What I Told Clients in 2023 That Turned Out Wrong (and Right)

Written in 2026, marking my own 2023 homework, with the answer key that only three years could provide.

At the end of 2023 I sat down and wrote out the main things I’d told clients that year, so that future me could grade them honestly. It felt useful and a little uncomfortable, which is usually how you know something’s worth doing. Now future me has arrived, three years on, with enough hindsight to actually score it. So here’s the report card, misses first, because the misses are where I actually learned something.

I’m grading each call from memory and against what actually happened. No fabricated metrics, just an honest verdict.

The ones I got wrong

“Volume still works if the content is decent.” Wrong. Early in the year I told a couple of clients that publishing a lot of solid, unremarkable content was still a fine strategy, that Google would keep rewarding coverage. Then the Helpful Content Update arrived in September and flattened exactly that kind of content, including on a site of my own. I wasn’t just wrong in theory, I paid for this one personally. Grade: a hard fail, and the most instructive one I have.

“AI answers in search will roll out slowly and stay marginal for years.” Half wrong. After Google I/O I told people not to panic, that the AI box would stay a limited experiment for a long time because Google couldn’t afford to break the click. I was right that they’d hesitate and wrong about the nerve. They leaned in faster and harder than I predicted. I under-read how frightened Google was, and fear moves faster than caution. Grade: partial credit, correct instinct, wrong magnitude.

“Don’t bother much with Reddit and forums as a channel.” Wrong, in a way I didn’t see coming. I told clients community platforms weren’t worth serious attention. Then the results filled up with them and the ground I’d waved off became prime real estate. Grade: a miss, though at least an interesting one.

The ones I got right

“Do not replace your writers with a prompt.” Right, and it’s the call I’m proudest of. All year, as clients asked whether ChatGPT could just write it, I said use the tool for scaffolding and never for the part that’s supposed to be you. The clients who listened came through the year’s chaos intact. The ones who generated everything got hit. Grade: pass, and it held up better than I dared hope.

“Export your Universal Analytics data before July, you will want the history.” Right, dully so. Boring advice, easy to ignore, correct. The clients who exported kept their past. The ones who didn’t lost it forever. Grade: a quiet pass, the kind nobody thanks you for.

“The safe place to stand is being genuinely known and first-hand.” Right, and it kept getting more right. I told clients that trusted, experienced, actually-been-there content was the durable bet. Every wave of the year, and the two years since, rewarded exactly that. Grade: pass, and if anything I understated it.

What the scores add up to

The pattern in my own report card is not flattering and it is useful. Every wrong call was me under-betting on change. Volume would keep working, AI would stay marginal, forums didn’t matter. I kept assuming things would move slower and gentler than they did. Every right call was some version of make real things and protect what’s genuinely yours.

So the correction I’m carrying out of 2023 is simple. When I catch myself predicting that a shift will be slow and mild, I now assume I’m wrong by default, because in 2023 I was, every time. And when I’m tempted to chase a volume shortcut, I remember a chart falling through September, on a site I built, and I don’t. It cost me some traffic and some pride to learn that, and I think it was cheap at the price.