Summary. Most enterprise software in retail is not rejected. It is installed, announced, and then quietly not used. This paper sets out ten practices that decide whether a rollout is adopted or merely deployed, who owns adoption once the implementation team leaves, what a stalled rollout costs, and how to tell the difference in your own reporting. Every external figure is footnoted.

Adoption is where the money leaks, and it is measurable

The gap between buying software and using it is now well documented. WalkMe’s State of Digital Adoption 2025, drawing on 3,700 executives and employees plus usage data from 1.5 million users across 2,400 enterprise applications, found enterprises lost more than $104 million in 2024 to underused technology, and that employees waste an average of 36 working days a year on technology frustrations.¹

The same research found only 28 percent of employees feel adequately trained on the tools they are given.¹ It also found executives believe an average of 37 applications are in use where the real number is 625, a seventeenfold gap between what leadership thinks is deployed and what is actually running.¹

The paradox is that spending has not fallen with usage. ATD’s 2025 State of the Industry research put formal learning at 13.7 hours per employee in 2024, down from 17.4 hours the year before.² Budgets held while the hours people actually spent learning fell. That gap is an adoption problem wearing a measurement disguise.

The seventeenfold application gap deserves a second look, because it is not really a finding about software inventory. It is a finding about how far leadership’s picture of the workplace has drifted from the one their people work in. If the count of tools is wrong by that margin, the assumption that people were trained on them is not worth much either.

The ten practices

1. Design for the end user, and start before you deploy. In enterprise software the end user is the customer. The Neochange and Oracle IT Adoption Insight Report found a direct correlation between beginning adoption work in the discovery phase and eventual adoption rates, with user-focused companies achieving 300 percent higher returns on their end-user programs than companies that started at or after deployment.³ By go-live the workflow decisions that determine whether anyone uses the thing have already been made. An associate should be recognized by role, channel and preferred language the first time they open the app.

2. Count the benefits that arrive later. Rollouts judged only on immediate operational cost savings get starved, because the effects that matter most on a sales floor (consistency, product knowledge, retention) show up over a defined period rather than in the first quarter. If a business case can only recognize headcount savings, it will not find a reason to invest, and the store will keep the binder.

3. Run change management as a permanent function, not a launch event. Retail turnover makes this structural rather than optional. A fleet that replaces a large share of its hourly staff every year is a fleet where a one-time launch campaign trains people who will largely be gone, and where the associates who most need the tool were hired after the announcement. Adoption has to live inside onboarding, permanently.

4. Deploy in weeks, not quarters. A learning system that takes eighteen months to design and launch is obsolete at kick-off, and it costs two holiday seasons. Modern configurable platforms go live in around twelve weeks. Speed is not only a cost argument. Content that reflects what is on the floor right now is the reason associates open the app twice.

5. Deliver in the format the workforce already uses. Associates adopt what resembles the rest of their phone. An app on a device they hold beats a portal they have to go find, and it beats an intranet built for head office.

6. Lead with video, and do not depend on the store network. Video shows correct selling, service and leadership behavior rather than describing it, and it replaces printed material that is wrong within a season. The critical engineering detail is that the video should be resident on the device rather than streamed. A module that stalls on store bandwidth teaches the associate that the tool is unreliable, and that lesson is very hard to unlearn.

7. Get associates out of the back room. Training that requires leaving the floor competes directly with selling and with payroll. Training that happens in between customers does not, and it converts a gap in traffic into an opening with the next person who walks in.

8. Keep modules short enough to finish. A retail shift is interrupted by design. Bite-sized modules survive it; a forty-minute course does not, and a half-finished course teaches nothing while still consuming the payroll.

9. Measure engagement per person, not completion in aggregate. Completion rates flatter everyone. What changes behavior is a manager able to see, by region, district, store and individual, who engaged and how well they scored, and to coach against that. This is also the only honest way to know whether adoption is real. Given that executives routinely misjudge what is running by an order of magnitude,¹ your own usage data is worth more than any vendor’s dashboard.

10. Choose a partner who has done it in stores. Retail rollouts fail in specific, boring ways: store bandwidth, device management, seasonal hiring waves, language coverage, and field managers who were not consulted. Experience with those failure modes is worth more than feature count.

“That is what the implementation team is for”

Every enterprise purchase arrives with a professional services engagement, and the reasonable expectation is that adoption is part of what you bought. It is worth being precise about which parts of it are and which are not.

An implementation team can configure, integrate and train the trainers. That is genuine work and it is well within scope. What it cannot do is decide which of your existing store channels this replaces, or persuade a district manager who was not consulted.

The engagement ends and turnover does not. An implementation runs for a defined period. Your hiring runs continuously, and the first practice above says adoption has to live inside onboarding. The day the implementation team leaves is the day the population that was trained begins to be replaced.

The vendor cannot fix an incentive. If the store scorecard does not reflect the behavior the tool enables, adoption will decay regardless of how well the software was configured, and no professional services statement of work covers changing your targets.

The fair version of the objection: a good implementation partner will tell you all of this, and the ones with retail experience will refuse to launch without it. That is exactly why the tenth practice is about experience rather than features. What the objection gets wrong is the assumption that adoption is a delivery phase. It is an operating responsibility, and it needs an owner with a name inside your business.

What failed adoption actually costs

The cost does not appear as a line item, which is why it survives so long.

• Productivity leaks continuously. The same Neochange research measured end-user productivity losses of roughly 17 percent, and concluded that at that level the economic value of user losses outweighed the entire IT spend. Technology-focused organizations lost 2.3 times more user productivity than user-focused ones, while user-focused organizations achieved 23 percent higher revenue per employee than their industry peers.³
• The business case gets measured with the wrong instrument. Forrester’s Andrew Bartel argued that firms trying to value new mobile and collaboration technology in terms of headcount or operating cost savings will not find the benefit, and therefore will not make the investment. The real benefit, he wrote, is reducing the probability of bad business outcomes and increasing the probability of good ones.⁴ A rollout judged only on payroll saved will always look marginal.
• Licenses get paid for and not used. Adoption is the difference between owning software and operating it, and only one of those changes what happens on the floor.

Costing your own stalled rollout

Industry loss figures are useful for framing and useless in a budget meeting. The version that works is your own, and it takes four steps.

1. Total annual cost of the rollout. License, implementation amortized, internal hours, content production and support.
2. Divide by active users, not licensed users. Active means someone who used it this month, not someone who was provisioned. This single division is usually the moment the room goes quiet.
3. Compare against your original per-user assumption. The ratio between the two is the adoption gap, expressed as the multiple you are actually paying per person served.
4. Add the value that did not arrive. Whatever outcome justified the purchase, measure it in your adopting stores against your non-adopting ones. That difference, applied to the non-adopters, is the return still sitting on the table.

The useful property of this calculation is that it points at a remedy rather than a write-off. Steps one to three are a cost you have already committed. Step four is recoverable, and recovering it is cheaper than any replacement system.

How to tell whether adoption is real

Four checks, all of which you can run on your own data rather than taking anyone’s word for it.

1. Repeat use, not first use. Launch-week logins prove a mandate was issued. Week-twelve logins prove the tool earned its place.
2. Depth by role. If engagement is strong among managers and thin among part-time associates, you have adopted a reporting tool rather than a training platform.
3. New-hire ramp. Compare the first ninety days of associates hired after launch against those hired before. Turnover means this cohort arrives constantly, which makes it the cleanest natural experiment you have.
4. Knowledge, then behavior. Quiz results tell you what landed. The KPI you expect it to move (units per transaction, average dollars per transaction, conversion) tells you whether it changed anything. Look at them in that order, because the second without the first is a guess.

Where to start

Take the last significant system you rolled out to stores and run step two of the costing above: total annual cost divided by monthly active users. It takes an afternoon and it will tell you whether adoption is a theoretical concern in your business or an expensive one.

Then name the person who owns adoption for the next rollout, before the contract is signed rather than after. The calculator will model what closing the adoption gap is worth across your own fleet.

References

1. WalkMe, The State of Digital Adoption 2025, Special AI Edition, February 25, 2025. Survey of 3,700 senior executives and employees plus proprietary usage data covering 1.5 million users across 2,400 enterprise applications. walkme.com
2. Association for Talent Development, 2025 State of the Industry, reporting 2024 data: 13.7 formal learning hours used per employee, down from 17.4 in 2023. td.org
3. Neochange with Oracle, 4th Annual IT Adoption Insight Report, as described by Neochange CEO Chris Dowse. Findings on discovery-phase adoption work, 17 percent end-user productivity loss, and user-focused versus technology-focused organizations.
4. Andrew Bartel, then Vice President and Principal Analyst serving CIOs, Forrester Research, on why headcount-based ROI models undervalue mobile and collaboration technology. forrester.com