spike demand above your baseline. If your forecast doesn't explicitly account for those events — adding a lift factor on top of the base forecast during promotional windows — every promotional period will show as a large error. The fix isn't better math; it's a promotional calendar built into the planning model.
The forecast is based on financial targets, not demand signals.
This is the most common problem at founder-led brands. The revenue goal gets translated directly into a production forecast without asking whether the demand actually supports it. You'll hit your MAPE target consistently only when the forecast comes from the data, not the growth plan.
You're measuring at the wrong level.
If you sell ten SKUs but you're measuring MAPE on total revenue, errors across SKUs are canceling each other out. A high-selling SKU running hot can mask a slow-moving SKU running cold — and you'll miss both reorder points while your blended MAPE looks fine. Measure MAPE at the SKU level, at minimum for your top ten items.
Lead time variance is baked into your error.
If your supplier's lead time runs from 35 to 60 days depending on the month, and you're planning to a fixed 45-day lead time, your inventory will be off even if your demand forecast is accurate. Lead time variance shows up in MAPE as demand error. It's worth separating the two: how accurate is my demand forecast, and how much of my inventory error comes from lead time variance?
New SKUs are pulling the average up.
New products are inherently hard to forecast — no history, uncertain velocity, promotional launch noise. If you're launching regularly, new SKU MAPE will consistently run higher than mature SKU MAPE. Track them separately. Don't let a 55% MAPE on a 6-month-old product distort your read on your core line.
How to Improve Your MAPE Over Time
A few practical moves that consistently bring the number down:
Run a monthly forecast vs. actual review. The single most effective improvement is a regular cadence where someone looks at last month's forecast, compares it to actual, and asks what drove the gap. Most brands that track MAPE don't do this — they calculate the number and move on. The learning is in the investigation.
Weight recent months more heavily. If you're using a simple average of historical demand, periods from 18 months ago are pulling your forecast away from current trends. A weighted moving average (more weight on the last 3 months, less on older periods) usually reduces MAPE on SKUs with trends or seasonality.
Build a separate forecast for promotional periods. Take your promotional calendar and flag every period with a planned event. Apply a lift factor based on past promotional performance. Treat it as a separate demand layer on top of your baseline. This alone can reduce MAPE by 5–10 percentage points for brands running regular promotions.
Stop adjusting the forecast to match the sales target. This is uncomfortable advice but it's the most important one. When the forecast gets adjusted upward to support a revenue goal, MAPE goes up because you're introducing bias that the data doesn't support. The forecast should reflect what you expect to sell. The gap between the forecast and the target is a business problem to solve — not a planning problem to paper over.
If you're measuring forecast accuracy and benchmarking against these ranges, you're running a more sophisticated planning process than most brands your size. The next step is usually building a monthly cadence where the MAPE number drives actual decisions — not just gets reported and filed.
Not sure what's driving your forecast error or how your MAPE compares to where it should be?
Izba's free 20-minute demand planning audit looks at your current forecasting method, your accuracy history, and where the gaps are. We'll tell you what's off and what to do about it.
Related Insights
Why Adding More People to Your Forecasting Meeting Makes It Worse
Adding leadership to a struggling forecast review doesn't fix the miss — it kills the honesty that could. Here's what's actually happening in that room.
How to Calculate Safety Stock for Small Brands
The standard safety stock formula assumes data you probably don't have. Here's how to calculate it as a founder-led brand — with or without demand history.
We Had 8 Years of Inventory in the Warehouse. It All Expired.
One founder produced too much, couldn't sell it, couldn't donate it, and had to destroy it. Here's what demand planning could have prevented — and what it cost.