It Passed Acceptance and Nobody Uses It: The Real Causes of Low Software Adoption and a 90-Day Rollout Playbook

The Client's Real Problem: Signed Off, and Gathering Dust

A 120-person instrument distributor spent NT$820,000 last year on an order and customer management system. Five months of development, all 47 UAT items passed, three training sessions delivered. On day 90 after launch, actual usage in the sales department was 22% — the other 78% of orders still ran through Excel and LINE. The owner asked the vendor, "Did you build it wrong?" The vendor pulled the logs: stable, no major bugs, 0.8 second average response time. Neither side was wrong, and the money was still gone. This article is about that gap.

Industry Myths, Broken

Myth 1: "Passing acceptance means the project succeeded."
Reality: UAT checks whether features match the specification, not whether employees will use them. Those are entirely different questions. A system can be 100% specification-compliant and 0% used. Between the end of the contract and the start of value there is usually a 90-day gap that nobody owns.
Myth 2: "They are not using it because it is badly designed."
Reality: across the cases we track, interface problems account for roughly a third of adoption failures. The more common cause is that the incentive structure did not change — the old process still works, the new system takes longer, and managers still read the Excel report. People are rational. If you do not close the old road, they will not take the new one.
Myth 3: "More training sessions will fix it."
Reality: seven days after a single training session, retention of operational steps is typically under 30%. What works is contextual onboarding — a prompt inside the system at the moment the person is about to do the thing. Training has a far lower return than good in-product hints plus one internal champion.
Myth 4: "People will get used to it eventually."
Reality: adoption has a clear tipping point. If you are not at 50% within 60 days, the odds of natural improvement afterwards are low; usage usually plateaus and then drifts back to the old process. Time does not solve the problem. Time hardens the old habit.

The Core Framework: The Four-Factor Adoption Formula

Adoption = (Clarity × Incentive) ÷ (Switching cost × Width of the back door)

Clarity: does the employee know when to use it and what for? Not what features exist, but which of their own actions is being replaced.
Incentive: what is in it for them personally. "Company data will be more complete" is not an incentive. "You stop typing the customer name three times" is.
Switching cost: learning time, number of required fields, device changes, whether they can remember the password. Every extra mandatory field costs you adoption.
Width of the back door: can the old process still be used? This is the most overlooked and most lethal of the four. As long as Excel still accepts orders, Excel will keep accepting orders.

How to use it: before launch, score each factor 1–5 and estimate adoption. Anything scoring below 2 is where your rollout effort goes. The value is not precision. It is that the formula forces "rollout" into four separately actionable pieces.

Three Typical Scenarios

Scenario	Typical failure point	Factor to attack first	Suggested rollout window
Small (under 20), single department	The owner does not use it, so nobody does	Incentive, starting with the owner	30 days
Mid-size (50–150), cross-department	Sales adopts, warehouse does not, data breaks in the middle	Back door width — close both ends together	90 days
Multi-site chain (5+ locations)	HQ pushes, each store interprets differently, execution varies	Clarity — one champion per store	120 days

The Hidden Cost List

The cost of adoption failure is never just the wasted development fee. The real invoice looks like this:

Parallel-running labour: with both processes live, each person loses 1.5–3 hours a week. For a 50-person company over three months that is 900–1,800 hours, roughly NT$600,000–1,200,000.
Data-split repair: half the orders in the system, half in Excel. Merging six months later typically takes 80–160 hours of data cleaning, roughly NT$120,000–240,000.
Credibility damage: this failure raises internal resistance to the next digital project. It cannot be priced, but in practice it adds 2–3 months to the next initiative.
Sunk maintenance: nobody uses it and the retainer still bills. NT$15,000 a month, NT$180,000 a year, for a URL nobody opens.
Opportunity cost: that NT$820,000 spent instead on the small thing that hurts 80% of staff daily typically pays back in 6–9 months.

Vendor Scorecard, Adoption Edition

Beyond technical capability, score these 10 dimensions 1–5 each. Below 30, be careful:

☐ Does the proposal state a post-launch adoption target number, not just a feature list?
☐ Did they proactively ask how the old process will be closed?
☐ Do they require named internal champions written into the contract?
☐ Did they plan a parallel-running period with an explicit end condition?
☐ Do they deliver in-product contextual hints and empty-state guidance?
☐ Do they deliver a usage dashboard showing who uses what?
☐ Do they treat the number of mandatory fields as a design constraint to minimise?
☐ Have they scheduled 30/60/90 day review meetings?
☐ Are they willing to tie part of the final payment to an adoption metric?
☐ Will they tell you honestly which processes should not be digitised?

Our Offer, and Where We Are the Wrong Fit

We package the above as an Adoption Assurance add-on within the retainer engagement model, from NT$18,000 per month with a suggested three-month term:

Pre-launch four-factor assessment and an old-process shutdown plan
In-product contextual hints and empty-state guidance implementation
Usage dashboard by department, feature and individual
30/60/90 day review meetings and adjustments

Where we are not the right fit, stated plainly:

Organisations where the owner does not intend to use the system or require managers to. No tool recovers adoption from that.
Projects that want to "launch first and figure out the process later." Without an existing process to compare against, adoption assessment has no baseline.
Companies in the middle of restructuring or department mergers. When process ownership is unclear, any rollout becomes a political instrument.
Budgets that cover development with nothing reserved for rollout. We would rather cut scope than cut the rollout.

The 90-Day Transition Playbook

Day −14 to 0: announce the switchover timeline and the old-process shutdown date. Name one champion per department and give them accounts early to run a real cycle. Prepare a one-page "here is what your three daily actions become" sheet for every desk.
Days 1–14: parallel running. Both processes accepted, but publish usage daily. Target 40%. The goal here is collecting friction points, not forcing use.
Days 15–30: close the first back door. Pick the easiest old entry point to shut (for example, no more order photos over LINE), announce it, then do it. Target 65%.
Days 31–60: close the main back door. Excel stops being updated, reports come only from the system, and every weekly management meeting uses system data. Target 85%. This step is the watershed of the whole plan.
Days 61–90: optimise and close out. Redesign the least-used feature, handle remaining exceptions. Target 95%, and hand a list of non-users to managers for one-to-one follow-up.

Decision Checklist

☐ Can you state the exact date the old process closes?
☐ Will the owner or most senior manager personally use the system?
☐ Has every department named a champion?
☐ Will managers' routine reports be switched to come from the new system?
☐ Does the same task take less time in the new system?
☐ Are there contextual hints inside the product, not just training?
☐ Can you see the usage number every day?
☐ Have mandatory fields been trimmed to the minimum?
☐ Is the parallel-running period capped at 30 days?
☐ Is there a named owner for handling exceptions?
☐ Are three review meetings scheduled within 90 days of launch?
☐ Is there budget specifically for rollout (suggested 10–15% of development)?
☐ Can you accept 40% adoption in month one?

Ten or more ticks and adoption usually hits target within 90 days. Six or fewer, pause development and close these gaps first.

FAQ

How much rollout budget should we set aside?

10–15% of the development fee. On an NT$820,000 project that is NT$80,000–120,000, covering champion time, contextual hint implementation, the usage dashboard and three review meetings. The return is usually better than spending the same amount on two more features.

If employees push back hard, should we compromise?

Distinguish the type. Pushback about extra time is a valid signal and you should optimise the flow. Pushback about visibility is an incentive and management issue, and compromising there turns the system into an ornament. Collect friction points publicly and respond with an action within 48 hours so resistance becomes a managed list rather than an emotion.

Can we tie the final payment to adoption?

Yes, and it is good practice, but the terms must be symmetric. A reasonable clause ties 10–20% of the final payment to 70% adoption at day 60, while the client commits to naming champions and executing the old-process shutdown. Loading the risk entirely onto the vendor gets refused by most serious vendors, because half the variables sit on the client side.

We launched six months ago and adoption is 20%. Can it be recovered?

Yes, but at 2–3 times the cost of getting it right initially. Re-run the 90-day playbook, but do one thing first: find the 20% who do use it and ask why. Their answers are your most effective rollout material.

We are a 20-person company. Do we need all this?

Not the full programme. Under 20 people, three things suffice: the owner uses it, one champion is named, and a firm old-process shutdown date is announced. Those three cost nothing and determine about 70% of the outcome.

Get in Touch

If you have a system that was built but never used, or you are starting a project and want to avoid this pit, we offer a one-off adoption diagnostic: an assessment against the four-factor formula and an executable 90-day rollout timeline.