AI Fit Shoes Plugin: Can a Size Finder Cut Returns for a Shoe Shop?

Possibly, but for a small shop the evidence has to come from your own orders. Size finders can cut wrong-size returns when shoppers actually use them and your size data is sound, yet vendor claims are not independent proof. Before paying, tag return reasons for a month, price your wrong-size returns, and trial one plan with a stop date.

Fit is the reason shoe shops most often blame for returns, and that is exactly why you should measure it yourself rather than assume it. A size finder that is switched on but ignored by most shoppers changes nothing. One that gives confident, wrong advice can raise returns. The method below tells you which of those you are paying for.

Follow me on Instagram@sagnikteaches

How a shoe size finder works

There are three common designs, and the data each one needs is different. Knowing which you are looking at tells you what you must supply.

Connect on LinkedInSagnik Bhattacharya

• Foot scan. The shopper points a phone camera at their feet, and the app estimates length and width, then recommends a size per product. The onefit.ai listing on the Shopify App Store describes this: a browser-based scan, with recommendations adjusted for brand, product and the shoe's intended activity.
• Measurement entry. The shopper types in foot length (and sometimes width, height or age), and the app matches the numbers to each product's size chart. The listing for one such app says it takes foot length and width for footwear recommendations.
• Quiz or "find my size". The shopper answers a few questions, such as the brand and size they wear now and how they like a shoe to fit, and the tool matches them to your products. SmartSize's listing describes an AI fit quiz with this approach, and says shoppers answer once and are remembered on later visits.

All three depend on the same raw material: accurate size data for each of your products. If your chart says a size 9 is 27 cm and the shoe is cut small, the best algorithm in the world will pass the error to the customer, with more confidence than a printed chart would.

Subscribe on YouTube@sagnikteaches

What data the tool needs from you

• A size chart for each product or product group, in the same unit the tool uses. If your charts are mixed or missing, fix this first.
• Fit notes per brand or last. "Runs half a size small" or "narrow in the toe box". Your returns will tell you.
• Order and returns history. The listings I checked ask for access to products, collections and the last 60 days of orders, including returns. That is a lot of data for a plugin; read what each permission is for.
• Return reasons. If your returns form says only "does not fit", you cannot tell too small from too large, and the tool cannot learn either.

The tutorial on AI size recommendation for a clothing brand from charts and returns data works through this same data question with clothes, and the points carry over to shoes, where width and half sizes add complexity.

Prices on the vendor pages, and what they imply

Prices change, so treat the following as a snapshot of the USD prices on two Shopify App Store listings when I fetched them in October 2026, and check the live page before deciding.

Notice the two different billing shapes. SmartSize charges a flat monthly fee, so cost does not rise with traffic. The scan-based app charges by use, so a busy month costs more, which is only a problem if the scans do not turn into fewer returns. Reading the onefit listing as a monthly fee plus per-scan charges beyond the free allowance, 300 scans a month would cost about $219 on Launch ($24 plus 260 scans at $0.75) or about $173 on Growth ($95 plus 130 scans at $0.60). Check how your chosen vendor counts a scan, and whether a repeat visit by the same shopper counts again.

Only two apps are shown, because I could verify their listing pages this session; there are others, and the Shopify App Store lists many. Name a tool in your own notes only after you have opened its current page.

What the claims are, and what I could not verify

Vendors and roundup sites quote impressive reductions, and one claims fit causes a large share of online returns. I could not find an independent study with figures I could verify, so I have not repeated any of them. The numbers that matter are the ones from your own shop, and you can get them within a couple of months.

Step 1: find out what wrong-size returns cost you today

Pull the last three to six months of returns. If your return reasons are poor, go through them by hand, or paste a table (without customer names or addresses) into an AI assistant and ask it to classify them. Do this only with data stripped of personal details and in a business account. A short prompt:

Here is a table of shoe returns with columns: order_id, product,
size_ordered, reason_text. Classify each reason as one of:
TOO_SMALL, TOO_LARGE, TOO_NARROW, TOO_WIDE, NOT_AS_DESCRIBED, OTHER.
If the text is unclear, use OTHER. Then give counts per product.

Illustrative output:
Product            Too small  Too large  Too narrow  Other
Leather boot A          9          2          4        3
Canvas trainer B        3          7          0        2
Sandal C                1          1          0       12

Read the result with suspicion and spot-check twenty rows. The illustrative output shows something useful: boot A runs small and narrow, which a note on the product page might fix, trainer B runs large, and sandal C's problem is not fit at all, so a size finder would do nothing for it. Many shops find that wrong-size returns concentrate in two or three products. If so, a better chart and fit note on those pages is the cheapest test, and it should come before any plugin.

For finding the same kind of pattern across complaints and defects, how to spot patterns in complaints, returns and defects with AI goes further.

Step 2: price a wrong-size return

Take an illustrative shop with 400 orders a month, a 22 percent return rate (88 returns) and 60 percent of returns tagged as wrong size (about 53). Work out the cost of one return:

• Return postage you pay: $9
• Handling, inspection and restocking time: $4
• Resale loss: say 15 percent of returned pairs cannot be sold at full price, on an average price of $85, so about $12.75 per return on average

That is about $26 per return. Fifty-three wrong-size returns cost about $1,380 a month. Now you know the prize. If a size finder cut wrong-size returns by 15 percent, that is 8 returns, or about $210 a month; by 30 percent, about $410. Compare those to a $19 flat fee and the answer looks easy. Against $173 to $219 of scan charges, the 15 percent case barely breaks even and only the 30 percent case is comfortably positive. The same tool is a good buy or a bad one depending on the plan shape and what it achieves, which is why you test.

Step 3: change one thing at a time

You cannot read the result if you change the size chart, switch on the finder and change your return policy in the same week. A sensible sequence:

1. Weeks 1 to 4: Record baseline. Add structured return reasons (a dropdown with too small, too large, too narrow, too wide, other) to your returns form or your returns spreadsheet.
2. Weeks 5 to 8: Improve the size charts and fit notes on the two or three worst products. Measure again.
3. Weeks 9 to 16: Switch on the finder, on the cheapest plan that includes the feature, with a note in your diary to cancel on a set date.
4. Week 17: Compare wrong-size returns as a share of orders in each period.

Two cautions. Seasons matter: sandals in spring and boots in autumn differ, so compare like with like, or compare against the same weeks last year if your volume is stable. And returns lag orders by a week or two, so do not judge a period until the returns from its last orders have come in.

Step 4: check that people use it

A finder that three percent of visitors touch cannot move your return rate much. Ask the vendor dashboard, or count from your analytics, the share of product-page views that open the finder, and the share of completed orders where it was used. Then compare wrong-size return rates for orders that used it against orders that did not. The comparison is not clean, because shoppers who use a finder may differ from those who do not, but a large gap is a useful signal, and no gap is a warning.

A quick illustration: of 400 orders in a test month, 120 used the finder and 280 did not. Wrong-size returns were 7 of the 120 (5.8 percent) and 46 of the 280 (16.4 percent). That would be encouraging, though the usual caution applies: people who bother with a finder may be the careful ones. Combine it with the before-and-after total, and judge both.

What goes wrong with size finders

• Bad chart data. The tool is only as accurate as your charts. If three brands in your shop are mislabelled, it will repeat the mistake.
• Wide and narrow feet. A scan that estimates length well but width poorly leads to confident wrong answers. Ask how it handles width, and check results on a few staff with known feet.
• Brand differences. A shopper's usual size in one brand does not carry to another. Quiz-based tools depend on how good their brand table is, and small independent brands are often missing.
• Privacy and camera permission. Scans need camera access in the browser. Some shoppers will refuse, and you must keep a manual route.
• Page speed. Extra scripts slow product pages. Test mobile speed before and after, because a slower page can cost more sales than the tool saves in returns.
• Lock-in. Check what happens to your data and your shoppers' saved sizes if you cancel.

A mistake that shows how it can backfire

Consider an illustrative shop that installed a size quiz and left the default settings on. The quiz recommended a size 8 for a shopper who wore size 8 in a famous trainer brand, and the shop's own trainers were cut a full size small. Returns for that product rose over the next month. The cause was a missing brand adjustment, not a faulty app: nobody had entered the shop's own fit notes. The fix was to enter "runs one size small" for that range and to re-run the test on six staff before switching the quiz back on. Without the structured return reasons from step 1, the shop would not have spotted that the returns were concentrated in one product line.

Fixes to try before paying for a size finder

• A better chart. Foot length in centimetres and inches on every product, with a short "how to measure your foot" instruction.
• A fit note on each page. "True to size", "half a size small", "narrow fit". Take these from your own returns data.
• Review prompts about fit. Ask reviewers to say what size they bought, what they normally wear and whether it fitted. Show the comparison.
• A free exchange policy. If most returns are fit-related, easier exchanges can keep the sale. Cost it first; it changes your return sums.
• A chatbot on the product page. A trained assistant can answer "what size should I get?" from your notes, with a hand-off to a human. See whether a chatbot should handle order tracking and returns before you rely on one.

The bigger picture of returns, stock and suppliers is covered in best AI for ecommerce operations.

Testing the finder on your own staff before customers see it

Before you switch it on live, spend an afternoon on a ten-minute accuracy check. Pick six people with known shoe sizes, including at least one with wide feet and one who sits between half sizes. Have each use the finder on three of your products, then try the recommended pair on. Record whether the recommendation fitted, was a size too small, or a size too large. A rough scoring: 15 or more correct out of 18 suggests it is usable; 12 or fewer means the chart data or the tool's width handling needs attention. The sample is tiny, so this is a smoke test, not a study. Its value is catching an obvious error, such as the size-8 problem above, before a customer pays for it.

Write down the three worst misses with the product name and the person's real size. These are the notes you send to the vendor's support team, and the way they respond is useful information about what you will get after you have paid.

Working through the numbers for one product

Return to the earlier table of illustrative classification output and take leather boot A. Suppose it sold 80 pairs in the baseline period and came back 15 times, with 13 returns tagged as too small or too narrow. That is a 16 percent wrong-size return rate on a single product, against a shop average of about 13 percent (53 wrong-size returns out of 400 orders). Now change only the page: add "Runs half a size small and narrow through the toe; most customers go up half a size" and replace the chart with foot length in both units.

In the next 80 sales, wrong-size returns fall to 8. That is 10 percent, a cut of 6 points, from one hour of editing. Only after that do you switch on a finder for the product, and compare the following 80 sales. If the rate drops again to, say, 6 percent, the finder has earned some credit; if it stays at 10 percent, it has not added much beyond the better page. In both cases you now know, and you have paid nothing until the last step.

Keep a simple results table, one row per period, with orders, wrong-size returns, the rate and what you changed. It is small enough to keep in a spreadsheet and clear enough to show a business partner.

How much data is enough

With a small shop, eight returns against five could be chance. As a rough habit, treat a change as real only if it is both large (at least a third fewer wrong-size returns) and repeated across two periods or two products. Do not read anything into a one-month movement of a couple of returns. If you sell under 100 pairs a month, rely more on the qualitative evidence, such as the reasons customers give and the staff accuracy test, and less on the percentages. It is perfectly reasonable to conclude "not enough volume for a paid tool yet", and to revisit when orders grow.

Deciding, in order of cost

Pay for a fit tool only when all of these are true: wrong-size returns cost you more than ten times the cheapest relevant plan each month; your charts and fit notes are already tidy; your return reasons are structured; and you have a stop date and a measurement plan. Otherwise, fix the data first. In many shops, that alone produces most of the gain at no cost, and the plugin question becomes much easier to answer honestly.

Questions shoe shops ask about size finders

Do shoe size finders work on any store platform?

Most listings I checked are for Shopify, and other platforms have their own plugins or none. Check that the app supports your platform and theme, and that it can show on a product page without breaking checkout. Test it on a duplicate or hidden product before switching it on for everyone.

How many orders do I need to judge whether a size finder works?

More than most small shops have in a fortnight. As a rough guide, aim for at least two months of before data and two months with the tool, and compare wrong-size returns as a share of orders, not raw counts. With under 100 orders a month, treat the result as an indication rather than proof.

Are foot-scan size finders safe to use with customers' data?

Read the privacy policy and the app's listed data permissions before installing. A camera scan raises questions about what is stored and where. Tell customers what happens to the scan, offer a manual alternative, and, if you have customers in the EU, ask a qualified adviser about the consent wording.

Is a better size chart enough?

Often it is the first thing to fix. A chart that shows foot length in centimetres and inches, a note on width, and a line such as 'runs half a size small' taken from your returns data can remove a large share of wrong-size orders for nothing. Do it before paying for any app.

Further reads

• How to Automate Returns and Refunds With Clear AI Rules — Rules for handling the returns that still arrive.
• Taking a Local Shop Online With AI: Product Listings in a Weekend — Write product pages that set fit expectations clearly.
• Boutique Inventory Management: AI for Size Runs and Reorders — Size-run stock planning once fit data improves.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Shopify App Store listings for onefit.ai and SmartSize (fetched October 2026); plan names and prices are the listed USD prices and may change.