On this page
Merchants shopping for "fit technology" often land on two very different products and assume they're interchangeable. A size chart generator builds a better table; a size recommendation engine tells one shopper their size. Confusing the two leads to buying the wrong tool for the actual return problem.
The confusion is understandable. Both categories market themselves with the word "size," both promise fewer returns, and both show up in the same Shopify app store searches. But they solve different halves of the same problem, and treating them as substitutes is why some merchants install a tool and see no real drop in returns.
What a Size Chart Generator Actually Does
A size chart generator is a content and formatting tool. It takes your product measurements — chest, waist, inseam, whatever applies — and turns them into a clean, standardized, on-page table. Some versions auto-populate from spreadsheet uploads, some convert between US/UK/EU sizing, some sync the same chart across every variant of a product line so you're not manually re-typing numbers for the fifth time.
The output is static. Every visitor sees the identical table. It doesn't know anything about the individual person looking at it — their height, their weight, their preference for a looser cut. It just presents information more legibly than a screenshot buried in a product description tab used to. That's a real improvement over no chart or a bad chart, but it's still asking the shopper to do the interpretation themselves: find a tape measure, compare their own body to a column of numbers, and guess which row applies if they fall between two sizes.
If your current baseline is an inconsistent, hard-to-read, or missing size chart, a generator is a legitimate fix. How to read a size chart correctly covers the shopper-side friction that a clean chart reduces, and it's worth understanding before assuming a table alone solves the problem.
What a Size Recommendation Engine Does Differently
A recommendation engine is a decision tool, not a formatting tool. Instead of publishing one table for everyone, it takes inputs from the individual shopper — height, weight, sometimes fit preference or a photo — and outputs a single answer: "you're a Medium in this brand." No comparison required, no guessing which row of a table applies to someone 5'7" and in between two chest measurements.
This matters because brands don't size consistently with each other. A Medium in one label can run a full size smaller or larger than a Medium in another, something covered in more depth in why clothing sizes are different between brands. A chart, however well formatted, can't correct for that ambiguity on its own — it still requires the shopper to self-measure and self-compare. A recommendation engine is built specifically to remove that step, which is also why it tends to move conversion and return-rate numbers that a static chart usually can't.
Here's the side-by-side:
| Size Chart Generator | Size Recommendation | |
|---|---|---|
| Output | One table for all shoppers | A personal size answer per shopper |
| Input needed | Product measurements | Shopper measurements (height, weight, etc.) |
| Solves | Inconsistent or missing charts | "What size am I in this?" uncertainty |
| Effort required from shopper | Self-measure, compare to table | Answer a few prompts, get an answer |
| Typical impact | Fewer support questions, cleaner PDP | Fewer wrong-size orders, lower return rate |
For a deeper breakdown of this exact contrast, see size chart vs size recommendation.
Where the Two Actually Overlap
Neither tool works in a vacuum. A recommendation engine still needs accurate, well-structured product measurements to calibrate against — garbage-in, garbage-out applies. If your product measurements are wrong or inconsistent across SKUs, a recommendation engine built on top of them will confidently give the wrong answer, which is arguably worse than no recommendation at all.
That's the practical case for treating a chart generator as infrastructure and a recommendation engine as the customer-facing layer. The generator standardizes and cleans the underlying data; the recommendation engine consumes that data to produce a personalized answer. Brands that skip the infrastructure step and bolt a recommendation widget onto messy measurement data tend to see inconsistent results and end up debugging data quality issues that look, from the outside, like a broken recommendation feature.
Why This Distinction Matters for Returns
Apparel and footwear already carry return rates well above the retail average, and sizing-related returns are the single largest driver within that category. US ecommerce returns totaled roughly $890 billion in 2024, with the online return rate running near 19-20%, according to the National Retail Federation. A chart generator can chip away at the portion of returns caused by shoppers who genuinely couldn't find or read the sizing information — a real but relatively small slice.
The bigger slice is shoppers who read the chart, measured themselves, and still guessed wrong, or who didn't bother measuring and ordered two sizes to "figure it out at home." That behavior — bracketing — is reported by roughly half of Gen Z shoppers according to the NRF, and it's a pattern a static table cannot interrupt because it doesn't ask the shopper anything; it just displays. Reducing that specific behavior is covered in how to reduce bracketing in fashion ecommerce, and it's a decision-tool problem, not a formatting-tool problem.
Where YourSizer Fits
YourSizer is a size recommendation and virtual try-on layer, not a chart generator — it assumes you already have (or are building) clean product measurements, and it uses them to answer the specific question a chart can't: "what size am I, in this exact product." Shoppers get a direct recommendation instead of a table to interpret, and can visualize the fit through try-on rather than guessing from numbers alone.
For merchants who don't yet have standardized charts, that's a separate, earlier problem worth solving first — a messy or missing chart undermines any recommendation layered on top of it. Once product measurements are in reasonable shape, a recommendation engine is what typically moves the return-rate needle, since it addresses the guessing behavior directly rather than just presenting information more clearly. Teams evaluating this path often start with how to choose a size recommendation solution to scope what "clean enough" data actually means before integrating.
Frequently Asked Questions
Can I use a size chart generator and a size recommendation engine at the same time?
Yes, and in most cases you should — they're complementary rather than competing. The chart generator keeps your underlying product measurements consistent and presentable, and the recommendation engine uses that same data to produce a personalized answer for each shopper. Brands that run both together typically get a cleaner PDP for shoppers who prefer to self-measure, plus a faster, more decisive path for shoppers who'd rather just be told their size.
Which one should I install first if I only have budget for one?
It depends on the state of your current sizing content. If your charts are missing, inconsistent across variants, or clearly a source of support tickets, fix that foundation first, since a recommendation engine built on bad measurement data will produce bad recommendations. If your charts are already solid and returns are still high, the gap is almost certainly decision-making, not information — that's a signal to prioritize a recommendation tool.
Does a size chart generator actually reduce returns on its own?
It can reduce a portion of returns, specifically the ones caused by shoppers who couldn't find, read, or trust the sizing information at all. It generally does not address the larger driver of sizing-related returns, which is shoppers who had the information but still guessed wrong or ordered multiple sizes as a hedge. For that behavior, a personalized recommendation tends to outperform a static table because it removes the guesswork instead of just presenting it more clearly.
Conclusion
Neither category replaces the other, and the merchants who see the best results tend to be the ones who stop asking "which tool is better" and start asking "which problem am I actually solving." Clean charts remove a low-effort source of friction and support tickets; personalized recommendations remove the harder, costlier problem of shoppers guessing wrong. As fit technology matures, expect the line between the two to blur further, with recommendation engines increasingly expected to sit on top of well-structured chart data rather than replace it.