Custom Patch Production Data 2026

The median custom patch order is just 25 pieces

The industry talks about patches as a bulk product — the data says otherwise. Across all 965 orders, the median order was 25 patches, 60.6% of orders were 25 pieces or fewer, and 78.2% stayed at or under 50. Genuinely large runs are rare: only 4% of orders exceeded 250 pieces, topping out at a single 9,600-piece job. (The mean — 71.67 pieces — says more about a handful of large orders than about a typical customer.) The median moved up from 20 in our first edition, the one headline figure that shifted between editions.

n = 965 orders. This is why we hold a 5-piece minimum — low-quantity demand is the market, not the exception.

Methodology

Inside our factory1:28 · embroidery, chenille, PVC, laser leather

The floor these orders came off. Every order counted here was produced at our own facility in Pakistan, operated by Panda Apparel & Technology — embroidery, chenille, PVC and laser-engraved leather, filmed September 2026.

• Sample: every order delivered through pandapatches.com for the window December 8, 2025 through August 5, 2026 — 965 orders, 69,166 patches. Includes both web-checkout orders and orders managed by our sales team; no sampling. Our factory also produces patches for other patch brands and resellers, and those orders are not in this dataset.
• What counts as a patch: finished patches only. Non-patch line items — challenge coins, pins, keychains, sample boxes, and digitizing-only services — are excluded from both the totals and every percentage. That removed 12 orders and 2,956 pieces from the raw window of 977 orders and 72,122 pieces, leaving the 965 orders and 69,166 patches published above. We quote the patch-only number because a page titled “patch production data” should not count keychains.
• Field completeness: quantity was recorded on 100% of orders, backing on 96.8%, and a parseable size on 99.3%. Percentages are computed against the orders where the field was specified.
• Size figures: free-text size entries were normalized to their largest listed dimension in inches. Our first edition withheld a measured share because we could not rule out that the 12-inch entries were a sales-team default; we have since checked, and the 217 twelve-inch orders span 44 distinct descriptions rather than one repeated value, so the large-format segment is real. The sized-order denominator is 958 (99.3% of the 965-order cohort), taken from this edition’s frozen cohort; an earlier prose figure of 921 predated that recompute. The 2-to-5-inch and 8-inch-plus shares and the multi-dimension count are carried from the prior sizing analysis and are being re-bucketed against this edition’s cohort. The remaining limitation is multi-dimension orders, described above. We still do not publish a “most popular size,” because attributing multi-size orders to one bucket makes a modal claim unsafe.
• Correction history. First edition, 18 July 2026: median 20 pieces, 61.3% at 25 or fewer, 5.1% rush, 81.6% iron-on. Second edition, 6 August 2026: 963 orders, 68,785 patches, 47 rush (4.9%). Third edition, 14 September 2026: 978 orders, 72,129 patches, 43 rush (4.4%) — those figures were the GROSS window, published as though the non-patch exclusions had already been taken off them, and its percentages were computed against three different denominators. This fourth edition, September 28, 2026, recomputes every figure from one script over one cohort: 965 orders, 69,166 patches, 43 rush (4.5%). The window did not change — the arithmetic did. The order count falls because 12 non-patch orders and one soft-deleted record are now genuinely removed rather than only described as removed. Rush is also now counted correctly. An earlier note here used a wider marker — urgent flag or a customer deadline date. That was wrong: a customer telling us when they need their patches is a deadline to plan around, not a rush request, and it carries no rush fee. Rush means the customer asked for rush or was charged the rush fee. All-time on that basis, to 9 September 2026, is 68 of 1,232 orders (5.5%). So this edition is reproducible. Backfill is why two editions of the same window disagree, and the fix is to freeze the sample rather than re-query it. The 965 order numbers behind every figure above are recorded as a fixed list at publication, so re-running these figures gives this edition’s answer whatever the database does next. The list is held internally — order records are not ours to publish — and is available to anyone auditing these numbers on request.
• Where the orders are made: every order counted here was produced at our own facility in Pakistan, operated by Panda Apparel & Technology. The processes behind these figures are on video: a walk through the floor covering embroidery, chenille, PVC and laser-engraved leather.
• What this is not: these are window figures, not lifetime totals, and they describe Panda Patches customers — a low-minimum-friendly supplier — so the order-size distribution may skew smaller than suppliers with 50-to-100-piece minimums.
• Updates: figures are re-pulled from source records for each edition by one script, so every number on this page, in llms.txt and in the glossary comes from the same computation. This is the fourth edition, published September 28, 2026. Next edition: January 2027.