Beyond Accuracy: What MAPE and MPE Really Tell You About Your Freight Network
When it comes to freight cost prediction, accuracy is everything—until it isn’t.
At Transfix, we believe accuracy isn’t just about hitting a number. It’s about understanding your process, managing risk, and ultimately, driving better business outcomes. That’s why we lean into two core metrics—MAPE and MPE—as tools for both measurement and insight.
But before we get into it, it’s important to note:
We’ll never promise a perfect MAPE—because freight is too dynamic for that. But we will promise to make sense of your data, help you understand your volatility, and build smarter, more stable processes that set you up for sustainable profitability.
Because in freight, MAPE doesn’t make you money, cost control and risk management do.
What Is MAPE?
MAPE (Mean Absolute Percentage Error) tells us, on average, how far off our predicted rate is from the actual rate a broker pays. It’s expressed as a percentage and is always positive.
Think of it as a measure of how volatile your cost structure is. A lower MAPE generally signals more consistency in your lanes, carrier base, and buying process.
For example:
Brokers who run highly repeatable lanes with stable carrier relationships consistently see MAPEs around 5–6% network-wide, and as low as 1–2% on specific lanes.
Brokers with more fragmented operations often fall in the 8–9% MAPE range. That’s not a bad model. It’s a reflection of a more dynamic, less predictable cost environment.
In other words, MAPE is a mirror. It shows how structured—or volatile—your operation really is.
What About MPE?
MPE (Mean Percentage Error) looks at bias. Are predictions consistently high or low?
Where MAPE measures magnitude, MPE tells us direction. Ideally, a healthy MPE is close to zero, meaning errors aren’t skewed one way or the other.
Why does this matter? Directional bias can impact decision-making. If you're consistently under-predicting costs, you're setting margin expectations too high. If you're over-predicting, you may be leaving savings on the table.
Why Most Models Stop Here—and Why Transfix Doesn’t
Traditional pricing tools often serve up a single rate—take it or leave it. But freight isn’t static, and neither are your costs.
At Transfix, we go deeper. Our cost models don’t just spit out a price. They provide a cost range, from aggressive targets (SCM10) to the median (SCM50) to conservative rates (SCM90). That range accounts for critical operational variables like:
Lead time
Number of stops
Carrier availability
Market volatility
This approach allows us to calibrate your pricing model based on your actual business conditions, not just generalized market trends.
MAPE as an Investigative Tool
Rather than using MAPE to claim blanket “accuracy,” we use it to start a conversation:
Where is your network consistent?
Where are you introducing unnecessary risk?
How can your buying process be improved?
Put simply: High MAPE = High risk. And that’s a problem we’re built to solve.
Your Business Is Unique. Your Model Should Be Too.
Every broker is different. Some are hyper-focused on one region. Others manage a complex national network. What works for one won’t work for another.
That’s why we train custom models on your data, reflecting the reality of your operations and empowering your team with a model that evolves as your business does.
MAPE and MPE are important benchmarks, but they’re not the destination. They're indicators that help us provide consultative insights into how you’re buying, where you can tighten your margins, and how to become more competitive in the market.
Ready to See How Your Network Measures Up?
Talk to our team to get a cost model tailored to your data and insights that go far beyond the rate.
