Jewelry Store Software: Serialized Stock, Memo and Repairs Without the Spreadsheet

Why jewelry inventory software makes or breaks a multi store jeweler

Walk the back of any independent jeweler doing $8M to $30M a year and you will find the same stack: The Edge running on a server in the manager's office, QuickBooks Online for the books, a shared Gmail inbox for custom design threads, Podium for text messages, a Punchmark or Shopify site that does not know what is actually in the case, and a spiral bound repair envelope book nobody has been able to kill. Vendor memo lives in a spreadsheet called MEMO_MASTER_v4_FINAL.xlsx. The GIA reports are PDFs in a Dropbox folder named by whoever scanned them.

The expensive version plays out on the Saturday before Mother's Day. A customer wants the 1.51 carat oval halo she saw on Instagram. The associate checks The Edge, sees quantity 1 at the Westside store, calls over, and the manager there says it went out on approval Thursday and the paperwork is in a drawer. Twenty minutes gone. The customer leaves. Meanwhile the piece is Stuller memo, day 58 of a 60 day term, and nobody has invoiced or returned it, so on Monday you own a $9,400 ring you did not choose to buy.

Multiply that. Across the four store, 14,000 piece operations we have worked in, memo drift runs into six figures, the repair promise date hit rate is a number the owner cannot state without pulling envelopes, and the annual physical count eats two closed days and still produces a shrink figure the accountant argues with. That is not a discipline problem. The systems were built for retail, where a SKU has a quantity. Your SKU has a serial number, a certificate, an owner who may not be you, and a bench jeweler holding it in a tray.

Problem: every piece is a one of one, and half of them are not yours

The Edge handles serialized stock better than a generic POS (Point of Sale), but memo is where it and Lightspeed Retail both fall apart. Memo goods sit physically in your case under a vendor's ownership with a term, a memo number, and a return or invoice decision at the end. Off the shelf systems either treat memo as owned inventory, which inflates your on hand value and your insurance schedule, or park it in a side ledger nobody reconciles. Transfers between stores make it worse: the piece moves, the memo record does not, and the vendor's statement no longer matches yours.

A custom build models the piece, not the SKU. One record carries the serial, the metal and stone breakdown, the certificate number, cost basis, current location down to case and tray, and an ownership state: owned, memo in, memo out to a customer on approval, at bench, at a trade show, at the setter. Every state change writes an immutable event with who, when, and where. Memo terms drive a daily reconciliation job against the vendor statement, and the system tells the buyer on day 40, not day 61, that this ring is about to convert. Transfers become a two sided handshake with a scan on both ends, so the piece is never in two places or no place. Cycle counting by case, three cases a morning, replaces the annual shutdown.

Problem: repairs and custom orders are shop floor jobs, not line items

A ring sizing is not a product sale. It is a job with a take in associate, a customer owned stone you are now liable for, a photo condition record, a bench jeweler, possibly an outside setter, a promise date, a quote the customer approved by text, and a pickup. The Edge and Jewel360 both have repair modules, and they do the envelope and the ticket. What they do not do is route work. Nobody can tell you at 9am which of the 63 open jobs will miss their date, which bench is over capacity, or which job has been sitting at "waiting on part from Stuller" for eleven days.

What we build instead: a job record with a status machine, capacity aware routing, and a parts dependency. Take in captures weight, photos from four angles, and the customer's stone description, all signed on a tablet, because the argument three weeks later about whether that chip was already there is the single most expensive conversation in the store. Each bench has a work queue and a real throughput number, so promise dates are calculated, not guessed. When a job goes red, it texts the customer before they call you. Outside vendor legs (rhodium, laser welding, an appraisal) are tracked as sub jobs with their own clocks. Custom design threads attach to the job: CAD files, the deposit, the wax approval, the setting photo.

Problem: appraisals go stale the moment gold moves

Your Graduate Gemologist writes an appraisal in Word from a template, prints it, and it goes into a drawer and to Jewelers Mutual. Gold moves through the year, and every replacement value you issued goes stale with it. When a client has a loss, the underwriter works from your document, and if the number is low your client is underinsured and it is your name on the letterhead.

A build makes appraisals a living record rather than a document. The item pulls its own metal weight and stone specs from the serialized record, values against a live spot feed and your current vendor cost tables, and generates the PDF with your GG's credentials and the required disclosure language. The system then knows every appraisal it ever issued and flags them for re appraisal when metal or market moves past a threshold you set, which is also the cleanest re engagement letter in the business. Document extraction pays for itself here: point it at a GIA report PDF or a vendor invoice and it pulls the report number, carat weight, color, clarity, cut, measurements and fluorescence into the item record instead of an associate retyping it and transposing a digit.

Problem: compliance paperwork lives in a drawer

If you buy and sell more than the FinCEN threshold in covered goods, you are a dealer in precious metals, stones or jewels under 31 CFR 1027 and you owe a written anti money laundering program with a designated compliance officer, training, and independent testing. Cash payments over $10,000 trigger Form 8300. Most states make you a secondhand dealer when you buy scrap gold off the street, with police reporting and a hold period before you can melt. The FTC Jewelry Guides govern how you describe lab grown and sustainability claims, and rough diamond flow sits under the Kimberley Process. None of this is optional, and none of it is in The Edge.

A custom build enforces it at the point it happens. A cash tender crossing the threshold blocks the sale until the 8300 fields are captured, and aggregation across related transactions is computed rather than remembered. A scrap buy captures the ID scan, generates the state report in the format your jurisdiction wants, and holds the lot in a quarantine state that physically cannot be sent to Hoover and Strong until the clock expires. Every risk flag lands in a compliance officer queue with an audit trail your independent tester can read in an hour instead of a week.

Problem: the client book is in three heads and one spreadsheet

Your top associate knows that the Hendersons buy every anniversary in October and that he likes yellow gold. When she leaves, that knowledge leaves. Podium sends texts but does not know a client's ring size, their spouse's birthstone, or that they looked at three tennis bracelets in the 4 carat range in March.

Build the client record properly and then let AI do the parts humans skip. Purchase history, sizes, preferences, wish list, and every appraisal on file in one place. An after hours booking agent answers "can I come in Saturday to get my ring sized" at 9:40pm, checks real bench capacity, and books it, rather than a form that emails a mailbox nobody reads until Tuesday. Follow up drafts go to the associate with the client's actual history in them, and the associate sends or edits, never a blast. Forecasting matters most at buy time: the model reads three years of sell through by category, price band, metal and store, and tells the buyer that the $2k to $4k bridal band at the Northside store turns in 47 days while the same band at Westside sits 190, before you write the JCK order.

What a build actually costs and how long it takes

These are Digital Heroes numbers from delivery across 2,000 plus projects, not a market survey. A focused first release, meaning serialized inventory with memo reconciliation, transfers, repair job routing and the migration off your current system, typically runs $60k to $130k and ships in 12 to 16 weeks. A full platform adding appraisals with live metal pricing, compliance workflows, clienteling, e commerce sync and vendor integrations runs $150k to $400k phased over 6 to 12 months.

What pushes price up in this category, specifically: the number of vendor integrations (Stuller, Quality Gold, Rio Grande and RapNet each behave differently, and some have no API at all, which means file drops), migrating dirty history out of The Edge where two decades of free text descriptions need normalizing, RFID hardware if you want case level counts, offline capable POS because your Wi Fi will drop on the busiest day of the year, and state by state secondhand dealer report formats if you operate across lines.

Build versus buy: an honest line

If you have one or two stores, under roughly 5,000 serialized pieces, little or no memo, and repairs under 150 a month, buy. The Edge with the repair module, or Jewel360 if you want cloud, plus a tight process will serve you better than a $90k build you cannot staff. Anyone telling a single store jeweler to build custom is selling hours.

The signals that it is time: you have three or more locations and transfers are a phone call; memo carrying value is over $250k and you reconcile it in a spreadsheet; a person on payroll spends more than ten hours a week moving data between The Edge, QuickBooks and your website; your promise date hit rate is unknown; you are buying scrap and doing state reporting by hand; or you have a genuinely differentiated operation, custom design at volume, estate buying, watch service, that the incumbent tools model as an afterthought. Two or more of those and the build pays back inside two years. The software should encode how you actually operate, and at your size how you operate is the only thing a chain cannot copy.

How to choose a developer for jewelry store software

Ask them to whiteboard the data model in the first meeting. If they cannot draw an item that is simultaneously serialized, on memo from a vendor, out on approval with a client, and half owned as customer supplied material on a custom job, they will learn it on your budget. Watch for whether they model ownership separately from location. That one distinction predicts the whole project.

Ask what they have shipped against vendor and certificate integrations. Stuller, RapNet and IDEX are not REST APIs with tidy docs. Ask specifically how they handle a vendor with no API, because you have at least two.

Ask them to name the compliance surface unprompted: 8300, the FinCEN dealer program, your state's secondhand dealer rules, FTC lab grown disclosure. A firm that has not built this before will say "we can add reporting later." Compliance is a state machine that blocks transactions, not a report.

Ask who owns the code, the repository and the deployment on day one, and get it in writing before the first invoice. If the answer involves their platform, their hosting, or a license you renew, you are renting again, only worse, because now it is also bespoke.