It was 6:30 in the morning. I'd barely walked into my office when the email landed — our biggest client's QA director needed the shrinkage detail on a shipment from last September: lot numbers, locations, who touched it, everything. I said no problem, pulled up the new WMS, and my stomach dropped. When we'd migrated off the old system, the historical transaction data came out as a pile of CSVs with cryptic internal field abbreviations. Lot genealogy was broken — which lot, received when, stored where, picked by whom, all of it unmatchable. That's when it hit me: during selection, everybody asks what the system can do. Nobody asks who owns your data when you leave a SaaS WMS. We in the warehouse call it something simpler — letting someone else hold the keys to your own house.
The Day You Switch Is the Day You Learn Who Owns Your WMS Data
Late 2022, our DC switched SaaS WMS providers. We're a 3PL in the Inland Empire — roughly 180,000 square feet, about 3,000 orders a day in peak season, two hundred people on the floor. Selection took three months. We negotiated price, sat through five rounds of demos, nailed down the go-live date. The one thing we never talked about was the exit. The contract had exactly one line about data export: "data export services provided in a reasonable manner, fees to be determined." Nobody blinked. Who thinks about leaving on the day you move in?
First quarter of 2023, the contract expired and we didn't renew. Migration time. Master data came out fine — items, locations, customers, no problem. But six years of inventory transactions — receiving, putaway, moves, picks, cycle count adjustments — came out as raw CSVs with the vendor's internal shorthand for column names and no data dictionary. Worse, lot lineage was severed. No way to trace a lot from receiving dock to pick slot to shipper.
Then came that morning. The client audit needed shrinkage detail on 3 SKUs from September 2023, a variance of over 400 units, and we couldn't produce the chain of evidence. What did we do? Three people, three full weeks, digging through paper records, cross-checking emails, reconstructing moves from memory, location by location. Let me be straight with you — every morning of those three weeks, I walked into the office embarrassed. The data wasn't stolen. We handed it over ourselves, with one vague sentence in a contract.
A Peer Paid Nearly $20K for the Same Lesson
A buddy runs an e-commerce DC in Fontana, about half our size. Summer 2023, he switched systems, and the old vendor had him by the throat: "Sure, we can export your data — at our standard professional services rate." The quote: $25,000, plus engineers at $800 a day, estimated 15 days. Do the math, folks — nearly forty grand to get back your own data. The contract said "reasonable fees." Reasonable according to whom? They argued for two months. He got it down to $18K, but the go-live slipped six weeks. Six weeks, right before peak. You know what six weeks costs in Q4.
I tell this story over beers now, and I always land on the same uncomfortable truth: when you buy SaaS, you think you're renting software. What you're really renting is a door that only locks from the outside. The salesperson dazzles you with features during selection. The implementation team rushes you through go-live. Nobody reminds you to sign the divorce papers. By the time you want out, you realize you never negotiated custody of the data.
Some folks say, just export everything — five, seven years of history. Honest truth? It depends on what it costs. The bigger the dump, the harder the schema mapping, the higher the consultant bill. We ran the numbers internally: rebuilding the schema for six full years of transactions, an outside consultant quoted $30K to $40K — and couldn't guarantee the lot chain would be reconstructable. Meanwhile, tax authorities and client audits almost never go back more than two or three years. So here's our trade-off, and it's deliberate: full master data, plus three years of transactions, plus lot genealogy exported and verified separately. Half the cost, everything auditable. More isn't better — verifiable and sufficient is better.
Who Owns Your Data: The 5 Clauses We Now Demand in Every Contract
Every contract we sign now goes through legal and IT together, and these five clauses are non-negotiable:
1. Ownership stated in the body, not an appendix. "All customer data is owned by the customer" — ownership, not a license to use. Include derived data too: reports, aggregated metrics, anything their algorithms generated from your data. Sounds fussy. It's the sentence you'll lean on when things go sideways.
2. Export format and schema in writing. Demand a complete data dictionary: field names, types, meanings, table relationships. Exports in CSV at minimum, with the relational structure documented. And format alone isn't enough — we run a full export drill every January. Export it, open it, match the lots. Half a day of work. Our first drill, January last year, caught a real problem: after a vendor upgrade, the cycle-count adjustment table was missing a field. IT made them fix it on the spot. Without that drill, we'd have found out on migration day. Too late.
3. Cap the transition assistance fee. Put a ceiling on it, or tie it to a percentage of the annual subscription. Never leave "reasonable fees to be determined" in there. Remember: the moment you have the most leverage is before you sign.
4. Data retention after termination. The contract must say your data stays on their servers at least 90 days after termination, exportable anytime during that window, and deletion requires your written confirmation. Don't let them wipe the database the week after you move out.
5. Self-serve API export rights. During the contract term, you should be able to pull all your data through the API yourself — no tickets, no begging. Many systems only expose a subset of tables via API. Get the API docs during selection and check them table by table: orders, inventory transactions, lots, cycle counts. Every single one.
Here's the honest math:
| Approach | Cost | Outcome |
|---|---|---|
| Negotiate the 5 clauses upfront | 1–2 extra weeks, a few thousand in legal fees | You export on your own terms, zero drama |
| Skip it, negotiate during exit | $25K+ export fees, 6-week delay | Six weeks before peak — losses you can't calculate |
| Lose the data, hire rebuilders | $30K–$40K in consulting | Lot chain may never be reconstructible |
Folks, during selection, the salesperson buys you lunch and the demo looks beautiful. That's all normal. But before you sign, flip to the data section of the contract and ask three questions: who owns it, how do I get it out, and what does it cost me to leave? If the salesperson hems and haws, make them bring someone who can actually commit. That one page is worth three years of peace of mind.
By Chen Feng, Warehouse Operations Manager
