# AI visibility: 89% spend more, 24% can prove a return. What makes the difference

> Survey of 700 companies: 89% spend more on AI visibility but only 24% can prove a return. Which metrics they use, why measurement is hard and what to do.

URL: https://geosnap.ai/en/blog/visibilita-ai-89-per-cento-spende-di-piu-24-per-cento-dimostra-ritorno-cosa-fa-la-differenza
Language: English
Version IT: https://geosnap.ai/blog/visibilita-ai-89-per-cento-spende-di-piu-24-per-cento-dimostra-ritorno-cosa-fa-la-differenza
Author: Rinald Sefa, CMO Geosnap · Category: Research and data
Publisher: Geosnap (Maind Group S.r.l.), https://geosnap.ai

## In short

A survey of 700 marketing decision-makers in the US, UK and Germany (November 2026) finds that 89% increased spend on visibility in AI answers, yet only 24% can prove a return and only 40% trust their own metrics. What makes the difference is not company size but organisation: companies that integrate AI search into their marketing strategy prove a return 43% of the time against 23%, and those with a dedicated owner 35% against 15%. Measurement also stays hard for technical reasons: Google has just confirmed that AI Mode and AI Overviews have no referer of their own.

- 89% of companies raised AI search spend and 81% will raise it again, but only 24% can prove a return; among those who raised spend significantly, 64% cannot.
- Visibility in AI answers is the most used (58%) and most trusted (41%) metric; revenue and leads attributed to AI are used by 42% and 39% but trusted by only 26% and 22%.
- The most cited barrier is the pace of change (33%): within one quarter, ChatGPT sessions with multiple searches went from 2% to 61% and Reddit citations collapsed.
- Google confirms that AI Mode and AI Overviews traffic carries the same referer as classic search: GA4 cannot separate it, and Search Console gives no clicks.
- An integrated strategy (43% vs 23%), dedicated roles (35% vs 15%) and a clear owner matter more than company size: mid-market and enterprise both sit at 24%.

Written by the Geosnap AI agent, reviewed and approved by Rinald Sefa.

Translated from the Italian original. [Read the original](https://geosnap.ai/blog/visibilita-ai-89-per-cento-spende-di-piu-24-per-cento-dimostra-ritorno-cosa-fa-la-differenza)

Nine companies in ten increased what they spend on visibility in AI answers over the past year, yet only one in four can show what it returned. That is the central finding of [Primary Sources](https://peec.ai/reports/primary-sources-q3-2026), a survey of 700 marketing decision-makers in the United States, the United Kingdom and Germany published on 11 November 2026, and it matches what another study found in June on a different sample. Here are the numbers, the reasons measurement is so hard and, above all, what separates the companies that can prove a return from those that cannot.

## How much are companies spending on AI visibility, and how much can they prove?

Almost everyone is spending more; few can prove the return. In the survey, run by Sapio Research in September 2026 among 700 marketing decision-makers at companies with at least 100 employees (300 in the US, 200 in the UK, 200 in Germany), [89% raised their AI search spend over the past year and 81% expect to raise it again](https://www.globenewswire.com/news-release/2026/11/11/3400710/0/en/ai-search-is-changing-faster-than-marketing-teams-can-keep-up-despite-increased-spend.html). Only 24% say they can prove a return on that investment. Among companies that raised spend significantly, 64% cannot show what it returned.

The pressure to invest is concrete rather than fashionable: [43% cite competitors appearing in AI answers and 42% cite customers using AI tools to research or buy](https://peec.ai/reports/primary-sources-q3-2026), 34% point to leadership pressure and 28% to industry media. 61% call growing AI visibility a top priority, and 60% checked what an AI assistant says about their brand several times in the past month. Interest is not the problem. Proof is.

The research was commissioned by a company that sells an AI visibility platform, as is true of many studies in this field, so read the figures with that in mind. The picture does match [a Semrush study from 3 June 2026](https://www.semrush.com/blog/the-operational-gap-ai-seo-study/) among 481 marketers and SEO professionals: there, only 9% said they could measure all the metrics that matter, and 49% named the impact of AI on pipeline and revenue as the hardest thing to measure.

## Which metrics do companies use, and which do they trust?

They use many metrics and trust few of them: overall, [only 40% of respondents say they have metrics they trust](https://peec.ai/blog/ai-search-marketing-planning). The gap between "used in the past three months" and "trusted" is wide everywhere, and it widens the closer a metric gets to the business.

| Metric | Used in the past 3 months | Trusted |
| --- | --- | --- |
| Visibility in AI answers | 58% | 41% |
| Brand perception or sentiment | 54% | 34% |
| AI referral traffic | 51% | 34% |
| Revenue attributed to AI search | 42% | 26% |
| Conversions or leads from AI traffic | 39% | 22% |
| Citation frequency | 35% | 19% |
| AI bot visits | 31% | 17% |

Source: [Peec AI, Primary Sources Q3 2026](https://peec.ai/blog/ai-search-marketing-planning). And yet 47% of companies trace a lead, a deal or revenue to AI search at least weekly: the link to the business is visible, but case by case, not as a channel measure. How the data is collected explains part of the distrust. In the Semrush study, [40% check visibility by typing prompts into AI tools by hand](https://www.semrush.com/blog/the-operational-gap-ai-seo-study/), 38% use their usual SEO platform, 36% a specialised tool and 13% have no consistent approach at all. A manual check by one person, on one day, from one account is not repeatable, which is why it pays to [fix a prompt list](https://geosnap.ai/en/blog/quali-prompt-monitorare-risposte-ai-come-scegliere) and record it the same way every time.

## Why is AI visibility so hard to measure?

Because the ground keeps moving and the basic data is missing. The most cited barrier to investing more is that [AI platforms are changing too quickly (33%)](https://www.globenewswire.com/news-release/2026/11/11/3400710/0/en/ai-search-is-changing-faster-than-marketing-teams-can-keep-up-despite-increased-spend.html), followed by brand safety (30%) and the lack of reliable measurement (28%). The same report shows how much changes in a single quarter: after GPT-5.6 became ChatGPT's default model on 8 August, sessions running multiple searches went from 2% in July to 61% in September, Reddit citations collapsed to an index of 11 against 100 in the second quarter, and product-page citations rose to 250. A benchmark measured in July was already stale by September.

An important part of the basic data is simply missing. On 12 November 2026 Google published [a guide to identifying Google Search traffic in analytics and server logs](https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic): the referer is `https://www.google.com/` or a local domain such as `https://www.google.co.uk/`, with no path or query string, and "other Google services served from the same origins" send the same values. AI Mode and AI Overviews are not named and have no referer of their own: in Google Analytics a click from AI Mode cannot be told apart from a click on the ten blue links. Google itself writes that the method "isn't exact", because privacy settings, redirects and different methodologies make analytics and Search Console diverge. The [Search Console AI report](https://geosnap.ai/en/blog/report-ai-search-console-senza-clic-come-leggerlo-cosa-fare), for its part, shows impressions but no clicks. ChatGPT is better thanks to the `utm_source=chatgpt.com` parameter, but in October [a bug made it disappear for a week](https://geosnap.ai/en/blog/traffico-chatgpt-crollato-utm-visite-perse-o-etichette-perse) and many companies watched the channel halve without losing a single visit.

## What separates the companies that prove a return from those that cannot?

Organisation, not size. Mid-market companies (100-999 employees) and enterprises (1,000+) prove a return at the same rate, 24%. What changes everything is how AI search sits inside marketing.

| How AI search is organised | Has trusted metrics | Proves ROI |
| --- | --- | --- |
| Strategy integrated into marketing | 72% | 43% |
| Defined but separate strategy | 39% | 23% |
| Dedicated full-time roles | 57% | 35% |
| Part of a broader role | 28% | 15% |
| Dedicated AI visibility platform | 58% | 39% |
| Features inside an existing tool | 42% | 24% |

Source: [Peec AI, Primary Sources Q3 2026](https://peec.ai/blog/ai-search-marketing-planning). Companies with an integrated strategy involve data teams (71%), product (44% versus 24%) and engineering (37% versus 20%) far more often, and [only 5% have stopped optimising for AI recommendations, against 31% of those still exploring](https://www.globenewswire.com/news-release/2026/11/11/3400710/0/en/ai-search-is-changing-faster-than-marketing-teams-can-keep-up-despite-increased-spend.html). Trust in the data protects the investment: among those who trust their metrics, 7% dropped this work; among those who do not, 31%.

The weak spot is ownership. AI search is led by the SEO team at 24% of companies and by executive leadership at 22%; where several teams share the work, 88% report coordination problems, with conflicting priorities and metrics (39%) and waiting on other teams' capacity (35%). Semrush tells the same story: [no function owns AI visibility at more than 18% of companies and one team in ten has no owner at all](https://www.semrush.com/blog/the-operational-gap-ai-seo-study/). Among teams seeing results, only 3% have no owner and 6% have no consistent way to measure; among those not seeing results, 22% and 27%. Country matters too: 31% of US companies prove a return, against 19% in the UK and 18% in Germany, where an integrated strategy is rarer (22% versus 38% in the US). There is no Italian data, and no reason to assume Italy does better.

## What to do in your company to move from spend to proof

Four organisational choices matter more than any tool.

1. **One owner, inside marketing.** Whoever runs SEO is the natural candidate, as long as AI visibility joins the same plan with the same goals rather than living in a side project. The 43% versus 23% figure says this is the single most important choice.
2. **Three layers of metrics, each with its own source.** Visibility: a fixed prompt list, recorded with the same method every week across ChatGPT, Gemini and Google AI Mode, tracking share of mentions, citations and sentiment. Traffic: a dedicated AI assistant channel group in GA4 ([how to set it up](https://geosnap.ai/en/blog/traffico-da-chatgpt-come-misurarlo-google-analytics-4)), knowing AI Mode will stay counted inside Google organic. Outcomes: leads and deals traced to AI, logged weekly as 47% of companies already do, plus a "how did you hear about us" question in your forms.
3. **Accept the imprecision and state it.** Google says traffic identification "isn't exact"; Search Console gives no clicks; ChatGPT can lose its parameters. Document the limits in the report instead of pretending the number is precise: the 40% who trust their data do not have perfect data, they have a stable method.
4. **Review the method every quarter, not the strategy every week.** Engines change models and behaviour within weeks; your prompt list, tracked competitors and metric definitions deserve a fixed review date, but the investment should not stop at every tremor.

A tool's job is to make the first layer repeatable and hold the other two together: Geosnap, for example, records ChatGPT, Gemini and Google AI Mode answers to the same prompts over time and connects to Google Analytics and Search Console, so visibility, traffic and impressions sit in one place. But as the data shows, a tool on its own lifts the share of companies proving a return from 24% to 39%; integration into marketing lifts it to 43%. Organisation first, platform second.

## Sources

- [Primary Sources, Q3 2026](https://peec.ai/reports/primary-sources-q3-2026) (Peec AI (indagine Sapio Research su 700 decisori marketing), 2026-11-11)
- [AI search moves faster than marketing planning](https://peec.ai/blog/ai-search-marketing-planning) (Peec AI, 2026-11-11)
- [AI search is changing faster than marketing teams can keep up, despite increased spend](https://www.globenewswire.com/news-release/2026/11/11/3400710/0/en/ai-search-is-changing-faster-than-marketing-teams-can-keep-up-despite-increased-spend.html) (GlobeNewswire (comunicato Peec AI), 2026-11-11)
- [Only 22% of marketers have fully integrated AI search and SEO. They're pulling ahead. [Study]](https://www.semrush.com/blog/the-operational-gap-ai-seo-study/) (Semrush, 2026-06-03)
- [Identify Google Search traffic in your web analytics or server logs](https://developers.google.com/search/docs/monitor-debug/identify-google-search-traffic) (Google Search Central, 2026-11-12)
- [Google Explains How To Identifying Traffic From Google Search](https://www.seroundtable.com/identify-traffic-from-google-search-42525.html) (Search Engine Roundtable, 2026-11-12)

## Frequently asked questions

### How many companies can prove a return on AI visibility?

According to the Primary Sources survey of November 2026, covering 700 marketing decision-makers in the US, UK and Germany, 24%. The figure is identical for mid-market (100-999 employees) and enterprise (1,000+) companies, but rises to 43% among those that integrated AI search into their marketing strategy and to 31% in the US, against 18% in Germany.

### Which AI visibility metrics are considered most trustworthy?

Visibility in AI answers, meaning how often the brand appears across a fixed prompt list: 58% of companies use it and 41% trust it. Sentiment and referral traffic follow (34% each). Revenue and leads attributed to AI are trusted by only 26% and 22%, because attribution data is incomplete.

### Why can't I see Google AI Mode traffic in Google Analytics?

Because, as Google's guide of 12 November 2026 explains, AI Mode and AI Overviews use the same referer as classic search (https://www.google.com/ or a local domain), with no path or parameters. In GA4 those clicks land in the Google organic channel and cannot be separated. The Search Console AI report shows impressions, but no clicks.

### Who should own AI visibility inside a company?

The data says a clear owner and a strategy integrated into marketing matter more than any specific function. Today the SEO team leads at 24% of companies and executive leadership at 22%; where several teams share the work without an owner, 88% report coordination problems. Among teams seeing results, only 3% have no owner.

### Should we pause investment until we can measure it?

The data suggests the opposite: companies that do not trust their metrics drop optimisation for AI recommendations four times as often (31% versus 7%), while the reasons to invest, competitors in AI answers (43%) and customers using AI (42%), remain. Better to fix a stable method, state its limits and review it every quarter.
