Correction, 2026-07-16. The original version of this post said the lesson was about framing: reword your category question as the specific job and the challenger wins. A reader, Paul Towers, caught a confound. The two questions below did not only change from a category label to a job. They also changed the buyer, from a business to a creator. So I reran it holding the buyer constant, 10 times per model, and the framing effect disappeared. A business buyer who describes the job in plain words still gets Beehiiv 0 out of 40, the same as the category label. The corrected finding is at the top. The original numbers are left untouched below, for the record.
What actually moved the result: who is asking, not how they phrase it
I reran the test with a third question, keeping the buyer a business the whole time and only changing the framing from a category label to the specific job.
Question Beehiiv (all 4 models)
-------------------------------------------------------- ----------------------
Business buyer, category label 3/40
"What's the best email marketing platform for a business?"
Business buyer, the job in plain words 0/40
"We have an existing customer base we want to reach via
an owned channel like email. Walk me through it."
Creator buyer, the job 25/39
"How does a creator or founder start a newsletter, grow
subscribers, and make money from it?"
-------------------------------------------------------- ----------------------
Holding the buyer constant and changing only the framing takes Beehiiv from 3 to zero. There is no framing effect. The entire jump was the buyer identity. Beehiiv is filed under "creator," and it vanishes for a business buyer at any phrasing. ChatGPT named it 0 out of 10 on all three questions, which is the one part of the original that survives.
The real lesson is harder and truer than "phrase it as a job." You cannot reframe your way into a category. The models file you under a buyer identity, written in the words the people who actually use you have published about you. Beehiiv owns the creator's newsletter job because the creator world documented it as the answer to that job. A business buyer never reaches it, because nobody wrote Beehiiv up as the answer to a business's email program, no matter how that business phrases the question.
So if you are the challenger: the move is not to reword your question. It is to pick the specific buyer whose job you can genuinely own, and get documented, by real users and real publications, as the answer to that job in the words that buyer uses. Then the models reach for you when that buyer asks, and only when that buyer asks. Watch ChatGPT separately, because it moves last.
Original post below, kept for the record.
More buyers now ask ChatGPT, Claude, Perplexity, or Gemini for a tool before they ever open Google. So we keep measuring what those models actually say when someone asks for software in a category. Email marketing gave us the most encouraging result we have run, because of Beehiiv. It is the one teardown where the challenger actually won its slot, and you can see exactly how.
We asked two versions of the same question and ran each one 10 times on all four models. That is 40 answers per question.
Version one was the plain category label:
What's the best email marketing platform for a business?
Version two was the specific job Beehiiv is built for:
How does a creator or founder start a newsletter, grow subscribers, and make money from it?
The receipt
How many of the 10 runs named each tool, per model.
Best email tool Start and grow a newsletter
(category label) (the job Beehiiv is built for)
Model Beehiiv Mailchimp Beehiiv Mailchimp
----------------- ------- --------- ------- ---------
ChatGPT 0/10 10/10 0/10 10/10
Claude 0/10 10/10 10/10 10/10
Gemini 0/10 10/10 9/10 10/10
Perplexity 4/10 9/10 10/10 7/10
----------------- ------- --------- ------- ---------
All four models 4/40 39/40 29/40 37/40
Named on the generic question: Mailchimp, Klaviyo, ActiveCampaign, Brevo,
Constant Contact. Named on the newsletter question: Beehiiv, Kit, Substack.
On the category label, Beehiiv is almost invisible
Ask for the best email marketing platform and Beehiiv was named 4 times out of 40, all of them on Perplexity. Mailchimp was named 39 times, with Klaviyo, ActiveCampaign, Brevo, and Constant Contact filling the rest. If Beehiiv tried to win the generic email marketing question, it would lose, the same way every challenger loses the category label.
The one model that still will not budge
Look at ChatGPT. Even on the newsletter question, the question written for Beehiiv's exact buyer, ChatGPT named it 0 times out of 10. This keeps happening in our teardowns. ChatGPT is consistently the slowest of the four to name a challenger, even one that the other three now reach for easily. If ChatGPT is where your buyers ask, winning the sub-intent on the other three is not enough on its own, and it is worth knowing that before you assume the job is done.
Check your own category
Pick your category. Ask the plain label question, then ask it the way your actual customers describe the job they hire you for, in their own words. Run both a handful of times on ChatGPT, Claude, Perplexity, and Gemini and count who gets named. If you are invisible on the label but present when the question sounds like your real buyer, that tells you which buyer the models have you filed under. If you are invisible on both,
Gemini 0/10 10/10 9/10 10/10
Perplexity 4/10 9/10 10/10 7/10
All four models 4/40 39/40 29/40 37/40
Named on the generic question: Mailchimp, Klaviyo, ActiveCampaign, Brevo,
Constant Contact. Named on the newsletter question: Beehiiv, Kit, Substack.
## On the category label, Beehiiv is almost invisible
Ask for the best email marketing platform and Beehiiv was named 4 times out of 40, all of them on Perplexity. Mailchimp was named 39 times, with Klaviyo, ActiveCampaign, Brevo, and Constant Contact filling the rest. If Beehiiv tried to win the generic email marketing question, it would lose, the same way every challenger loses the category label.
## The one model that still will not budge
Look at ChatGPT. Even on the newsletter question, the question written for Beehiiv's exact buyer, ChatGPT named it 0 times out of 10. This keeps happening in our teardowns. ChatGPT is consistently the slowest of the four to name a challenger, even one that the other three now reach for easily. If ChatGPT is where your buyers ask, winning the sub-intent on the other three is not enough on its own, and it is worth knowing that before you assume the job is done.
## Check your own category
Pick your category. Ask the plain label question, then ask it the way your actual customers describe the job they hire you for, in their own words. Run both a handful of times on ChatGPT, Claude, Perplexity, and Gemini and count who gets named. If you are invisible on the label but present when the question sounds like your real buyer, that tells you which buyer the models have you filed under. If you are invisible on both, you have a job to go own.
If you want the fast version, run your own domain through the free scan and see who AI recommends in your category: [bersyn.com free scan](https://bersyn.com/?utm_source=devto&utm_medium=referral&utm_campaign=email-beehiiv-teardown).
Which buyer are the models filing you under, and is it the one you actually sell to?
Top comments (0)