Recommender Suggests the Wrong Size
When the recommender’s suggestion looks off, it’s almost always one of four causes — each takes a minute to check with Fit Lab.
1. Body vs garment is set wrong (most common)
Section titled “1. Body vs garment is set wrong (most common)”The chart’s measurement type changes the math completely:
- Body — numbers matched as-is.
- Garment — ease subtracted and +10% stretch applied before matching.
If you entered body measurements but the chart says Garment (or vice versa), every recommendation shifts by roughly a full size. Open the chart in the editor — the choice sits right under the chart name — and verify it matches how the numbers were actually measured.
2. Cells were excluded from scoring
Section titled “2. Cells were excluded from scoring”If a cell isn’t a measurable range — free text, a single number without a range, “≈86” — it’s excluded from recommendations while still displaying. The save toast warns “Saved with N warnings — check excluded cells”.
Fix: open the chart; find the excluded values and enter proper ranges (86–89, or a single value the app can pair). Re-save — no more warning means everything scores.
3. The wrong columns drive recommendations
Section titled “3. The wrong columns drive recommendations”Check Size recommendations use in the editor. Only measurable columns (cm, in, kg, lb) can be selected — and only the ones you tick are scored. A chart that scores only “Waist” while shoppers think in bust sizes will feel wrong.
4. The ranges themselves are off
Section titled “4. The ranges themselves are off”Run 3–4 typical shoppers through Fit Lab:
| Symptom in Fit Lab | Likely cause |
|---|---|
| No size reaches Perfect for a typical shopper | Ranges shifted — re-check the source numbers |
| Everything lands Too tight / Too loose | Measurement type wrong (see #1) |
| Two adjacent sizes always nearly tie | Adjacent ranges overlap — fine, ties go larger |
| Verdicts jump around between columns | One column’s numbers use a different unit than its header says |
How ties and edge cases behave
Section titled “How ties and edge cases behave”- Ties go to the larger size — deliberate: slightly loose beats too tight.
- Tight is penalised about twice as hard as loose — published scoring method, see How the recommender works.
- A shopper exactly on a range boundary scores as inside it.
Height & weight vs exact measurements
Section titled “Height & weight vs exact measurements”Height & weight mode estimates measurements via regression and is labelled “Estimated” for shoppers. If a shopper reports a bad result from height & weight but a good one from exact measurements, that’s the estimation gap — expected behaviour, not a bug.
Still wrong after these checks?
Section titled “Still wrong after these checks?”Send us from the app’s Contact us page: the chart name, the shopper’s inputs (height/weight or measurements), the suggested size, and the size you expected. Fit Lab reproduces any case exactly, so screenshots of both help.