IDSPIRATION STUDIO · GUIDE · AI & DIGITAL ADOPTION
The adoption audit.
A one-afternoon check on whether your team actually uses the tools you paid for, and what’s getting in the way.
THE SHORT ANSWER
Seat counts and logins tell you who can use a tool, not whether it has changed how work gets done. Pick one tool and one real task. Talk to five people who should be using it. Score four signals: use, habit, trust and value. The lowest score is where to start, and it’s usually a human problem, not a technical one.
01
Access isn’t adoption.
Most rollouts are measured by what’s easiest to count: licenses bought, accounts activated, people trained. Those numbers can all look healthy while the actual work carries on exactly as before.
Adoption means a tool has become part of how someone does their job, without a reminder. That’s a change in behavior, and behavior doesn’t show up in a license report. You have to ask.
02
Pick one tool and one task.
“Are people using AI?” is too broad to answer. Narrow it to a single tool and a single, recurring task it was meant to help with. For example:
- Drafting a first version of campaign briefs with your AI assistant
- Logging customer conversations in the CRM after each call
- Pulling the weekly performance report from the new dashboard
If you can’t name the task a tool was bought for, that’s your first finding.
03
Ask five people, not the dashboard.
Choose five people who do that task regularly. Mix roles and seniority, and include at least one skeptic. Fifteen minutes each is enough. Keep it a conversation, not a test, and make clear there are no wrong answers.
- The last time you did this task, how did you do it? Walk me through it.
- Did you use the tool? Why, or why not?
- When it works, what does it save you or improve?
- When do you avoid it, double-check it or redo its work?
- Does anyone around you use it well? What do they do differently?
The first question matters most. Asking people to describe what they actually did gets closer to the truth than asking whether they use the tool.
04
Score four signals.
After the conversations, score each signal from 0 to 2 for each person, then look at the pattern across all five.
| Signal | The question | 0 | 1 | 2 |
|---|---|---|---|---|
| Use | Did they use it for this task in the last two weeks? | No | Once or twice | Most times |
| Habit | Is it their default, or a detour? | Forgotten | When reminded | Without thinking |
| Trust | Do they rely on the output? | Redo it | Check everything | Check what matters |
| Value | Can they name a real benefit? | None | Vague | Specific |
Low use with high value means people know it helps but something blocks them. High use with low value means they’re complying, not adopting. Both point to different fixes.
05
Name the friction.
When adoption stalls, the cause is rarely the technology. Listen for these patterns in what people told you:
| What you hear | What’s really happening | A first fix |
|---|---|---|
| “I’m not sure when to use it.” | No clear link to their real work | Show two examples from their own tasks, not generic demos |
| “I don’t want to get it wrong.” | Low psychological safety | Create a low-stakes space to practice and share mistakes |
| “It’s quicker to just do it myself.” | The tool adds steps to the workflow | Redesign the task around the tool, or drop the tool for this task |
| “Nobody else really uses it.” | No visible social proof | Make good use visible: short demos, shared wins |
| “I can’t trust what it gives me.” | Unclear human responsibility | Agree what a person must always check, and what they don’t need to |
FROM EXPERIENCE: START WITH UNDERSTANDING
In a previous role, the official approach to AI was to give people the tools and hope for the best. So Steven started running AI literacy workshops from the basics up: how large language models actually work, what they’re good and bad at, and how to work with them on everyday tasks.
The aim wasn’t to train people on features. It was to make the tools feel understandable, so people could judge for themselves when to use them. One session went further, with teams playing the AI themselves to see how it chooses sources.
What followed mattered more. People from different teams started coming forward with their own ideas, and that grew into an AI advocate program: colleagues building real use cases and sharing them with their own teams. That’s the fourth row of the table in action. Adoption spreads fastest when it comes from someone you work with.
06
Read the results honestly.
Five conversations are a signal, not a statistic. Look for patterns that repeat across people, not percentages.
ILLUSTRATIVE EXAMPLE, NOT CLIENT DATA
A marketing team scores high on value (“it saves me an hour on every brief”) but low on habit (“I forget it exists until someone mentions it”). The tool works. The workflow doesn’t include it. The fix isn’t more training; it’s adding the tool to the brief template itself.
07
Fix one thing, then check again.
Pick the lowest signal and the most common friction behind it. Make one change, give it 30 days, then repeat the same five conversations. If the scores move, keep going. If they don’t, the friction was somewhere else, and that’s useful to know too.
Resist the urge to roll out another tool before this one works. Each new tool that goes unused makes the next one harder to introduce.
About this guide: written by Steven Klimek from experience leading digital and customer change in large organizations. The scoring is a practical starting point, not a validated research instrument.