Does 'Play' Foster Innovation?

And do you need to measure its ROI?

I went to a breakfast meeting last week where the subject for discussion was the likely impact of AI on agencies. And, without betraying any confidences, one person described themselves as being in the middle of feeling massively excited and massively terrified. Excited for the opportunities it provided to innovate and terrified of the existential threat to the agency business model itself.

But one thing that came up several times was the need for everyone within the agency to become experts in the use of AI: to develop efficiencies (and therefore reduce costs); to help their clients understand the value of AI on their businesses; and because it might result in innovations that the agency could productize and sell to their clients.

The consensus was that agency leaders needed to give their teams the permission to play with AI. As a product person, I was most interested in how playing with AI could result in products that could be monetised. And whether simply letting people play with AI would result in innovation.

Permission to Play is a good start, but it’s not a strategy

Getting people to play with AI as part of their job is a good idea. Expecting people to do it as an extra curricular thing might be tempting: you might consider that they are investing in their own careers and should be doing this out of professional curiosity, but you can’t assume this. So, mandating people to carve up some of their working day to spend learning about and playing with AI might make sense. Putting them on a course might be even more sensible.

It will build up internal ability and hopefully keep your team one step ahead of your clients, many of whom might be looking to you to teach them what they need to know. It should also reveal practical use cases, specifically in the area of automation and improved efficiencies.

But then, the interesting thing is whether it can also uncover potential product ideas. Could letting your team spend, say, 10% of their time thinking about the possibilities that AI brings spurn an idea that could be turned into a product? Maybe, but...

Going back to my product days, play without direction can easily end up becoming just a feel-good distraction. There has to be some expectation of what people are going to achieve by doing it. Play is a necessary condition but not a sufficient one. What do I mean by that? You need to give people the space to experiment, but without constraints and expectations, most play will stay in the sandbox. As we all know, ideas are worth pretty much nothing: it’s the execution that matters.

From Internal Efficiency to Product Led Growth

Playing with AI to improve internal efficiencies might give birth to product ideas. In the same way as products pre-AI did. ScreenCloud (my SaaS) came about when we wanted to show KPIs on a couple of public screens in our office and we couldn’t find an obvious and cost effective way to do it. We decided to hack something together using an AppleTV device, then discovered it was a harder technical solution than we first thought, which then made us think that others might want to find an answer to the same not-so-easy-to-solve problem - and the rest is history.

I suspect for many agencies this might also happen as they play around using AI to develop increased efficiencies. They might stumble on something really useful and the obvious next question will be “if this can work for us, why couldn’t it work for others, too?”

Showing clients the tools you have developed might demonstrate that there is a need you are meeting and your could productise what you have created for your clients and beyond.

ROI on Mucking About with AI?

How do you measure ROI and should you even try? My short answer is yes: but not straight away.

The problem with trying to measure ROI on curiosity and investigation is that it can stifle creativity and experimentation. Someone might think, I’ve got a cool idea but if I’m going to get shouted at because it didn’t generate enough revenue I’d rather just play around with it on my own time.

But ultimately, if your entire workforce is giving up a tenth of their time to play with AI, and there is no expectation, no direction, then people will drift off onto their own pet projects or rabbit holes that have no commercial value.

So, my suggestion would be to measure KPIs around ‘learning velocity’. Instead of applying a $ number to your time, especially early on, could you apply a learning velocity number? For example,

• Number of prototypes produced and demo’d to the wider company

• Number of tasks or even time saved creating efficiencies

• Prototypes that were spun into client-facing tools or services

In this way, there is an expectation that there is a commercial outcome to experimentation but it’s more about thinking about the practical output of an idea and, critically, getting it to a point where it’s actually tangible enough to show your peers.

AI Innovation Squad?

This is just an idea that I had and I’ve no idea how practical it would be. But, when I was still at ScreenCloud, I created a Growth Squad. We were made up of people from across the company: sales, marketing, customer success, product and engineering. We met for an hour once a week and our sole focus was on identifying small experiments we could run to make improvements to our growth trajectory. We figured that if we could improve, say, one part of the funnel by even a fraction of a percentage and then repeat that approach across every faction of our business, then those small improvements would together compound to make a significant impact.

What would happen if we did the same thing with AI innovation and included people from across the board. Maybe this team could curate ideas and bring the most promising ones to life: taking them out of the sandbox. Maybe they could articulate the biggest problems from the different operational functions within an agency and collectively see if they could work out how they could use AI to solve them.

The AI Innovation Squad could have a “Failure is Expected, Shipping is Celebrated” mantra. In this way, you could move quickly and efficiently, getting the most from your investment in ‘play’.

AI is clearly going to have a huge impact, but saying to your team, “here’s a subscription to Lovable, go and knock yourself out” and hoping that magically some brilliant product will come out of it, I would say, is missing an important step. A step that really, as a leader, that’s down to you to create. Good luck.