List7 min readPublished

Why your team stopped using the AI tools you rolled out (and 5 fixes that work)

A commercial team working together around a table

The short answer

Most teams stop using new AI tools because nobody is appointed custodian of the new way of working once the launch is over. Managers haven't been shown how to judge the quality of output or coach improvement, and people aren't given time to practise on real work, so frustrations build and teams revert to what worked before. None of this needs a new tool.

Key takeaways

  • In 2025, regular AI use among frontline employees had stalled at 51%, while more than three quarters of leaders and managers were using it several times a week (BCG, 2025).
  • Manager support makes the biggest difference. Employees whose manager supports AI use are 7.4 times as likely to say it helps them do their best work (Gallup, reported by HR Dive).
  • The five fixes are about ownership, coaching, measuring quality, better briefs and time to practise.

You've bought the licences, run the launch session and sent round the prompt guide. For the first few weeks there was a real buzz, with people sharing what they'd tried in the team chat and a few surprisingly good first drafts. Employees enjoy the first flush of success. Then things gradually trickle back to how they were before. A couple of people still use it every day, most have gone back to the way they've always worked, and nobody's entirely sure whether the investment is paying off. New models are released and everything stays as it was.

If that sounds familiar, you're not alone. BCG's 2025 AI at Work survey found regular use among frontline employees had stalled at 51% (BCG, 2025). Their 2026 survey shows use has grown since, with 74% of frontline employees now describing themselves as AI users, but only a third say leadership's communication about AI is clear (BCG, 2026).

In my experience, the reasons have much more to do with what happens after launch day than with which tool you chose. These are the five I see most often in commercial teams, with a fix for each that you can start on this month.

01

Nobody owns the new way of working

Launches tend to be owned by whoever bought the tool, often someone in ops or IT, or a forward-thinking member of the team who spotted the opportunity. Once the training session is done, their part feels finished. Nobody in the sales, marketing or BD leadership picks it up as part of how the team runs week to week, so it slips off the agenda. When things get busy, people fall back on the way they know works.

The Fix: Give each team a named owner, ideally the line manager, and make AI-assisted work a standing five-minute item in the weekly team meeting. What did we try, what worked, and what should we stop doing? It keeps it visible without turning it into another project. Ask managers how they are checking in on AI use, opportunities and outcomes in their 1-to-1s with their team. If they keep asking the same questions, they'll revert back to the same way of doing things.

02

Managers haven't been shown how to coach it

Most managers I speak to are using AI themselves, but very few have been shown how to coach someone else who is. They can see the output, but they can't always tell whether a proposal or a piece of outreach is good because someone thought hard about the customer, or because the tool produced something that looks polished. So they either leave it alone or they count how much of it there is. Managers need to look at how human judgement is being integrated into the AI tools, output and impact together. What did the output tell them and how did they build on it? Not, did they simply press send. It's that gap in between where the best opportunity to coach is being missed.

This matters more than most people expect. Gallup's 2026 State of the Global Workplace research found that employees whose manager supports AI use are 7.4 times as likely to say it helps them do their best work, and Gallup's CEO described a manager actively championing it as the strongest predictor of adoption after technical integration (Gallup, reported by HR Dive). BCG found the share of employees who feel positive about AI rises from 15% to 55% with strong leadership support, however, only around a quarter of frontline employees say they get that support (BCG, 2025).

The Fix: Spend one session with your managers reviewing real AI-assisted work together: a proposal, an outreach sequence, a set of call notes. Agree what made the good ones good. That conversation does more for their confidence than another tool demo, and it gives them the language to coach their own teams.

Leadership support changes how people feel about AI

Share of employees who feel positive about generative AI

Source: (BCG, AI at Work 2025)

From experience

One founder I worked with wanted to bring AI agents into their business development process. Rather than switching everything on at once, we picked a handful of tasks where it would save real time, and I trained them live, on AI-assisted working, as we went. They gradually felt empowered to explore further themselves, confident enough to try, which is the behaviour you need that ensures ongoing adoption.

03

The numbers you track reward volume

When AI makes it easy to send more, the easiest thing to report on is how much more, whether that's emails sent, posts published or sequences launched. Activity goes up, the dashboard looks healthy, and conversion stays exactly where it was. People soon work out that what's being measured is quantity, and the ones who were using the tool thoughtfully stop seeing any reason to put in the extra effort. In so many cases, the greatest revenue opportunities lie in improving conversion. Almost every founder says they need more leads. Many actually don't. They need to get smarter at converting what's already flowing into the system.

The Fix: Pick one quality measure to sit alongside your activity numbers and track it for AI-assisted work against how things were before. It could be reply rate on outreach, meetings booked from first calls or proposal win rate, whichever fits the team best. If you can't yet tell which pieces of work were AI-assisted, start there.

04

The brief going in is too generic

This is the one I see most often, and I've written about it before as the Skills Gap. Someone asks the tool for a follow-up email to a Head of Marketing, gets something bland back, and decides the tool isn't much use. They get frustrated with the time it takes to rewrite from scratch. The tool can only work with what it's given, and a job title tells it very little about that person's actual situation.

Good briefing is a skill most of us were never taught, because until recently we were briefing colleagues who could fill in the gaps for us. This holds true for content creation generally, so applies across sales and marketing. Many have never studied how to break down the elements of a brief, the structure, a writing style, and what's needed in terms of explicit instructions, so the prompts end up being too thin.

The Fix: Give the team a simple three-part brief to use every time: who the customer is and what's going on for them right now, the goal or constraint that matters most, and what a good result would look like. What good looks like isn't just a list of instructions. Start here first with multiple examples and reference files and analyse what makes them good, using the tools, building the skills and memory as you go. It takes longer, and the difference in what comes back is obvious straight away.

A three-part brief to use every time

Part 1The customer's situation

Who they are and what's going on for them right now, beyond their job title.

Part 2The goal or constraint

The one thing that matters most for this piece of work.

Part 3What good looks like

Examples and reference files to analyse, so the tool can see the standard you want.

05

Nobody was given time to get good at it

Learning to work well with AI takes practice, and practice takes time that most commercial teams don't have spare. If the only training was a launch session, people are being asked to learn a new way of working on top of a full target, often in front of customers. Under that kind of pressure, most of us will choose the approach we already trust. Once you've noticed more volume isn't getting you what you want, and you've pivoted to quality, 'going slow to go faster' becomes good practice and conversion becomes the mantra. I can't stress this enough. Focusing on this almost always pays off.

BCG found regular use is sharply higher among employees who've had at least five hours of training and access to in-person training and coaching, but only a third of employees say they've been properly trained (BCG, 2025).

The Fix: Protect at least 30 minutes a week for the first six weeks so the team can practise together on live work: a real account, a real proposal, a real follow-up. Short, regular and practical sessions (spaced repetition) work far better than a single long training day, because without review we forget roughly 75% of what we learn within a day or two, based on Hermann Ebbinghaus's forgetting curve (Auburn University, The Forgetting Curve).

A quick self-check

Score each statement for your team as it is today: not yet, sometimes, or yes, consistently.

Your answers stay on this page. Nothing is saved or sent anywhere.

1We have a deliberate approach to using AI, rather than individuals experimenting on their own.
2Managers can tell the difference between strong and weak AI-assisted work, and actively coach their teams on it.
3We track whether AI-assisted work converts better or worse than what we did before.
4Our outreach and proposals read as though they were written for that particular customer.
5There's a consistent review step before AI-assisted work goes out to a customer.

Where to start

You don't need to tackle all five at once. Pick the one that made you wince a little as you read it and start there this month. If it would help to talk through which one is holding your team back, that's exactly what the first call is for.

Sources

Where this fits in WISE: this is part of Embedded adoption. See how the full WISE system works.

Questions people also ask

How long does it take a team to get comfortable using AI tools?

It varies with the team and the tool, but short, regular practice over several weeks works far better than a single training day. BCG found regular use was sharply higher among people who'd had at least five hours of training and access to coaching.

Should AI adoption be led by the commercial team or the people team?

Both need to be involved. The commercial leader owns how the work gets done and what good looks like, and the people team helps with skills, changing roles and how managers are developed. When only one side owns it, it tends to stall.

How can you tell if AI tools are being used well?

Look at quality alongside usage. Compare reply rates, meetings booked or win rates for AI-assisted work with what the team achieved before, and review a handful of real examples with your managers each month.

What do you do when AI tools aren't working for your team?

Start by finding out why, because it's rarely the tool itself. Check who owns the new way of working, whether managers know how to coach it and whether your measures reward volume over quality. If the problem runs wider, the five gaps AI exposes in commercial teams will help you spot where it's breaking down.

Sarah L Morgan

About the author

Sarah L Morgan has led sales, marketing and BD teams for more than 20 years, from FTSE 100 media to £4m professional services firms. She now helps £5m to £30m businesses fix the people and process gaps AI exposes.

More about Sarah · LinkedIn

Seeing this in your team?

30 minutes on what's getting stuck. No pitch deck, no prep needed.

Book a 30-minute call

Keep reading

← All insights