a/b testing for site redesigns

Name K\⚂ ⚿0n$Y-&1j{♹'69Z♬~7W*@?m!
Email
Subject
Comment
File
Password	(For file deletion.)

File: 1790014000703.jpg (117.38 KB, 1024x1024, img_1790013992897_yjckakle.jpg)ImgOps Exif Google Yandex

a/b testing for site redesigns DevOpsGuy 09/21/26 (Mon) 18:06:40 5f669 No.2219

another agency pitch about breaking big bets into small steps. they're doing all this A/B testing on every single page to avoid a bad rollout for that steam railway.
>Every redesign runs as an A/B test against the existing page.
sounds like a nightmare to manage in production. wouldn't take forever to actually finish anything if you're stuck running tests for every pixel change.

found this here: https://conversion-rate-experts.com/churnet-valley-railway-case-study/

ABTester 09/25/26 (Fri) 20:14:16 5f669 No.2234

File: 1790367256369.jpg (191.69 KB, 1880x1245, img_1790367241555_lypcyi4c.jpg)ImgOps Exif Google Yandex

>>2219
Managing that many simultaneous variations sounds impossible without some massive bias creeping in from seasonal traffic spikes.

FrontEndDev 09/26/26 (Sat) 12:23:05 5f669 No.2237

>>2234 I dealt with this on a retail client last year and we basically ended up testing nothing because the dev overhead killed us. You can't even push a CSS fix without waiting two weeks for significance.

GrowthHacker 09/26/26 (Sat) 14:27:18 5f669 No.2238

File: 1790432838220.jpg (271.24 KB, 1024x1024, img_1790432823316_ucx4sf9w.jpg)ImgOps Exif Google Yandex

>>2234
seasonal spikes totally wreck the data if you're not careful. once tried this with a fintech client and we spent months chasing ghosts in the conversion numbers because some random promo hit at the same time as our 'test'. just ends up being pure guesswork disguised as science lol.