ai as a thinking partner is such a buzzword

Name
Email	81636
Subject
Comment
File
Password	(For file deletion.)

File: 1790963901008.jpg (116.39 KB, 1024x1024, img_1790963894763_4iawc9zg.jpg)ImgOps Exif Google Yandex

ai as a thinking partner is such a buzzword GrowthHacker 10/02/26 (Fri) 17:58:21 e1933 No.2240

the ai part is mostly just using it to summarize analytics outputs so you don't have to stare at spreadsheets all day lol. craig kistler from signet is basically saying the real work is still just judgment and deciding what's worth changing. feels like everyone is trying to ~~automate~~ outsource the actual thinking part of cro.

article: https://www.convert.com/blog/optimization/think-like-cro-pro-craig-kistler/

PixelPusher 10/05/26 (Mon) 19:51:05 e1933 No.2256

File: 1791229865117.jpg (145.33 KB, 1024x1024, img_1791229849398_rk50lekw.jpg)ImgOps Exif Google Yandex

>>2240
Kistler is right about the judgment part. I used ChatGPT last week to parse some heatmaps, but it couldn't tell me why users were ignoring the CTA, just that they were.
The tool doesn't know our brand voice or the actual user psychology.

ScriptKid 10/06/26 (Tue) 07:24:45 e1933 No.2259

File: 1791271485186.jpg (118.97 KB, 1024x1024, img_1791271467741_aywxt0fy.jpg)ImgOps Exif Google Yandex

>>2256
>it couldn't tell me why users were ignoring the CTA, just that they were

exactly. it basically just acts like a glorified mirror for what's already there. i tried feeding some user session recordings into 'claude' last month to see if it could spot patterns in friction. it flagged the same obvious stuff i saw in five seconds. it lacks that gut feeling you get when a layout just feels "off" or clunky. without that human intuition, you're just optimizing for numbers while ignoring the actual vibe of the site lmao.

Anonymous 10/06/26 (Tue) 21:46:45 e1933 No.2266

>>2240
i'm using it as a rubber duck for edge cases, not to replace my brain. if you feed it your actual hypothesis alongside the data, it's slightly better at finding contradictions in your logic. works okay for sanity checks on testing plans but don't expect it to find anything you didn't already point out.