Case study · Soapbox Bulletin · SMS & messaging
480% more verified AI wins in 11 weeks.
Soapbox Bulletin went from 5 verified first-place AI recommendations to 29, on identical prompts, and ended the period as the most recommended brand in its benchmark.
+480%
verified AI wins
5 → 29 of 80 queries; win rate 6.25% → 36.25%
25% → 71%
correct-domain attribution
the brand resolved to soapboxbulletin.com
58 vs 34
verified #1 finishes
most in the benchmark field, full period
Every number URL-verified against soapboxbulletin.com
What moved it
We made soapboxbulletin.com more authoritative. That is what changed the AI’s opinion.
The 20 benchmark questions never changed. Neither did the competitor field: Peerly, Wonder Cave, Twilio, DirectSnd, and SimpleTexting, the five alternatives the AI platforms themselves surfaced during discovery research. The variable we controlled was the content on Soapbox Bulletin’s own domain, engineered articles built from what the benchmark showed the platforms weigh in this category, each one edited and approved by a human before it published. As that library grew, the AI-perceived authority of the domain rose, and the platforms’ own reasoning started to favor Soapbox Bulletin in comparisons they had answered differently for months.
Same questions. Same competitors. A different recommendation. The win rate is the receipt.
The trajectory
Attribution moved first. Preference followed.
April to May was the attribution phase. Several unrelated companies share the “Soapbox” name, and at baseline the platforms connected the brand to soapboxbulletin.com in only 25% of answers. Correct resolution nearly tripled as the structured content was indexed.
May to June was the preference phase. Attribution held while first-place recommendations kept climbing, which is what you expect when platforms start choosing a brand over alternatives they also know.
25% → 51%
attribution-to-win conversion
Share of voice roughly doubled over the period. The win rate grew nearly sixfold. When the two diverge like that, the platforms are doing more than mentioning the brand more often. They prefer it.
The field
Category leader across the full measurement period.
Verified, URL-grounded first-place finishes across all 240 query-answers. Every entity is counted by canonical domain, the competitors included, so name variants inflate neither side.
Soapbox BulletinClient
58
SimpleTexting
34
Wonder Cave
31
Peerly
28
Twilio
20
DirectSnd
11
Verified #1 finishes, 240 head-to-head query-answers, April 8 – June 24, 2026
By June the lead was wider still: 29 first-place finishes in the month’s run against 11 for the nearest competitor.
Platform detail
Perplexity: from absent to 11 wins. ChatGPT: the honest gap.
Grok
5 12
The only platform that had surfaced soapboxbulletin.com at baseline.
Perplexity
0 11
Zero verified attributions at baseline. By June it cited soapboxbulletin.com in 19 of 20 answers.
Gemini
0 6
From brand-absent to six verified wins.
ChatGPT
0 0
Zero verified wins across the period. A strategically important platform, and the clearest next target.
Verified #1 recommendations, of 20 head-to-head prompts per platform
We report the gap as plainly as the gains. ChatGPT still connects the brand to soapboxbulletin.com in only 2 of 20 answers. That is the single largest untapped opportunity, and the same content methodology applies to it.
Ongoing authority building
21 engineered articles in seven weeks, across multiple named experts.
The blog was empty at the April baseline. Every article was written for how AI reads: structured claims, direct answers, evidence a machine can parse and cite. And publication ran on a deliberate multi-author strategy, which is what made this velocity possible.
Named experts, real people
Each byline belongs to a real human counterpart with their own voice and their own area of the comparison evidence. Nobody’s name is borrowed: the person behind each byline edits and approves every post published under it before it goes live. There is always a human in the loop.
A larger footprint, faster
A single byline can limit publishing velocity and reduce the credibility of a high-volume program. Parallel experts remove that ceiling: the domain builds a much larger digital footprint, much faster, and attacks different facets of the head-to-head evidence at the same time.
Human approval, every time
Nothing publishes on its own. Every piece passes human curation, editing, and sign-off before it goes live. Velocity comes from the system; judgment stays with people.
Each publishing burst preceded higher measured visibility in the next benchmark.
Empty blog
0 articles
next measurement · April 8
6.25%
First burst, one day
10 articles
next measurement · May 29
30.0%
Second burst, two days
11 articles
next measurement · June 24
36.25%
Methodology & verification
Locked prompts
All trend claims use the April 8 prompt set, rerun without modification on May 29 and June 24. Like-for-like, every time.
URL-grounded verification
Every prompt requires each company's official domain, so every answer carries its own verification key. Answers that pointed at similarly named companies, 20 to 27 per run, were identified and excluded.
The AI picked the competitors, not us
Peerly, Wonder Cave, Twilio, DirectSnd, and SimpleTexting are the alternatives the platforms themselves surfaced during discovery research. The benchmark tests the field the AI already recommends, including brands far larger than the client.
Measured again next month
The benchmark reruns monthly on the same prompts, so a gain only counts if it holds. Month over month, the record shows whether the domain's authority is maintained, compounding, or slipping.
The benchmark demonstrates that sustained, structured content publication can shift AI recommendation preference in controlled head-to-head comparisons against the same LLM-surfaced competitors.
Who does the AI recommend first in your market?
A walkthrough shows you, live, how ChatGPT, Perplexity, Gemini, and Grok answer your buyers' questions today, and what we would publish to change their answer.