Three Months After Go-Live, Back to Spreadsheets
A home-goods warehouse spent $60K on a WMS. Three months after go-live, the warehouse supervisor led everyone back to Excel. The owner asked me if it could be saved. After a day on site, my conclusion: the system was fine — the people and processes were broken. That is the shared script of nearly every failed WMS implementation.
The industry's open secret: WMS project failure rates run far higher than vendors admit. And the causes cluster tightly around five reasons.
Reason 1: Starting Without Clear Requirements (The Biggest Killer)
The most common way to die. Requirements were vague at selection; consultants configured "standard processes"; at go-live nothing matched reality: your wave rules aren't supported, your billing logic needs custom development, nobody ever mentioned your returns flow.
Fix: before implementation starts, run a "process walkthrough" — take 20 real historical orders and run each one end-to-end, from order to shipment, inside the system. Anything that doesn't run becomes a change request on the spot. Don't wait for go-live.
Reason 2: Dirty Data Goes Live
A WMS is a data system: locations, SKU masters, inventory quantities — if any one is wrong, go-live is a disaster. The worst I've seen: a warehouse went live at 60% inventory accuracy. The system assigned picking tasks against phantom stock, the floor descended into chaos, and operations shut down within three days.
Fix: complete one full physical inventory before go-live; don't cut over below 95% accuracy. Remember the formula: garbage in, garbage out — the system amplifies data quality, good or bad.
Reason 3: No Executive Sponsorship
WMS implementation is organizational change, not an IT project. Projects where "the boss just signs checks" fail at alarming rates: departments deadlock with nobody to decide, veteran staff resist the new system with nobody to overrule them, and consultant decisions sit unapproved for a week.
Fix: the owner must hear a project briefing monthly and show up for key milestones (process sign-off, cutover). Executive time invested is the single best predictor of project success.
Reason 4: Drive-By Training
Many projects "train" with a two-hour PowerPoint the day before go-live. Operators pick up the PDA the next morning lost — they scan wrong and don't know how to undo it, hit an exception and don't know who to call. So everyone quietly reverts to the old ways: Excel and paper.
Fix: train in three tiers — operators (SOP per role + hands-on test; no pass, no go-live access), managers (reports and exception handling), IT (basic administration). "Test to qualify" is the line between training and theater.
Reason 5: Overaggressive Cutover — The Big Bang
"Full cutover next Monday!" — the most dangerous sentence in warehousing. Big warehouses especially can't big-bang: one hidden bug hits every shipment in the building.
Fix: pilot first. Cut over one zone, one shift, or 20% of order volume; run 2–4 weeks stable, then expand. Running dual systems during the pilot is exhausting — and it's the cheapest insurance you'll ever buy.
| Failure Cause | Typical Symptom | Prevention Cost vs. Failure Cost | |---|---|---| | Unclear requirements | Processes don't match at go-live | 2 weeks of discovery vs. 3 months of rework | | Dirty data | System assigns impossible tasks | 1 full count vs. 3-day shutdown | | Absent executive | Delayed decisions, turf wars | 2 hours/month vs. abandoned project | | Drive-by training | Staff revert to Excel | 1 week hands-on training vs. idle system | | Big-bang cutover | One bug paralyzes the warehouse | 1-month pilot vs. full shutdown |
Early Warning Signs: Five Red Flags in the 30 Days Before Go-Live
Many failures are sealed before go-live. If you're mid-implementation and spot any of these, stop and fix immediately — never gamble on "we'll sort it out after launch":
- The consultants understand your processes less than you do: if they ask "what does this document mean" more than three times in a meeting, discovery wasn't done properly.
- Master data keeps slipping: two weeks before go-live and SKU masters still aren't complete — chaos at launch is nearly guaranteed.
- Key users skip training: the warehouse supervisor says "too busy, send Xiao Li instead." The first person to resist the new system after launch is usually that supervisor.
- Testing covers only the happy path: test cases exercise standard flows only — no exceptions, no returns, no count variances. Day one of go-live, the exceptions will humble you.
- The executive misses two project reviews in a row: the most dangerous signal — the project has already fallen off their priority list.
The 30 days before go-live are your last braking opportunity. Delaying launch by two weeks always costs less than three months of failed operations.
Training pays for itself: one week of hands-on operator training costs a fraction of a single day of post-go-live chaos. And make the training stick — print the top 10 most-used PDA workflows on laminated cards and attach one to every device. The best SOP is the one within arm's reach.
Case Study: A Project Pulled Back From the Edge
A Texas auto-parts warehouse saw on-time shipping fall from 98% to 82% two weeks after WMS go-live; the owner was ready to kill the project. I found three problems: 15% of SKUs lacked weight/dimension data (the system couldn't cartonize), the night shift had never been trained on exceptions, and the big-bang cutover left no isolation. The rescue: two weeks to complete master data, dedicated night-shift retraining, and moving returns back to the old process. On-time rate recovered to 97% in a month, 99% in three. Most "failed" projects don't need a new system — they need a new playbook.
Conclusion
The five killers — unclear requirements, dirty data, absent executives, drive-by training, aggressive cutover — are not one bit technical. They're all management problems. Which is actually good news: technical problems need the vendor; management problems you can solve yourself. Before your next WMS project, print these five pitfalls and pin them to the meeting-room wall.
