Talk to MapsExperts

Information about how we process your data is available in our Privacy Policy.

Insights

AI & Search · Research

YouTube Shorts and AI visibility: what we observed across 10 local businesses

In September 2026, ten small local businesses published no YouTube Shorts; in October, they started. Across the two months, the average change in their composite MapsExperts AI Visibility Score was +17 points on a 0–100 scale. Separately, we observed about one UTM-attributed lead for every 13,000 Shorts views. What those results show, what they do not, and how to test short-form video properly.

By
, Founder, MapsExperts
Published
Reading time
17 min
Tags
AI visibility · AI search · Google

In short

  • In September 2026, ten small local businesses published no YouTube Shorts; in October, they started. Between the two months, the average change in their composite MapsExperts AI Visibility Score was +17 points on a 0–100 scale. Separately, we observed approximately one UTM-attributed lead for every 13,000 Shorts views.
  • The two results measure different things. UTM-attributed leads are a commercial outcome on one tracked path, not every lead the Shorts influenced. The AI Visibility Score summarizes how the businesses appear in AI answers.
  • A composite score shows that overall AI visibility moved, not which part of it. Our measurement framework reads it alongside its components, never in place of them.
  • Both are aggregated observations, and the score comparison is a before-and-after observation, not a controlled experiment. It shows a measurable change in the month the businesses published Shorts, but not that the Shorts caused it, and neither result establishes a direct effect on AI recommendations or Google Maps rankings. Together they justify further investigation of short-form video in local discovery.
  • A video that states opening hours or prices is one more copy of location data, and it has to be kept current like any other.

What we observed across ten small businesses

Two claims about YouTube Shorts are common: that posting them helps a business rank on Google Maps, and that AI assistants learn about businesses from YouTube. The first can be checked against Google's documentation, which does not support it. No AI provider documents the second, so we looked at ten small businesses we work with that started publishing Shorts. We measured them in two different ways: AI visibility, with the composite MapsExperts AI Visibility Score, read the way our guide to measuring AI visibility for local businesses recommends, and commercial outcome, with UTM-attributed lead tracking. This article reports both, keeps them apart, and sets out how to test short-form video properly.

In September 2026, the ten businesses in our observation group did not publish YouTube Shorts. In October, they introduced Shorts publishing. Using the same MapsExperts AI Visibility Scoring methodology, questions, AI platforms and measurement settings, we compared each business's AI Visibility Score across the two months. The average change across the ten businesses was +17 points on a 0–100 scale.

Separately, we observed approximately one UTM-attributed lead for every 13,000 YouTube Shorts views, with views measured through YouTube Analytics and attributed leads tracked through Google Analytics 4.

The two results are different measurements, and we keep them apart. UTM-attributed leads are a commercial outcome attributed to the videos through tagged links, counted on one tracked path. The AI Visibility Score summarizes how the businesses appear in AI answers. These leads were attributed to the Shorts, not to AI answers, so they are not the outcome layer of our framework. One result can improve while the other does not, and neither explains the other.

About one UTM-attributed lead per 13,000 Shorts views

A UTM-attributed lead is one that arrived through a tagged link from a video and was recorded as such. That makes the ratio a conversion measure for one path: from a Short, through a tagged link, to a lead. It is not a count of every lead the Shorts influenced. Someone who watched a Short and later searched for the business by name, called it or walked in would not be counted, so the ratio may understate any wider contribution of the videos, by an amount it cannot show.

The ratio is useful precisely because it is narrow. It is a clean, repeatable measure: a business that tags its links the same way can compare its own ratio from one period to the next, or against other channels tracked the same way. It puts the scale of video attention next to the leads that came through that path. And it marks where the harder question begins: what the attention does that a tagged link cannot see.

Two things limit it. View counts can include people far outside a business's catchment, because YouTube recommends videos on performance and on each viewer's watch and search history[1]. And as an aggregate across ten businesses, it does not show how much individual businesses differ, and it is not a benchmark for what another business should expect.

An average 17-point increase in the composite MapsExperts AI Visibility Score

The AI visibility result is the one most local businesses will care about: from September, without Shorts, to October, with them, the average change in the ten businesses' composite MapsExperts AI Visibility Score was +17 points on a 0–100 scale.

A composite score is the right number for tracking whether overall AI visibility is moving, and the wrong number for explaining why. It combines several AI visibility metrics, and it can rise because one of them improved strongly while another held still or fell. That is why our measurement guide treats the score as a summary KPI to be read alongside its components, never in place of them. This article reports the average change only, so by the standard our guide sets it is a summary: the component-level results that explain it stay with the underlying records. Read the average 17-point increase accordingly: as a measured average rise in the businesses' composite scores, not as evidence of which dimension moved, whether the right locations were recommended more often, or whether the facts in AI answers became more accurate.

Did YouTube Shorts cause the average 17-point increase in the AI Visibility Score?

We cannot say. The average 17-point increase on a 0–100 scale came in the month the businesses started publishing Shorts, but a before-and-after observation cannot separate the videos from everything else that changed: other work, platform changes and ordinary variation in AI answers could each account for some or all of it. What it justifies is a properly controlled test, described below.

What we can and cannot conclude

Both results are aggregated observational results. They do not show that the Shorts caused either result, and they do not establish a direct effect on AI recommendations or Google Maps rankings. Without a randomized comparison, they cannot show how much of either result would have occurred without the videos. The businesses are clients, so other work may have continued during the observation, and AI products change often enough that an update can move every business's numbers at once. Businesses that publish Shorts regularly may also be the most active on everything else, and an aggregate across ten businesses can hide how differently they moved.

What the observations do justify is further investigation. Across ten businesses, leads came through the Shorts' tagged links at a measurable rate, and the average change in the businesses' composite AI Visibility Score was +17 points on a 0–100 scale, from a month without Shorts to a month with them. That puts short-form video on the list of things a local business should test, with a design that can separate a real effect from everything else. The rest of this article sets out what YouTube is and is not in local search, how video could plausibly affect AI visibility, and how to run that test.

How YouTube content becomes discoverable

The mechanics are mostly documented, and mostly about clarity. Titles and descriptions should say what the video shows and where, in customers' words. For a hypothetical optician, "Your first eye test: what happens in 20 minutes, two minutes from the main station" tells a search system more than "A day in the shop". Automatic captions are available for Shorts in many languages, German included, though quality varies[10]; correcting them keeps the business name, street and services accurate. Shorts have been able to run up to three minutes since October 2024[11], long enough to say all of this on camera.

Location can be stated explicitly. A creator can tag a video or Short with a location, which is shown on the watch page and lets viewers find other videos tagged in the same area, and YouTube may highlight places mentioned in a description, transcript or frames as "featured places"[12]. Neither feature is documented as connecting a video to a Business Profile or as affecting local ranking; both make the video's geographic context explicit.

Three ways video could affect AI visibility

No AI provider publishes how much its assistant draws on YouTube for local recommendations, and we do not claim to know. What can be described are the plausible pathways and what each would look like.

Three pathways by which video could affect AI visibility
PathwayWhat would happenWhat you would observeState of the evidence
Direct: the video is shown or cited An AI answer cites or shows the video itselfThe video's URL among an answer's citations or video resultsDocumented for Google's AI features[2]; how often it happens for local questions is not documented
Information: the video adds facts Title, description or transcript adds facts or associations ("evening appointments", "step-free entrance") that a system retrievesAnswers begin to mention attributes that appear only in the videosPlausible, not documented for local answers. Google says its generative AI features can show what is said about products and services across the web, including in videos[2], and its Gemini app can find public YouTube videos and answer questions about their content[13]; no AI provider documents whether video informs local business recommendations
Demand: the video creates interest Viewers search for, review, link to or mention the businessBranded search, reviews and mentions rise before AI answers changePlausible, through prominence-type signals such as links and reviews[7]; slow and hard to attribute

Video can be a source of AI answers, at least in Google's own AI features[2]. For a local question, though, Google's guidance points businesses to their Business Profile for visibility in AI responses[2], and no AI provider documents whether a business's own Shorts change what an assistant recommends.

The key distinction is between content an AI system cites and answers that change while a business is posting. The first is direct evidence that a video was used. The second, like the average 17-point increase in the composite MapsExperts AI Visibility Score that we observed, is an association in time, which can arise for reasons unrelated to the videos.

Can YouTube Shorts get a business recommended by ChatGPT?

That has not been shown. OpenAI documents that ChatGPT search can use third-party search providers and, for local questions, may share location with providers of local information such as nearby business listings; it does not say which sources carry weight[14]. For a local business, treat it as a hypothesis to test, with a measurement design large enough to separate a real change from noise.

Which Shorts are useful for local businesses

The test for a local Short is simple: does it show something true about a specific place that a customer would want to know before choosing or visiting it? Six types pass.

Six types of Short that pass the local test
TypeExampleWhat it can addWatch out for
The premises and how to find them "From the station exit to our door in 40 seconds": entrance, parking, step-free accessWayfinding and accessibility factsIdentifiable passers-by and neighboring properties
A product or service, shown "What a full service includes, step by step"The service described in the business's own wordsPromised results; prices that will change
Answers to real customer questions Questions that come up on the phone, in reviews and in messagesLanguage that matches how customers askInvented questions no customer asks
Specialist expertise "How we measure for a custom fitting"Evidence of competence for a specific needMedical, legal or financial claims that need review
Location-specific information Seasonal hours, parking during road construction, holiday arrangementsTime-bound practical factsStaleness: date it, avoid stating standard hours, unlist it when facts change
Procedures and customer experiences, with consent A first visit, from arrival to the endRealistic expectationsWritten consent from everyone identifiable; no health information; extra care with minors

Location-specific information needs the most care. A video that states opening hours, prices or a phone number is another copy of the business's location data, and copies drift; a companion article shows how outdated information about real places ends up in AI answers. A video library needs the discipline of a listing: an owner, a review date and a rule for updating, unlisting or removing videos.

A practical 60–90-day test

This is the design we recommend for the next round of measurement; it addresses the main weaknesses of an aggregate before-and-after observation like ours. The most useful design for a single business is a control inside the business. Choose two topics, each a group of customer decisions, of similar commercial importance and with no services, wording or season in common, such as "contact lens fitting" and "children's glasses" for a hypothetical optician. Toss a coin to decide which gets videos, so the choice does not fall on the weakest topic, which tends to recover anyway. Before publishing, pick two or three true, specific facts that will appear only in the videos, not on the website, the profile or review sites; we call them marker facts. Whether AI answers start to mention them is the most direct test a small business can run.

  1. Before week 1: set up. Two topics, each of two or three customer decisions with five to eight phrasings per topic, built as in a companion guide. Add the channel to Search Console as a platform property, note YouTube Studio baselines, and fix in advance the follow-up weeks (for example 9–12) and what change will count as meaningful.
  2. Weeks 1–4: baseline. Run every phrasing three to five times a week, on different days, on two or three AI surfaces; record Business Profile and Search Console figures. Publish no new video, or keep an existing rhythm unchanged.
  3. Weeks 5–12: publish. Two to three Shorts a week on the treated topic only, with place-specific titles and descriptions, corrected captions and the video location set; link the channel from the website and, where available, the Business Profile. Keep other work topic-neutral and log every change to profile, website, reviews and advertising. In the follow-up weeks fixed in step 1, run the panel exactly as in step 2: same phrasings, places, surfaces, account state and runs per week.
  4. Week 13: evaluate. Compare the change in the treated topic with that in the untreated one, with one confidence interval for the difference; check whether any video was cited and whether the marker facts appear; then decide whether to continue, change the content or stop.

If visibility rises equally for both topics, a topic-specific effect is unlikely, though a business-wide effect through branded searches or reviews remains possible, as do unrelated causes; this design cannot separate them. If it rises only for the treated topic, by more than the panel's noise, a topic-specific link becomes more plausible, and more so if a video is cited or a marker fact appears in answers about that topic.

This is a screen for large effects. At a baseline around 30%, with weeks 9–12 as follow-up and one run per phrasing a week, the difference between the two topics' changes would need roughly 30 to 40 percentage points per surface to be statistically significant, and about 45 to 55 to be detected reliably; running the panel three to five times a week in both periods brings the latter to roughly 20 to 35. These figures assume independent runs, so treat them as best cases, and a 60-day version is weaker still. Accept a null result as a result: learning that video brings YouTube search traffic but does not move AI visibility is worth knowing.

How to measure the impact

For the AI layers, follow our guide to measuring AI visibility for local businesses: recommendation rate per surface for each topic, plus field accuracy and critical-error rate against the business's own record, and a composite score if you track one, read alongside them. Three measures are specific to video: YouTube Studio's traffic sources, which separate YouTube search, Shorts and external sites and apps[6]; the Search Console platform property, with the channel's impressions, clicks and queries on Google Search[3]; and whether any of the business's own video URLs appears among an answer's citations, the only direct evidence that a video was used. Put a UTM-tagged link in every video description and track UTM-attributed leads per view for each business, not only in aggregate. For outcomes, Google Analytics' AI Assistant channel covers assistants such as ChatGPT and Gemini but not Google's AI Overviews or AI Mode[15], and, because it measures the website, cannot see clicks that go to YouTube instead; add Business Profile actions and a "How did you hear about us?" question at booking. None of this alone shows that video caused a change; set against the untreated topic, it shows whether a link is plausible.

What this means for multi-location businesses

For a single business, video is a question of the owner's time. For a network of hundreds of locations, it is a question of governance.

Video for a single business and for a network of hundreds of locations
TopicA single independent businessA network of hundreds of locations
Production and approval The owner, on a phoneLocal teams under central guidelines, or a central team; brand, legal and data-protection review, with consent records
Channel structure One channelWe recommend one brand channel with location playlists and consistent location identifiers in titles; separate channels per location each need an owner to stay current
Keeping videos current The owner remembersA video inventory linked to the location record, with review dates and unlisting when facts change
Measurement Before and after, with a control inside the businessLocations assigned to video at random, or in randomly ordered waves
Economics The owner's timeProduction and governance cost times the number of locations, justified only where measurement shows a gap video can plausibly close

Measurement and economics differ most. A network can assign video to locations at random, or in randomly ordered waves, and compare them with the locations still waiting, so platform changes and seasonality hit both sides alike. Compared locations should not share a catchment, and because any effect of a brand channel would reach every location, including those still waiting, the comparison captures only the location-specific effect. Where randomization is impossible, matched locations over the same weeks are the next best basis. Either is a stronger basis for an investment decision than aggregate before-and-after observations, ours included. At network scale, video is also one more channel on which location facts are published, and it belongs in the same operating model as profiles and location pages: one record, many outputs, monitored.

Before testing a new channel, check that the basics agree: the name, address, phone number and opening hours on your website should match your Business Profile and your other listings.

References

  1. YouTube Help. Search & discovery tips (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  2. Google Search Central. Optimizing your website for generative AI features on Google Search (opens in a new tab). Last updated 10 July 2026. Platform documentation.
  3. Search Console Help and Google Search Central Blog. About platform properties in Search Console; Platform properties roll out globally, plus a new social and video performance guide (opens in a new tab). July 2026; available globally since 29 July 2026. Platform documentation.
  4. Google Search Central. Video SEO best practices (opens in a new tab). Last updated 18 December 2025. Platform documentation.
  5. YouTube Help. Understand your YouTube video reach (traffic source types) (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  6. Google Business Profile Help. Tips to improve your local ranking on Google (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  7. Google Business Profile Help. Manage your social media links (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  8. Google Business Profile Help. Manage your Business Profile photos & videos (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  9. YouTube Help. Use automatic captioning (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  10. YouTube Help and YouTube Official Blog. Understand three-minute YouTube Shorts; Tall updates coming to Shorts (opens in a new tab). Blog post 3 October 2024; applies to uploads from 15 October 2024. Platform documentation.
  11. YouTube Help. Tag your location on videos and live streams; Edit video settings (featured places) (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  12. Gemini Apps Help. Find & ask about YouTube content in Gemini Apps (opens in a new tab). Accessed 9 November 2026. Platform documentation.
  13. Google Analytics Help. What's new in Google Analytics (entry of 13 May 2026); [GA4] Default channel group (opens in a new tab). 13 May 2026. Platform documentation.

How this article was sourced. Platform functionality was checked against current Google and YouTube documentation on 9 November 2026 and OpenAI's on 6 November 2026. The two observations are MapsExperts' own aggregate results across ten client businesses: UTM-attributed leads per YouTube Shorts view (views from YouTube Analytics; leads from Google Analytics 4 with UTM tracking), and the average change in the composite MapsExperts AI Visibility Score from September 2026, when the businesses published no Shorts, to October 2026, when they did (the arithmetic mean of the ten businesses' individual changes, on a 0–100 scale), measured with MapsExperts' proprietary scoring methodology, using identical questions, AI platforms, methodology and settings in both months. Individual businesses are not reported, and the score's components and weights are not published. The results come from client work, not a controlled experiment, and are not evidence of a causal effect. The thresholds in the test design are standard two-proportion difference-in-differences calculations (two-sided 5% level, 80% power) at a 30% baseline, assuming independent runs. The content types, the test design and the recommendations for networks are ours, not platform rules. Sources last checked . Research standards

Related insights

Next step

Get the location facts right before adding a channel.

MapsExperts maintains the source record, the profiles, the location pages and the monitoring behind multi-location networks, so the same current facts reach Google, Apple, Bing, your Store Locator and your own pages.

Google Business Profile management