That AI feedback that finally made me stop over-explaining

I had a client last month who straight up told me my AI chatbot responses were 'too wordy and sounded like a legal disclaimer.' They said it made them not trust the tool at all. So I went back and cut every output down to 3 sentences max, zero fluff. The engagement rate went up 40% after that change. Anybody else had a user call them out on something that actually improved their AI work?

3 comments

3 Comments

mason.brian22d ago

Hang on, is shorter always better though? I get that cutting fluff can help, but three sentences max every time sounds like you're just swapping one problem for another. Some things need a little more context to actually make sense, especially if you're dealing with complex questions. I've seen bots that are too short and they come off as robotic or dismissive, like they're brushing you off. Maybe the issue wasn't the length but that your original writing sounded like a legal disclaimer. Could be the tone was the real problem, not the word count.

angela68722d ago

Literally three sentences max is almost always the way to go though. I've tested this across like 20 different writing samples on my own and every single time the shorter version got better responses. The key is you have to pick your words carefully, not just cut stuff randomly. If your bot sounds robotic at 3 sentences that means you wrote the wrong 3 sentences, not that you needed more of them. Context is overrated honestly, most people skim anyway and they'd rather get a clear direct answer they can ask a followup on than read a whole explanation they might not need.

the_alice22d ago

Doesn't this whole thing remind you of how everyone talks in general now? Like even when I text my friends, anything longer than a few lines and they're like "TL;DR" at me.