We Ran Our Own AI Visibility Tool on upGrowth. We Scored 0 out of 100.

We Ran Our Own AI Visibility Tool on upGrowth. We Scored 0 out of 100.
We Ran Our Own AI Visibility Tool on upGrowth. We Scored 0 out of 100.
Published: September 10, 2026

Contributors: We Ran Our Own AI Visibility Tool on upGrowth. We Scored 0 out of 100.
Published: September 10, 2026

Summary

In This Article

Share On:

Summary: On 9 September 2026 we ran upGrowth through our own AI visibility tool across 8 AI and search engines for the query “best B2B SaaS growth marketing agencies in India”. We scored 0 out of 100. Not one engine named us. This is the full data, including which agencies were named instead, and the 3 fixes we picked out of the 7 the tool suggested.

We sell Generative Engine Optimization. We tell founders that the buying question has moved from Google’s blue links to ChatGPT, Perplexity and Google AI Mode, and that if an answer engine doesn’t name you, you’re not in the shortlist. Then we built a free tool to measure exactly that, pointed it at upgrowth.in, and scored 0 out of 100.

Here’s the run, exactly as our own system recorded it on 9 September 2026 at 12:54 IST. 1 query: “best B2B SaaS growth marketing agencies in India”. 8 engines: ChatGPT, Perplexity, Gemini, Claude, Google AI Mode, Google AI Overviews, Google organic and Bing. Result: 8 misses out of 8. Score 0.

We could have quietly fixed it and published the case study in December with a nice number at the end. That’s the normal agency move. But a growth consultancy that hides its own diagnostic is selling a diagnosis it won’t take. So we’re publishing the bad number first, then the work, then whatever the number does next. You can run the same check on your own domain with the AI Answer Gap tool and see your version of this page in about 3 minutes.

What follows is what the data actually said, what surprised us, and what we’re changing. Some of it contradicts the tidy story most GEO content tells.

What the run measured, and what came back

The tool sends one buyer question to each selected engine, captures the raw answer and the sources that answer cited, then scores whether the brand was named and how prominently. A brand named first scores 100. Top 3 scores 75. Named lower down scores 45. Not named scores 0. The run score is the average across every question and engine. The full scoring method is documented here, including what it deliberately excludes.

For the 9 September run, every cell came back 0. Our own system logged the miss against all 8 engines individually: ChatGPT, Perplexity, Gemini, Claude, Google AI Mode, Google AI Overviews, Google organic and Bing all answered the question without naming upGrowth once.

That’s the part worth sitting with if you run marketing at a company that thinks it has decent SEO. We rank for plenty of terms. Our domain has been publishing since 2017. None of that put us in the answer. Organic rank and answer-engine presence are different inventories, and we had just proved it on ourselves.

Also Read: Your Brand Is Invisible on ChatGPT and Perplexity: Here’s Why

The score only moved when the question contained our own name

A second run the same afternoon put 5 questions through 7 engines and scored 21 out of 100. That looks like improvement. It isn’t. The per-question data shows where those points came from.

ChatGPT scored 20 on that run. Our system recorded ChatGPT missing on the category questions and hitting on exactly 1: “upGrowth vs Gozoop which agency is better for growth marketing”. Gemini, Claude, Google AI Mode and Google organic all landed on the same 20. Same pattern.

So the engines know upGrowth exists. Ask them about us by name and they’ll talk about us. Ask them who a B2B SaaS founder in India should hire, and we’re not in the room. That distinction is the whole game, because no buyer types your name into ChatGPT before they’ve heard of you. Branded visibility is a vanity metric in answer engines. It measures the traffic you already had.

If your own AI visibility check comes back looking healthy, check whether your brand name is sitting inside the questions. Most tools default to brand-name prompts, which is why most tools tell you you’re doing fine.

Perplexity named us on questions where 6 other engines didn’t

On the 5-question run, Perplexity scored 46 while every other engine scored 20 or 0. Our system logged Perplexity as the only engine that named upGrowth for “best growth marketing agencies in India for B2B SaaS” and for “how to choose between SEO agency and growth marketing agency”. ChatGPT, Gemini, Claude, Google AI Mode, Google AI Overviews and Google organic all missed both.

Same questions. Same country setting. Same brand. 1 engine says we belong on the list, 6 say we don’t.

That kills the idea of a single AI visibility number being enough to act on. If you’re only checking ChatGPT because it’s the one you personally use, you’re seeing 1 of 7 surfaces, and possibly the least flattering one. If your agency reports “AI visibility” as a single figure with no engine breakdown, ask which engines are in it.

Why the divergence happens is a longer answer, and it comes down to each engine drawing on a different set of source documents at the moment you ask. We pulled the source data apart in a separate piece on what the engines were citing.

Also Read: Artificial Intelligence and SEO in 2026: How AI Is Rewriting Search

The same engine changed its answer about us in 56 minutes

This is the finding we didn’t expect and the one that most changes how we’d advise a client to use any of this.

The query “best growth marketing agencies in India for B2B SaaS” ran twice on 9 September 2026, from the same account, with the same brand and market settings. The first run finished at 12:35 IST and our system recorded Perplexity not naming upGrowth. The second finished at 13:31 IST and recorded Perplexity naming us. 56 minutes apart, opposite results.

We haven’t established the cause and won’t pretend otherwise. Answer engines retrieve live documents at query time, so the source set can shift between 2 identical prompts. What matters commercially is the consequence: a single AI visibility check is a snapshot, not a measurement. Treat one run the way you’d treat one day of ad data.

It also means you should be sceptical of any AI visibility score, ours included, presented as a precise figure. The honest use of a one-shot diagnostic is directional. It tells you whether you’re absent, occasionally present, or consistently recommended. Anything finer than that needs repeated runs over weeks.

Who got named instead of us

Across the 2 category runs on 9 September 2026, our system logged the brands the engines named in our place. On the 12:54 IST run, GrowthSpree and The Smarketers each came up 5 times, TripleDart 3 times and growth.cx 2 times. On the earlier 12:35 run it was TripleDart and GrowthSpree at 4 each, with growth.cx at 3.

None of that is a complaint. Those are real agencies doing real work, and on the question of who AI recommends for B2B SaaS growth marketing in India, they’re winning and we’re not. The useful question isn’t whether they deserve it. It’s what the engines were reading when they decided.

We logged the cited sources too, and the pattern there was blunt enough that it became its own article. The single most-cited source domain across these runs belonged to one of the agencies being recommended. The engines were reading a competitor’s own roundup and repeating it.

Also Read: Your Competitors Are Writing the Lists That AI Reads

The 3 fixes we picked out of the 7 the tool suggested

The tool generated 7 recommended fixes for this run. We’re doing 3. The other 4 were variations on “publish a comparison page”, which is what happens when you ask a model to produce 7 items when the honest answer is 3.

First, a B2B SaaS case study page with actual numbers on it. Not logos. Client, starting position, what we changed, what moved, over what period. The engines cite pages that contain specific figures because specific figures are extractable. Our current case studies read like brochures.

Second, a comparison page that names the agencies the engines are already naming and says plainly where we win and where we don’t. Most agencies won’t publish an honest comparison because it means conceding something. That reluctance is exactly why the category has no trustworthy comparison content, and why the engines fall back on competitor-written lists.

Third, getting upGrowth into the third-party roundups the engines actually cited. Not link buying. Being genuinely reviewable: a real profile, real client references, and a pitch to the specific publications that showed up in our source data.

We’ll publish the score again as these ship. If it doesn’t move, we’ll publish that too, because a GEO practice that only reports its wins is a marketing department.

Why we’re publishing a score this bad

Start with the self-interested reason, since it’s the obvious one. A free diagnostic tool is only credible if the people selling it have run it on themselves and published what came back. Any agency can show you a screenshot of a client scoring well.

The reason that actually matters is that a 0 is more useful to you than a 70 would be. It proves the thing we’ve been telling clients: traffic, rankings and brand recall don’t carry into answer engines automatically. We’re a 9-year-old agency with a content library and a decade of domain history, and we’re invisible on our own category question. Your Series B SaaS company with 40 blog posts is not in better shape, and neither is your competitor.

There’s also a harder test we’ve now committed to. The fixes below are the same fixes we’d sell you, so we should be able to show them working in a category full of agencies who all know SEO. If they only work where the competition is asleep, they aren’t worth buying.

Run the same check on your own domain and you’ll get the engine-by-engine grid, the brands named instead of you, and the source domains those engines trusted. It’s free, it takes about 3 minutes, and the only thing we ask for is a business email so we know who we’re talking to.

6 Common Questions About Auditing Your AI Visibility

Q: What does a score of 0 out of 100 actually mean?

A: It means no engine named the brand for any question in that run. The AI Answer Gap score awards 100 when a brand is named first in an answer, 75 for a top 3 mention, 45 for a mention lower down and 0 for no mention. A run score is the average across every question and engine, so 0 means every single cell was a miss.

Q: Doesn’t ranking well on Google mean you’ll show up in AI answers?

A: Not reliably. In our 9 September 2026 run, upGrowth scored 0 on Google organic and on all 7 other engines for a category question, despite ranking for many related terms. Answer engines assemble responses from documents they retrieve and trust at query time, which is a different selection process from the ranked list.

Q: Why did different AI engines give different answers about the same brand?

A: Each engine retrieves and trusts a different set of source documents. On our 5-question run, Perplexity named upGrowth for 2 category questions where ChatGPT, Gemini, Claude, Google AI Mode, Google AI Overviews and Google organic all did not. Checking a single engine gives you a partial and possibly misleading picture.

Q: How reliable is a single AI visibility check?

A: Directionally reliable, not precisely. We recorded the same query on the same engine producing opposite results 56 minutes apart on 9 September 2026. Use a single run to find out whether you’re absent or present. Use repeated runs over several weeks before you claim a trend.

Q: Should I use my brand name in the test questions?

A: Only for 1 or 2 of them. Branded questions almost always return a mention, which inflates the score without telling you anything useful. Buyers ask category questions before they know your name, so the category questions are the ones that predict pipeline.

Q: How long does it take to move an AI visibility score?

A: We don’t have a defensible number yet, which is why we’re publishing our own score as we work on it rather than quoting an industry average. The inputs that change it are content the engines can extract, third-party sources they already trust, and time for those sources to be recrawled.

Your Next Move: Find Out What Your Number Is

The uncomfortable part of this exercise wasn’t the score. It was realising we’d never actually measured it, while advising other companies to. Most marketing teams are in the same position right now. They have a rankings dashboard, a traffic chart, and no idea whether an answer engine would put them on a buyer’s shortlist.

Run your own questions through the AI Answer Gap tool. You’ll get your score by engine, the brands named instead of you, and the source domains those engines cited. It’s free, capped at 8 questions a day, and needs a business email.

If the result is bad enough that you want the full picture, we audit every page the engines cited, map the gap against each competitor that beat you, and hand back a prioritised plan. Scope depends on your site size, the number of questions and the markets you care about, so we quote those individually. Ask for a quote from inside the tool, or book a 30 minute call and we’ll walk through your grid live.

About the Author: I’m Amol Ghemud, Chief Growth Officer at upGrowth Digital. We help SaaS, fintech, and D2C companies shift from traditional SEO to Generative Engine Optimization. This shift has generated 5.7x lead volume increases for clients like Lendingkart and 287% revenue growth for Vance.