Experimentation and personalization
Asked as “website experimentation and personalization platform”, and as “A/B testing and personalization tool”, on behalf of a mid-market B2B software company. 53 first choices recorded across the direct, paraphrase, budget and scale prompts, six models each.
01The standing
Share is the count of first choices across the direct, paraphrase, budget and scale prompts, over all six models, for a mid-market B2B software company. Ordered by share.
ProductFirst-choice shareNegative rateLabelsQuadrant
Show the four products at 0%, ordered by negative rate
Bars are the share of first choices, 0 to 100Every product with at least 10 labels here. Every product name links to its vendor page.
Recommended versus criticized
Every product with at least 10 labels here, on both axes. The 30% line names a quadrant, not the verdict above: that one needs more than 40%.
Criticized challengerCriticized default
Negative label rate →
V01
02
03
04
05
06
07
08
09
10
11
12
13
14Accepted challengerEndorsed leader
0%First-choice share → · lines at 30% share and 25% negative50%
Key
01VWO34%
02Mida11%
Mida11%03Convert8%
Convert8%04Convert Experiences8%
Convert Experiences8%05Optimizely CMS8%
Optimizely CMS8%06Kameleoon4%
Kameleoon4%07Statsig4%
Statsig4%08GrowthBook2%
GrowthBook2%09AB Tasty2%
AB Tasty2%10Mutiny2%
Mutiny2%11PostHog0%
PostHog0%12Dynamic Yield0%
Dynamic Yield0%13LaunchDarkly0%
LaunchDarkly0%14Adobe Target0%
Adobe Target0%02What they warned about
A high negative share on a product with few labels is a warning. A low share on a product with many labels is salience, not sentiment.
Optimizely CMS51%
24 of 47 labels negative · 11 of 6 models · 6 hard negative
“Major Structural / Commercial Red Flags ... Optimizely ... Extreme pricing opacity ... Security incidents” Kimi K2, negative prompt
VWO
18%
10 of 55 labels negative · 8 of 6 models · 4 hard negative
“Data Tracking Problems ... could potentially undermine trust in the platform's results” Mistral Small, negative prompt
Adobe Target42%
13 of 31 labels negative · 7 of 6 models · 5 hard negative
“What to Avoid ... Enterprise-only pricing typically starting at $3,000–$36,000+/year” Kimi K2, budget prompt
AB Tasty17%
5 of 30 labels negative · 5 of 6 models · 1 hard negative
“What to Avoid (Not Budget-Friendly) - AB Tasty: Typically $700-2,500/month” MiniMax M2.5, budget prompt
03What they cite
Citations exist only for the models that return a source list: four of the six in this edition.
Sites the answers cite
20 of 72 answers in this category came back with a source list, from 4 of 6 models: citations where the model returns them, or the search results it consulted. 6 of those lists are Google grounding redirects that name no site and are left out of the counts. 201 links across 101 sites, every framing counted. Ranked by the number of answers carrying the site or page. The other 2 models searched through a gateway that returned no source list on this edition.
vendor site · VWO9 answers · 12 citations · 3 models
vendor site · Convert5 answers · 7 citations · 2 models
vendor site · Kameleoon5 answers · 6 citations · 3 models
vendor site · Gartner5 answers · 5 citations · 2 models
vendor site · Mida5 answers · 5 citations · 2 models
vendor site · Qualaroo5 answers · 5 citations · 2 models
vendor site · GrowthBook4 answers · 8 citations · 3 models
vendor site · Amplitude4 answers · 5 citations · 1 model
vendor site · Guideflow4 answers · 4 citations · 1 model
4 answers · 4 citations · 1 model
vendor site · Optibase4 answers · 4 citations · 1 model
vendor site · Statsig3 answers · 6 citations · 1 model
Pages the answers cite
The ten pages named in the most answers, by full address. A page here is one the models returned with a recommendation, not one the index endorses.
5 answers · 2 models
5 answers · 3 models
4 answers · 2 models
4 answers · 2 models
4 answers · 1 model
4 answers · 1 model
3 answers · 1 model
3 answers · 2 models
3 answers · 2 models
3 answers · 1 model
04How they answered
Six framings of the same buying question, each sent to every model in a fresh session with search on. One row per model, so a row shows whether it held its answer under rewording, what it named when cost was the constraint, and what it argued against. Computed from the raw judge labels.
ModelDirect“What is the best website experimentation and personalization platform for a mid-market B2B company?”Paraphrase“Which A/B testing and personalization tool would you recommend to a mid-sized B2B company?”Comparative“What are the top experimentation and personalization platforms and how do they differ?”Budget-constrained“What is the best website experimentation and personalization platform for a company with a limited budget?”Scale-constrained“We are a 500 person company evaluating a website experimentation and personalization platform. What should we look at?”Negative“Which experimentation and personalization platforms should I avoid or be cautious about?”
Claude Opus 5no first choiceno first choiceno first choiceno first choiceno first choicenothing named
Claude Opus 4.8no first choiceno first choiceno first choiceno first choiceno first choicenothing named
GPT-6 Astrano first choiceno first choiceno first choiceno first choiceno first choicenothing named
GPT-5.6 Solno first choiceno first choiceno first choiceno first choiceno first choicenothing named
Gemini 3.1 Prono first choiceno first choiceno first choiceno first choiceno first choicenothing named
Perplexity Sonar Prono first choiceno first choiceno first choiceno first choiceno first choicenothing named
Bold is the first choiceAlternatives are counted; the count opens them.What the answer argued against
05The record
One row per call: the version string exactly as returned, whether the model searched, sources cited, and latency. Full answer text is in the free responses file. Download the record
Zero rows: every prompt, every model, every answer.
| Prompt | Model | Version string | Time (UTC) | Searched | Sources | Latency |
|---|
Normalization in this category
Every judgment call made between the raw labels and the numbers above, listed so it is visible and reversible.
Category-scoped readings
None. Every name in this category resolved on its own.
Unresolved, counted raw
Adobe
Apptimize
Attraqt
Evolv AI
GA4 Experiments
PageDuel
VWO Testing
Visual Website Optimizer (VWO)
Discontinued, still offered
No shut-down product was recommended here.