Marketing won the adoption race. Optimizely surveyed more than two thousand marketers and nearly half report artificial intelligence fully integrated into their daily work.
Here is the trophy. Seventy-five percent of them now spend at least three hours every week editing, fact-checking and repairing what it produced. Only four percent say the technology saves time across the whole process.
The time the machine saved is being spent supervising the machine.
I want to be careful here, because that is not an argument against the tools. It is an argument about how they were introduced.
The second shift, itemised
Nobody planned this work, so nobody costed it, so it does not appear anywhere except in people's evenings.
Checking that the figures are real. Removing the tells. Putting your actual voice back in. Confirming that the thing it stated with total confidence about a property, a tax, or a programme is true. Rewriting the opening, which is always the weakest part. Then the fair housing pass, which is not optional and which nobody scheduled.
Each individually is small. Together they are most of a day a week, and it landed on whoever was closest.
Everybody sounds like the machine now
The second cost is subtler and it compounds.
When everybody in a market uses the same tools with the same defaults, everybody's market update reads the same. The differences between three agents shrink to the differences between their photographs.
Which is a strange outcome for a technology bought to help you stand out.
The corner office is grading its own homework
Ask a leader whether the rollout worked and they will say yes, because they authorised it. Ask the person doing the second shift and you get a different answer, delivered carefully, because nobody wants to be the person who is bad at the new thing.
This is the failure I care about most, and it is entirely a management failure. If the person doing the fixing cannot say out loud that they are doing three hours a week of fixing, you do not have a technology problem. You have a company where inconvenient information does not travel upwards, and you will find out about that in a much more expensive way later.
The playbook
Measure the second shift. One fortnight, everybody writes down the minutes spent fixing output. Not to blame anybody, to price it. You cannot manage a cost you have never counted, and this one is invisible by construction.
Point it at the chores, not the thinking. Reformatting, resizing, summarising, twenty subject lines. Those genuinely save time. Anything requiring judgement about your market costs more to supervise than to write.
Write down what it is not allowed to do unsupervised. Anything with a number in it. Anything describing a neighbourhood or the people in it. Anything with your licence number underneath. That list takes ten minutes and prevents the only failures that actually matter.
Ask the person doing the work, privately. Then act on what they tell you, visibly, so that the next honest answer costs them nothing.
Any road up
Four percent say it saves time. That is not a verdict on the technology, it is a verdict on the implementation, and implementation is a thing you control.
Find out what your second shift actually costs. Then decide whether you are buying leverage or just relocating the work into somebody's evening.