When a Recommendation Fails

Push something hard enough and often enough and it will eventually land badly on somebody whose opinion matters. This is not a risk to be minimised. It is the operating cost of a format that refuses to hedge, and the way it gets handled determines whether the format is worth anything afterwards.

The first mistake is retroactive hedging. Well, I did say it was not for everybody, when no such thing was said, is the fastest available way to destroy the credibility of every previous recommendation at once. If the sentence was unqualified, it stays unqualified. The claim was made and the claim missed.

The second mistake is escalation, which is more tempting and more damaging. This is the move where the failure gets reassigned to the listener: they did not stay with it long enough, they were in the wrong mood, they would need more background. Sometimes all of that is even true, and it does not matter, because the recommendation is judged on the encounter it actually produced.

The useful response is duller. Find out where it went wrong, precisely, and treat that as information about the recommendation rather than about the person. Usually the answer is one of three things: the commitment was larger than stated, the entry point was rougher than remembered, or the reason given was not the actual reason it worked. That third one is much the most common.

That last case is worth dwelling on, because it is the failure mode that repeats. A thing is loved for a complicated reason and recommended for a simple one, because the simple one is easier to say on a deadline. The listener arrives looking for the simple thing, does not find it, and leaves. Nothing was wrong with the object, and everything was wrong with the sentence attached to it.

It follows that a failed recommendation should change the next attempt at the same thing rather than retiring it. The object is probably fine. The packaging needs rebuilding, and the failure is the only source of information about what to rebuild. Retiring the recommendation entirely is the emotionally comfortable option and throws away the one useful thing produced by the whole episode.

There is also the plain case where the thing was simply not as good as it seemed at the time. Saying so, out loud, in the format that pushed it, is the most valuable thing the format can do. It costs one item and buys back the assumption that the enthusiasm is real, which is the only asset the list has.

The rate matters more than any individual miss. A list where everything lands was recommending safe and obvious things. A list where nothing lands is not paying attention to its audience. Somewhere in between, with the misses acknowledged plainly and the reasons examined, is a format doing its job.