I think a lot of marketing teams approach AI the same way they approach buying software. See a demo, run a trial, perhaps buy it, then decide whether it works for the team. If it doesn’t, stop using it and try something else.
That makes sense when you’re choosing a CRM. If Salesforce doesn’t work for you, you can look at another CRM. But “we tried AI and it doesn’t work for us” is a much bigger conclusion to draw from one unsuccessful project.
It gets confusing because we use “AI” to describe different things. A product such as Jasper uses AI to help with a particular job, in this case creating content. You can evaluate that product for that job, just as you would other software. General AI use, through models and the tools that let you work with them, covers a much wider range of possibilities.
If one AI application doesn’t work for your team, there are plenty of other applications to consider. You might need a different tool, a different approach, or a different task to start with. You haven’t established that AI has nothing useful to offer your work.
AI ability can develop across the whole team
With a CRM, it makes sense to have one or two administrators who know the product deeply. Most people on the team use a small part of it for a small part of their day. They need to know how to do their jobs in the system, but we wouldn’t expect everyone to become a Salesforce administrator.
There’s also a fairly clear set of things an administrator needs to learn. Someone who knows the product can teach those things, and much of that expertise will still be useful at another company using the same software.
Some of the knowledge carries to other products too. A team that’s used Salesforce for years has a head start when moving to another CRM, even if learning the new product takes a few months. Getting good at an email marketing tool also means learning about email marketing.
AI opens up a different possibility. Everyone on the team could develop a high level of ability in applying AI to their work. The person planning campaigns, the person analysing results and the person creating content all have problems they could learn to tackle with AI.
For me, someone who’s good at using AI can look at a problem and quickly assess which parts AI could help with. They can apply that understanding to different kinds of work and find ways to do the work faster or more effectively.
Someone starting out often has a much earlier question: “I chat with ChatGPT, but beyond that I don’t know what to do with it. How do I use it to solve the problems I have?”
Learning the product’s features only takes that person so far. They also need experience connecting what AI can do with the work in front of them. I think of that as a career skill, something people can keep developing as their work and the technology change.
Useful processes take more than one attempt
Think about a marketing team using AI to analyse campaign results and suggest improvements. A process that gives the team useful recommendations probably hasn’t been built in one go. There have been several iterations, with the outputs improving as the team learns how to make the process work.
The team needs to judge whether the analysis is useful, work out what needs changing and try again. A polished explanation can still be weak. When we’re editing human work, we’re familiar with signs such as awkward writing, typos and grammar mistakes. AI can produce a weak paragraph without those signs, so learning to review AI-generated work can involve different checks.
That’s part of what the team learns while building the campaign analysis process. They’re learning how to use AI within their own work, including how to recognise when an output needs more attention. A course can help, but planning, building and improving the process gives them experience they won’t get from understanding the theory alone.
And the process can keep improving across campaigns. The team can use what they learned from the previous attempt instead of starting with the same request and making the same corrections each time.
Of course, “we’re building capability” can also become an excuse. We should be able to explain what we’ve learned and what we’re doing differently because of it. Spending time on one project, then going back to our old habits, doesn’t give us the same benefit as continuing to practise and build on that experience.
Start small and keep building on the experience
If you haven’t started, there’s more help available now than early users had: courses, tutorials, documentation and other people sharing what they’ve learned. You don’t need to go through every difficulty they encountered before you can start doing useful work.
For an individual or a small team, the first projects can use the model subscriptions or low-cost tools they already have. The main commitment is making time to try something, review the results and improve the approach.
A few things I’d focus on:
- Choose a task the team already does. Analysing campaign results is one example. Give someone responsibility for developing the process, with the people who use the analysis helping to judge whether the outputs are useful.
- Use the feedback on the next attempt. Discuss what worked, what didn’t, and what to change. Look for whether the recommendations become more useful over successive campaigns, rather than just counting how often people use AI.
- Write down what people know. Record what the team trusts AI to do, which tasks need a person, and what reviewers check. Get that knowledge out of people’s heads so it isn’t lost when someone leaves or a colleague takes over.
- Try the workflow with a different model. Once you’ve got a process working, see how quickly the team can adapt it to another model. The more experience the team gets doing this, the faster those adjustments can become.
Give people time to carry that learning into the next project. As they gain experience, they should get better at looking at their own work and seeing another problem AI could help them solve.