Tandon Gems

How often to cycle-count bead strands, and how to set ABC classes

Anish Tandon, Founder, Tandon GemsPublished October 17, 2025· Reviewed September 7, 2026· 4 min read

This article is a reference guide. Confirm the current strand, price and quantity with Anish before ordering.

Short answer

Count your A items weekly, B items monthly, and C items quarterly. Assign A to any SKU selling 5 or more units a week, B to 2 to 4, and C to 1 or fewer. Accept a variance of one strand or 2 percent per SKU, whichever is larger. Recount the same SKU when variance passes 3 percent.

In short

  • ABC by velocity: A is 5 or more units a week, B is 2 to 4, C is 1 or fewer. Reclassify each quarter from your own sales report.
  • Cadence: A weekly, B monthly, C quarterly. Earring-pair trays follow the A cadence because display trays drift fastest.
  • Tolerance: pass at 1 strand or 2 percent per SKU; recount immediately above 3 percent.
  • A pair tray fails when any single earring sits unmatched, even if the total piece count is right.
  • Keep each session to 15 to 25 SKUs, sequenced by shelf location, so the count actually gets done.

How do you assign ABC classes by sales per week?

Pull 12 weeks of sales by SKU and divide units by 12. That average sets the class, not gross dollars. A SKU that moves 5 or more strands a week is an A. Two to 4 a week is a B, and 1 or fewer is a C.

Dollar value can promote a SKU one class, never two. A slow but expensive strand belongs in B because the shrink hurts more. Rerun the report every quarter and let SKUs move up or down on their own.

  • A: 5 or more units per week
  • A by value: a high value strand you cannot afford to guess at, even when it moves slowly
  • B: 2 to 4 units per week
  • C: 1 or fewer units per week, including odd sizes and one-off leftovers

How often should you cycle-count strands versus earring-pair trays?

Strands live in hanks and bins, so they count fast and drift slowly. Pair trays drift fast. A customer picks up an earring, holds it against another one and sets it back in the wrong slot, which is why we tell shops to count pair trays on the A cadence even when the pairs sell like C items.

A practical rhythm: strands on the ABC schedule above, pair trays weekly for open display and monthly for anything behind glass. Do trays at open, before the first customer touches them. Do strands at close.

  • A strands: weekly, same day each week
  • B strands: monthly, split across four weekly sessions so no week is heavy
  • C strands: quarterly, one bin section per session
  • Open pair trays: weekly; glass-case pair trays: monthly

What error rates are acceptable before you recount?

Set the pass line at one strand or 2 percent of the on-hand quantity per SKU, whichever number is bigger. On a small bin, one strand off is a pass. On a large bin, the 2 percent figure gives you a little more room.

Above 3 percent, recount that SKU the same day with a second pair of eyes. If the recount confirms the gap, adjust the record, write down the reason, and move that SKU up one class for the next two cycles. Chronic offenders usually have a location problem rather than a theft problem.

  • Pass: within 1 strand or 2 percent
  • Recount now: variance above 3 percent
  • Recount now: any pair tray holding an unmatched single
  • Escalate: three failed cycles in a row on one SKU means split it into two locations

How do you count matched-pair trays without checking every piece?

Count pairs first, then verify matching on a sample. Take the square root of the pair count and round up. That many pairs get opened and checked for size, drill and color match.

One unmatched single fails the whole tray. Then you check every slot, because a single mismatch almost always means two mismatches sitting apart from each other. Log the fix. Repeat mismatches point to a slot layout that invites confusion.

  • Record pairs as pairs in your system, never as loose pieces
  • Keep singles in a labeled orphan box with a date on the label
  • Re-pair orphans monthly against new arrivals
  • Retire orphans older than two quarters to component stock

How do you build a cycle-count worksheet?

Split each class list into weekly slices so no week lands heavy. A items repeat every week. B items divide into four slices, one per week of the month, and C items spread across the weeks of a quarter, one bin section at a time.

Keep the sheet to seven columns: SKU, class, location, system quantity, counted quantity, variance percent, initials. Fill the system quantity from your report before you walk to the bin, and never look at it while counting. Blind counts are the only ones worth the time.

  • Sequence the sheet by location instead of SKU number, so you walk each aisle once
  • Cap a session at 25 SKUs; longer sessions produce sloppier numbers
  • Log the variance percent even when it passes, so you can watch drift build
  • Initial every line so every count has an owner you can ask

How do incoming strands from us fit your count records?

Standard strand length from our Jaipur cutters runs 13 to 16 inches, and faceted rondelles often arrive at 13 inches. Receive by strand count and hank, then check length before you shelve. A 13 inch strand and a 16 inch strand of the same size hold different bead counts and should not share a SKU.

Every listing in our strand inventory states stone, shape, size and cut, which is the same four-part naming pattern that keeps your own SKUs clean. When you build new codes, borrow that order and check the stone guides for spelling so your shelf labels and your reorder emails match.

Anish here: shops that reconcile the packing slip against the invoice on the day the box lands catch nearly every discrepancy while it is still easy to fix. We confirm stock before we send an invoice, so a mismatch is usually a receiving error, and it takes two minutes to find on day one.

Common questions

Do I still need a full physical inventory if I cycle-count?

Yes, once a year for your books. Cycle counting keeps daily numbers honest and makes the annual count quick, because most SKUs were verified within the past quarter. Your accountant will want an ending quantity and value regardless of how you got there.

Should I weigh beads instead of counting them?

A counting scale helps with small findings and loose beads of one size. Strands are better counted as strands, since length varies within a hank. Weight also drifts with stone density, so keep separate tare weights per SKU if you use a scale.

How long should one cycle-count session take?

Short enough to finish before staff get pulled back to the counter. If a session drags, your bin labels or your sheet order are working against you. Fix the walk path before you add staff hours.

What if the same SKU keeps failing?

Three failures in a row means the SKU sits in two places, shares a bin with a lookalike size, or gets sold as a partial without a record. Split the location, relabel the bins, and count it weekly for a month.

Can I use one tolerance for strands and pairs?

No. Strands tolerate a small quantity gap because a miscount corrects itself next cycle. Pair trays are pass or fail on matching, since one unmatched earring makes two items unsellable at full value.

Sources and checks

  1. Counting Scales and Doing Inventory - Jewelry Discussion - Ganoksin Orchid Jewelry Forum Community for Jewelers and Meta
  2. Inventory for tax purposes - Jewelry Discussion - Ganoksin Orchid Jewelry Forum Community for Jewelers and Metalsmiths
  3. Putting Jewelry into the POS & QuickBooks - Jewelry Discussion - Ganoksin Orchid Jewelry Forum Community for Jewelers an

Catalog details are reference information. Ask Anish to confirm the current strand, price and availability.

Want a strand like the ones in this post?

Message Anish with the stone, shape and size you have in mind. Photos of actual strands on request.

Ask on WhatsApp

Keep reading