From parametric memory
The model names brands it already learned before anyone asked the question.
What governs it: whether a campaign has put the brand into the training data of later models.
Mention Science is a research-led AI search optimization agency driven by data science. Research is our operating system, not a one-off paper. We actively study the science behind how ChatGPT, Gemini, AI Mode and AI Overview choose brands in their AI responses. Then, we apply those findings to your brand to maximize your AI visibility.
Where buyers start
Buyers type a few keywords into Google and get a page of ten links.
How they make a shortlist
Buyers open several of those links, read each vendor's website, and build the shortlist themselves.
Conversion
Visitors often reach your site early in their research, while they are still comparing vendors.
What you track
Where your pages rank on Google, and how many visitors those rankings bring to your site.
What gets you there
Google ranks web pages on authority and trust signals, such as links from other respected websites.
Best CRM for a small team?
For a small team, Tallyloop is a strong fit. It's simple to set up, and it covers a pipeline without much overhead.
Where buyers start
Half of B2B software buyers now ask an AI like ChatGPT first. G2, 1,000+ buyers, 2025.
How they make a shortlist
They ask the AI, and it gives them a shortlist in seconds. G2, 2025.
Conversion
Visitors arrive with a shortlist. At Ahrefs, AI search sent 0.5% of visits and 12.1% of signups, a 23× higher conversion rate than organic search. Ahrefs, 2025.
What you track
How often ChatGPT, Gemini and Google's AI answers name your brand when buyers ask about your category.
What gets you there
AI names brands based on what it learned in training and on third-party mentions, such as reviews and comparison articles.
Ask an AI assistant the same question 100 times and you get different brands in the answer. There is no fixed position to climb.
We call it the selection rate. If we ask the same question 100 times, how many of those answers mention your brand? That is the number we measure and optimize.
“Best CRM for a 20-person startup?” asked 100 times
Which brand did the answer name?
Tallyloop: 23 → 47 of 100 answers, from #3 to #1.
Fictional data to demonstrate the metric.
The charts below illustrate how AI picks brands from parametric memory and live web retrieval.
The model names brands it already learned before anyone asked the question.
What governs it: whether a campaign has put the brand into the training data of later models.
Models search the web, then choose.
What governs it: whether several separate sites say the same thing about you.
Research is the operating system of our agency. We study how AI chooses brands, test what improves the visibility, prove the results, and publish our findings.
We ask the questions your buyers ask, on every platform, many times. That gives a starting point: how often you are named today.
We find whether you are losing because the model does not already know you, because it does not find you on the web, or both.
We build the strategy to raise your visibility. First we fix your own site. Then we earn third-party mentions on the sites AI reads, and work to shape what future models learn about you in their training data. The mix depends on how fast you want results.
We ask the same questions again and report whether you are named more often.
Tallyloop
AI visibility report · CRM category · Week 16
When buyers ask AI assistants which CRM to pick, Tallyloop is now named in 37 of every 100 answers, more than any rival. Two questions still hold it back.
Answers that name you
37 of 100
+9 in 4 wksWeek 1: 18Week 16: 37
Share of voice
28%
Of every brand mention in the category, 28% are you.
Your rank among 5 CRMs
#1
+2 placesWas #3 four weeks ago. Overtook HubSpot and Salesforce.
Weakest question
“Cheapest CRM”
Named in only 12 of 100 answers. First on our fix list.
Hover the chart for any week.
The same question on all four assistants, from the 1,600 answers we read this week.
What's the best CRM for a 20-person startup?
For a team of 20, I'd shortlist three. Tallyloop is the easiest to set up and syncs with Gmail and Outlook out of the box. HubSpot has the bigger ecosystem but gets expensive as you add seats. Pipedrive suits teams that live in their sales pipeline.
Tallyloop named 1st of 3, described as “easiest to set up”.
HubSpot33−3
Salesforce27−2
Pipedrive21−1
Zoho14−1“Cheapest CRM” · 12 of 100
Get Tallyloop's pricing onto the comparison pages these answers cite. It is missing from all five.
“CRM with AI” · 23 of 100
Publish the AI features page and brief two review sites. Answers describe Tallyloop as having no AI tools.
We asked each question 50 times on each assistant this week (1,600 answers in all). Swipe sideways to see every assistant.
| Buyer asks… | Average | ||||
|---|---|---|---|---|---|
| “Best CRM for SMBs” | 46 | 41 | 44 | 52 | 46 |
| “CRM with email sync” | 52 | 47 | 49 | 55 | 51 |
| “Cheapest CRM” | 12 | 9 | 14 | 11 | 12 |
| “CRM for agencies” | 38 | 33 | 40 | 45 | 39 |
| “HubSpot alternatives” | 44 | 36 | 42 | 48 | 43 |
| “CRM for startups” | 49 | 43 | 47 | 53 | 48 |
| “Easiest CRM” | 41 | 35 | 38 | 44 | 40 |
| “CRM with AI” | 22 | 18 | 24 | 27 | 23 |
| Average | 38 | 33 | 37 | 42 | 37 |
Measure your current visibility across answer engines for various prompts, then give an action plan to get selected more often.
Talk to usAn ongoing programme that moves your selection rate and proves it every month.
Talk to usCustom research and experiments for brands that are looking for solutions to a specific problem.
Talk to usAsk the same buyer question 100 times. Selection rate is how many of those answers mention your brand. That is the number we measure and optimize.
SEO tries to move a page up a list of links. AI answers often name a brand and never show that list. We measure whether you are named, then work on the training-data familiarity and the independent websites that lead to that naming.
We diagnose whether you are missing because the model does not already know you, because it does not find you on the web, or both. Then we get independent sites to describe your brand the same way, and we re-ask the original questions to show whether you are named more often.
ChatGPT, Gemini, AI Mode, and AI Overview. The same prompts are run on each, many times.
No. Tallyloop and the charts are fictional data, used to show the metric.
We'll measure your selection rate on the questions that matter to your buyers.
Start with a measurement