• by Maarten Albarda , Featured Contributor, Yesterday

Whenever a massive technological shift hits, everyone starts chasing the elusive shiny new tool. Until someone starts asking questions like “is it working? Is it doing us any good?”

Oh ye of little faith. Of COURSE we need to bet all our money on AI -- or be vilified, just like those questioning Google Glass, Second Life or its big brother Metaverse (R.I.P.). To be honest, this is actually a really good question. Companies and investors are throwing billions into AI, agentic, LLMs and more to feed the hungry beast. Agency networks are rebuilding and reframing themselves. Smart CMOs are asking those pesky questions and are beginning to explore metrics that asses exactly how much money the machine just saved them. Or not…

Sadly, the perfect scorecard doesn't exist yet. AI is still in its absolute infancy. And it’s completely fine that we don't have every single micro-interaction perfectly tracked and measured right now. The tech is moving too fast for rigid KPIs to keep up.

But... and this is a massive "but"... that doesn't mean you just throw your hands up, toss a bunch of beta tools at your team, and say, "Go explore!" You can't just wing it. Marketing budgets are too tight for operational chaos. You need structure, even if the math is a little fuzzy.

A lot of confusion right now stems from the platforms themselves. Google, Meta, Amazon, and Walmart are all aggressively pushing out their own end-to-end AI solutions. They promise to set your strategy, generate the creative, place the media, and optimize it in real time. If you are a small to mid-sized advertiser, this is a dream come true. You can basically outsource your entire funnel (which is exactly what they want).

But if you are a blue-chip brand? For marketers in this category, these solutions are a massive operational trap. All of them are walled gardens. Google’s AI is optimizing for Google's ecosystem. Amazon’s AI only cares about Amazon. If you let each platform's AI run autonomously without oversight, your brand identity will fracture into five completely different, conflicting personalities by next Tuesday.

To manage this, you must take or retain control, whether through a heavily beefed-up in-house team or a highly competent agency partner. Someone has to coordinate the chaos, stitch the fragmented data together, and ensure your core messaging actually looks and sounds like your brand across the entire ecosystem.

So, how do we know which AI initiatives to expand and which to kill? Look through a broader strategic lens. Sure, track the time saved on mundane tasks. But more importantly, track where that time goes. If your agency uses AI to cut production time by 30%, did they reinvest those hours into bolder, high-level strategy? Or did they just pocket the margin? Or did you consciously add it back to the bottom line?  If you aren't measuring what happens to the freed-up human capital, you can’t make an informed decision about where the gains go.

Equally, if an AI tool requires too much manual human babysitting to maintain brand consistency, it’s not efficient. Kill it.

For now, stop obsessing over perfect efficiency scorecards. Accept that the tracking will be a bit messy for a while. Focus your energy on building the human operational structures needed to manage these walled gardens, and ensure the AI is actually giving you a strategic advantage, not just a cheaper way to make noise.