Paradox

One of the main topics as it relates to AI is about the future of our jobs. I already wrote about my stance that AI is not coming for our jobs, so I’m not going to re-open that box per se. 

But there is this concept that I’ve learned about recently that supports my original point: That is that more AI leads to more efficiency, which leads to more demand, which means we all keep our jobs. The academic principle underlying this potential is called Jevons Paradox, and it’s been talked about a lot as it relates to AI. 

Without getting too nerdy, Jevons Paradox is an economics concept that basically says that efficiency doesn’t reduce consumption, but rather expands it.

As Drew explored last week, if AI lowers effort, opportunity costs will disappear and the work worth doing will actually increase. This pretty much fits the bill. 

What has stuck with me, though, is that Jevons Paradox and the counterintuitive nature of efficiency actually shows up a lot. Without trying to fit a square peg in a round hole, I was trying to come up with some of these paradoxes in the nature of ecommerce, and specifically with profit optimization. 

A few examples:

• If you improve CAC, you’ll spend more money on advertising. Shocking? Maybe not… as CAC improves, you find that you’re able to drive more sales with less spend. Which means that your incremental spend is now more efficient to a point where it finally makes sense to increase those marketing budgets! So yes, lower CAC means you’ll end up spending more (that’s still a very good thing)

• If you improve your conversion rates, they will end up getting worse again. Imagine that you’re a site optimizer and you figure out how to double conversion. Amazing! But that now makes your marketing dollars more efficient, and as with the example above, your VP of Marketing hikes spend and sends a bunch of lower quality traffic to the site. It’s still profitable, but your conversion rate is back to where it was before. Again, this is a great thing, let’s celebrate!

• If you invest in page speed, the weight of your pages will get heavier. The expected outcome is that faster page speeds will require lighter pages. But as pages load faster, you can now afford “heavier” elements that aid in conversion, like richer personalization, videos, and more. 

As we optimize, we often make calls on data at a snapshot in time, when we’ve reached a level of confidence. That confidence level, though, doesn’t take into account the knock-on effects of the metric moving, and the analysis of data in the near-term might miss some long-term realities. 

It doesn’t mean the analysis is wrong. It more means that it is correct in the moment and may be incomplete for the future (depending on how the business leverages those improvements for additional gains).

This is where experimentation meets Jevons Paradox. The paradox itself is that the metric not changing doesn’t mean there was a lack of progress.

In fact, a brand that takes advantage of the compounding effects of an experimentation program is likely to see healthy progress without much change to the metrics associated with it.