AI Adoption Programmes
Training is not adoption. One is a day out, the other changes the work.
What is AI adoption?
An AI adoption programme changes how work is actually done so the tools stay in use after training ends, measured by whether people are still using them months later rather than by how many attended.
Most AI training works on the day and fails within a month. People enjoy the session, return to their desks, and the work is exactly as it was, so the habit that would have replaced the old one never forms.
The reason is that training teaches a tool while adoption changes a task. If nobody has decided which specific piece of work is now done differently, there is nothing for the new skill to attach to.
That is why adoption programmes are measured on usage months later rather than on attendance, and why the answer is rarely more training.
Why we report the ninety day number
Duara runs adoption programmes from Nairobi for organisations that have already bought AI tools and are not seeing them used.
We measure adoption at ninety days and report the number even when it is poor. Training providers are usually paid on delivery and measured on satisfaction sheets, which is why so much training is enjoyable and changes nothing.
Who this is for
- You bought AI tools nobody uses
- Training was delivered and nothing changed
- A few enthusiasts use it and nobody else does
- You cannot tell whether the investment worked
Industries we serve
- Financial services
- Professional services
- Manufacturing
- Healthcare groups
- NGOs and development
- Government suppliers
Problems this solves
These describe tools that were bought, taught and then quietly abandoned.
- Licences paid for and unused. Seats are being billed monthly for people who opened the tool twice.
- Training that changed nothing. A session was delivered, feedback was positive, and the work is exactly as it was.
- Adoption concentrated in enthusiasts. Three people use it constantly and everybody else does not, which looks like partial success and is not.
- No measurement at all. Nobody knows the usage rate, so there is no basis for deciding whether to expand or stop.
What we deliver
A defined change to specific tasks, a small pilot proved before any rollout, usage measured at ninety days, and an honest decision about whether to expand or stop.
Specific tasks identified for changing
A pilot group proved before wider rollout
Usage measured at thirty and ninety days
Internal champions identified and supported
An honest expand or stop recommendation
Licence spend matched to actual use
Features
- Tasks changed, not tools taught. A named piece of work is done differently, because a skill with nothing to attach to does not survive the week.
- Ninety day measurement. Adoption is judged months later, not by attendance, and we report the figure whatever it says.
- Pilot before rollout. One team proves the change is real before the organisation is asked to adopt anything.
- Champions rather than mandates. People adopt what a respected colleague uses far more readily than what a policy instructs.
- Stopping is a legitimate outcome. Where a tool is not earning its licence cost, we recommend dropping it rather than prescribing more training.
Benefits
- Tools you pay for get used. Licence spend stops being a subscription to unused software.
- You find out what actually works. A measured pilot tells you which tasks genuinely benefit before you commit the whole organisation.
- Adoption spreads sideways. Change carried by respected colleagues moves faster and lasts longer than change announced from above.
- Bad investments get stopped. Knowing a tool is not being used is worth more than a satisfaction score saying the training went well.
Technologies we use
- Usage measurement. Reading real activity from your tool licences rather than relying on self reported use.
- Task analysis. Identifying the specific pieces of work where a tool genuinely saves time.
- Pilot design. Running a small group properly so the result means something before rollout.
- Champion support. Equipping the people others actually copy, rather than training everybody equally.
- Baseline and comparison. Recording how the task was done before, so the change can be judged.
The measurable question is not whether people can use the tool but whether they still are in ninety days. Almost all reported training success is measured before that point.
How the process works, step by step
An adoption programme usually runs eight to sixteen weeks, because the ninety day measurement is the point rather than an extra.
-
Step 1
Free adoption review
We look at what you have bought, what is actually being used and where the gap sits.
-
Step 2
Tasks chosen
Specific work is identified for changing, since teaching a tool without a task changes nothing.
-
Step 3
Pilot with one team
A small group runs the change properly, with the previous way recorded for comparison.
-
Step 4
Measured at thirty days
Early usage tells us whether the change held once the novelty passed.
-
Step 5
Ninety day decision
We report actual usage and recommend expanding, changing or stopping, including when stopping is right.
Quality assurance and how we work
The commitments attached to a programme whose success is easy to fake.
- We measure at ninety days and report it. Training providers are paid on delivery and measured on satisfaction sheets, which is why so much training is enjoyable and changes nothing.
- We recommend stopping where it is not working. A tool not earning its licence cost should be dropped, not given another round of training.
- Attendance is not a success measure. How many people came tells you nothing about whether the work changed.
- We pilot before asking for rollout. One team proving the change is real is worth more than an organisation wide programme built on an assumption.
- We use real usage data, not self reporting. People report using tools more than they do, and the licence data settles it.
Frequently asked questions about AI adoption
We already did training, why did nothing change?
Because training teaches a tool and adoption changes a task. If no specific work is now done differently, the new skill has nothing to attach to.
Would more training help?
Usually not, and we would say so. The answer is rarely more sessions, which is inconvenient for anybody selling training by the day.
How will we know adoption actually happened?
Real usage at ninety days, read from your licence data rather than self reported, and we report it whatever it says.
Why ninety days?
Because almost all reported training success is measured before that point, when enthusiasm rather than habit is still carrying it.
What if the tools are not being used at all?
We recommend stopping and reclaiming the licence spend. That is a legitimate outcome rather than a failure of the programme.
Should we roll out to everybody at once?
No. One team pilots the change properly first, because organisation wide rollouts built on assumptions are expensive to reverse.
What is a champion?
A respected colleague others copy. People adopt what a peer uses far more readily than what a policy instructs.
Can you make people use the tools?
No, and mandates produce compliance rather than adoption. Changing the work is what makes use rational.
How long does a programme run?
Eight to sixteen weeks, because the ninety day measurement is the point rather than an optional extra.
Do you provide the training too?
Yes, where it is needed, though the training is the smaller part and the task change is what makes it stick.
What about staff who fear replacement?
That is a real concern and pretending otherwise damages trust. It needs addressing directly rather than through reassurance nobody believes.
Can this work with tools we already bought?
That is the usual case. Most clients come to us with licences already purchased and unused.
Which tasks work best to start with?
High frequency, low risk ones where the time saving is obvious to the person doing it.
Do we need to buy anything else?
Frequently not, and we would rather you used what you have than added another subscription.
How do you measure usage?
From your tool licence data. People report using tools more than they do, and the data settles it.
Will you tell us the programme failed?
Yes. A ninety day figure showing no change is more useful to you than a satisfaction score saying the sessions went well.
Can it work across several departments?
Yes, though tasks differ by function and the change has to be specific to each rather than generic.
What if managers do not use it themselves?
Adoption stalls. That is worth raising early, since teams follow behaviour rather than instruction.
Is our usage data confidential?
Yes, and we would also advise against using it to discipline individuals, which turns adoption into avoidance.
What does an adoption programme cost?
Group size and duration determine it, quoted after the adoption review.
Glossary of terms
Adoption terms in plain language.
- Adoption
- Whether a tool is still in use months after training, as distinct from whether people can use it.
- Champion
- A respected colleague whose use of a tool persuades others more effectively than a policy.
- Pilot
- A small group running a change properly so the result is meaningful before rollout.
- Usage rate
- The proportion of licensed users genuinely using a tool, measured rather than reported.
- Shelfware
- Software paid for and not used, which is the normal outcome of training without adoption.
- Baseline
- How a task was done before the change, without which improvement cannot be judged.
Related services in AI Training
Talk to us about AI adoption
Check how many of your AI licences were used last week. That figure is the honest state of your adoption.
Start your project