“I love deadlines. I like the whooshing sound they make as they fly by.”

2025 proved to be a fascinating time to be alive for many in marketing. It felt a lot like being Mr. Burns trying to find the difference between Ketchup and Catsup, everything was the same and yet wholly different. We saw seismic level shifts in how we work with automation and busy work being done to a marginal level of competency by AI. What we work on changed too as we now had to sort out how to talk to a new kind of robot. Not only that but consumer behavior shifted too… what they trust changed, as did how they shopped. 

Further, we needed to contend with the ethical challenges of working with tools that plagiarize, replace labor, and use more fresh water than anyone can imagine so it makes you think twice before asking it something frivolous and fun. The sword had a second edge and it certainly felt like it cut. This added up to a pretty eventful year worth reflecting on as an industry. 

Beep Boop… Profit?

“Pivot!” Ross yelled while moving a couch up the stairs. 

Well, we certainly did. We had to make AI a cornerstone of the business in early 2025, hiring a new head of AI adoption. We kept track of the changes, joined Perplexity as a Fellow, and developed a whole suite a tools both internal and external for AI measurement, optimization, and workflow assistance. This was no small investment on DG’s part. For a business our size to lean so heavily into it was a gamble, but one that paid off. We stayed on the bleeding edge and I credit my team and our AI lead (Andrew) for keeping us honest and on track. This came with guardrails, though. We swore to never replace jobs with AI, never to expose client data to AI models without permission, and never to replace critical thinking and strategic work with AI. Thus far those guardrails hold and kept us on track. The path to ethical use of AI (as much as that is possible environmentally speaking) is a narrow one so we need to tread lightly.

Am I Talking to a Robot?

“Customer service agent!” he yelled at the phone automated assistant.

Email communication has been a big part of my life professionally, but lately you find yourself in a sort of uncanny valley with what you get back from people. Mediocre writing is so accessible and much quicker to produce, that you find yourself asking “am I even communicating with a real person” some of the time. You even the odd time get someone you forgot to remove the “should i write this differently for you” from ChatGPT in their copy and paste. On some level, you’re sad about this since it adds a layer between our social interactions. On another level, things do move more quickly, and the emails are better organized… even with the emojis and em dashes.

The hardest part of this isn’t that there are emails to answer, but we do notice scope creep here. Clients who get AI to write their emails include new questions they didn’t think of so instead of a single request, an email may have five or six. This isn’t necessarily a bad thing, but it is something we all need to stay on top of.

Below the Radar Deck

“Lies, damn lies, and statistics”

Welp, we certainly had it on easy mode before. Before we could draw a pretty straight line between action and revenue. We changed X, ranked better for Y, got more traffic to page Z and traffic that started on Z led to more revenue. Now it’s a bit of a jumble. First, AI Overviews on Google answers questions quicker and sooner especially long tail questions. This leads to people never touching the website at all, and having a new generative place to “rank” for links. Suddenly ranks matter a bit less than before, especially up the funnel. Second, we see AI tracking being extremely nascent. People don’t use keywords in their conversations with AI since… well… it’s conversational. Plus the answers are generative so there is no “ranking” but rather being cited. However, since citations are only tracked with AI tools that use sampled data usually from chrome plugins that scrape conversations with AI, we are left in a murky spot. We do still trust source/medium traffic to a degree, and we still have click data from Google Search Console, but we’re in the gray otherwise as an industry hazarding bets on samples.

Old School is the New School?

“Don’t count your chickens before they hatch”

We may just have to look back to look forward now. Once upon a time, when we used TV and Radio or other OOH advertising, we had to track runoff effects. It could take a full quarter or more in some cases to know how well a campaign performed. This required patience but also required intense scrutiny. If you were making a 7 or 8 figure advertising bet, and knew you didn’t know right away if it worked… you put way more work up front to give it its best shot. We need to do that here too. Similarly, the metrics we look at will need to change. Things aren’t so linear now that we have exposure goals rather than simply specific traffic targets. Bringing back adaptations of models like AFF or GRP based metrics could be in the cards going forward.

What’s Next?

“A bend in the road is not the end of the road… unless you fail to make the turn.”

Anyone who says they know what’s coming is a liar. But I do think we’re note quite done yet. Legislatively, there’s a lot to settle regarding AI. There’s anti-trust suits, erosion of traditional publications, and what would incentivize creation of new content to contend with. Further, we still wait on the launch of ChatGPT advertising which seems inevitable and fast approaching, as well as the far off true agentic models. I’d bet in the near term, we will see consumer behavior continue to shift to conversational, transactional changes shift, and LLMs adapting to less spam influenced results. Influencer marketing may continue to find a home too as referral sales take more heed. Email marketing may slowly die too if everyone uses AI to produce them, they’ll become noise soon enough. However, in the long term… who knows. Will all shopping be done inside LLMs? Will we hit a water crisis sooner than we expect? Will the economy continue to be held up by speculative value of AI rather than revenue collapse? What we do know is, like a good infielder in baseball, stay on your toes, not your heels.

If you want to join this ride with us, let’s chat.