Everyone is concerned about AI taking over everything. It will take over jobs, and we will all lose our ability to think. Problem-solving will be a thing of the past. Just let the AI take care of it. I call BS! I want to present a case study of how good old-fashioned ingenuity solved an issue with Google Ads, and I am convinced that AI may not have figured it out on its own.

First, the Deluge and Then the Crickets

One of our clients spends a good amount of money on Google Ads, both in search ads and local service ads. When we recently had a heat wave, which kicks their business into overdrive as they are an HVAC company, they had a deluge of calls as expected. But something was off. 

We noticed that the leads were coming in, but the volume didn’t seem right. We went around and around trying to come up with reasons. It was a gut feeling because all the comparative numbers looked good. The client was busy, and they generated a ton of calls. Then came the crickets.

After the weather dropped below the temperature of the sun, the calls dropped off. Initially, we thought it was one of those situations where it was so busy that once it slowed down, it felt really slow. Kinda like when you are driving 90 mph up the highway, and you have to slow down to 65. It feels like you are crawling even though you are still moving rather quickly. (not that I would know about this)

Again, the 30-day looked good, and the month-over-month as well as the year-over-year looked good. That was when we started peeling back the layers in the onion. There was a drop-off, no doubt, but the daily average of calls seemed lower than it should have been. 

Suddenly the Light Bulb Went On

What we realized was that we weren’t comparing heatwave to heatwave. The previous year’s July was a stellar month. The client had a ton of calls. We compared Julys, and there was a drop, but not a precipitous one until after the heatwave was over.  Then the light bulb went on: the heatwave last year was late June, not early July as it was this year. We were comparing the wrong things!

Once we compared the heatwaves, we noticed that there was a 30% dropoff. Then we looked at the budget increase this year versus last year, and it was only a 15% increase. Which helped, but it didn’t match the success we had the previous year. I credit my partner’s “Spidey Sense,” as he kept saying it didn’t feel right even though the numbers weren’t terribly off. So we kept looking. 

Unless You Know What You Are Looking for, You’re Not Going to Find It

Back to how AI wouldn’t have come up with the same conclusion. If we had gone to the AI of choice to evaluate the campaign and figure out why our numbers didn’t match year over year, it wouldn’t have cited a ton of empirical data, reported on what others have reported, and come up with a number for a budget increase. There’s a chance that the number would have been the same, but I believe that it would have been a scientific wild-ass guess. 

With the data we found, we calculated the difference in impressions between this heat wave and the last heat wave. The drop-off was 30%, and when we calculated a 30% increase in conversions we got this year, the number of conversions between the two years was almost the same. Within three conversions! Now we had a real number based on real data that we could go back to the client and explain that we needed to increase the budget. If the client asks how we came up with the number, we could show our work. 

Don’t get me wrong, I use AI in many different ways and continue to explore how to use AI better. (Full disclosure: I used AI to refine the headline of this article). I love the expediency with which AI can do a task; however, AI is only as smart as the instructions it is given. Garbage in, garbage out. I also am convinced that, at least for the foreseeable future, human ingenuity will still outthink AI.