A customer films a 20-second story in your store. She says your brand name twice, holds up the receipt, and adds a caption sticker: "never again." There is no text post and no hashtag. Your monitoring tool, which searches captions and post text, finds nothing.
Most teams still run monitoring on a few comfortable beliefs about visual content. Some were true five years ago. Most were never quite true. This guide takes the four most common ones, checks each against the evidence, and then shows you how the technology works, where it fails, and what to test before you buy.
Four myths, four facts
Myth: If your brand is not in the caption, nobody is talking about you.
Fact: The caption is often the least informative part of the post. A reel captioned "honest review" can name your product out loud for a minute. A meme can carry your slogan in the image and a single emoji as text. A story often has no caption at all, and it disappears after a day.
Myth: Visual mentions are a niche. Your audience reads text.
Fact: Video is now a mainstream way to get news. The Reuters Institute Digital News Report 2025 found that the share of people consuming social video news rose from 52% in 2020 to 65% in 2025 across its markets. The same audience that reads about your brand also watches it.
Myth: "Image monitoring" is one feature, and every tool means the same thing by it.
Fact: It is three different technologies. OCR reads written text in pictures. Speech-to-text transcribes spoken words. Visual recognition looks for marks, objects and scenes with no words at all. Each finds a different kind of mention, so ask every vendor which of the three they run, and on which plan.
Myth: AI reads every image and every soundtrack perfectly.
Fact: Accuracy varies with fonts, resolution, background music and language. Even the widely used open-source Whisper model's documentation states that its "performance varies widely depending on the language." Plan for noise from day one.
The numbers behind the shift
All four figures come from the Reuters Institute Digital News Report 2025, which surveys news audiences across many markets.
Where the missing mentions hide
- Stories. Text stickers, overlays and spoken commentary, often with no caption.
- Reels and short video. A creator reviews your product out loud. The caption says almost nothing.
- Memes. Your name or slogan is written across the image.
- Screenshots. A customer posts your support chat, a price tag or an internal email. The words are pixels.
- Voice and video notes. In messengers like Telegram, people send round video notes and voice messages. Channels repost them. Nothing is typed.
What missing them costs you
- Volume. Your share of voice looks smaller on the platforms where your audience spends the most time.
- Sentiment. Complaints filmed in a store or shown in a screenshot never reach your negative count.
- Speed. A crisis that starts as a video gets noticed only when a news site writes about it, hours or days later.
Vendors that sell visual listening make the same point. YouScan's Visual Insights page says brands can "get up to 80% more mentions" by collecting images that feature their logo. That is a vendor claim, not an independent study, but the direction is clear.
The three techniques, explained
OCR: text inside images and video frames
Optical character recognition (OCR) reads printed or written text in a picture. In monitoring, it picks up story stickers and burned-in captions, words on memes, screenshots of chats and receipts, and signs, banners, packaging and jersey text in a photo or video frame.
Speech-to-text: spoken mentions
Speech-to-text, also called automatic speech recognition, turns audio into a transcript. It finds your brand in a creator's spoken review, a presenter on a YouTube channel, a story where someone talks to the camera, or a video note reposted in a Telegram channel.
Visual recognition: marks with no words
Visual recognition uses image models to spot a brand's mark, objects or scenes, even when no word appears. Talkwalker's image recognition page describes finding "all mentions of your logo" and measuring social impressions of brand logos. YouScan's Visual Insights recognizes logos, scenes, objects and activities; it sits on YouScan's quote-only Unlimited plan, per its pricing page.
If wordless sightings are your core need, for example sponsorship valuation, test that capability on your own images during each trial. Our Talkwalker alternatives guide compares several vendors.
How a video becomes a mention
Every visual mention goes through the same pipeline, and each step can drop it.
- Download. The crawler saves the image or video, not only the caption. Stories must be fetched before they expire.
- Split into frames and audio. The video is sampled into still frames at intervals, and the audio track is separated.
- Run OCR and speech-to-text. Frames go through OCR; audio goes through speech-to-text. The output is plain text attached to the post.
- Match your query. Your keywords, exclusions and language filters run against the caption plus the extracted text.
- Score sentiment. The combined text gets a sentiment label, so a spoken complaint counts as negative.
- Alert. If the mention matches an alert rule, or pushes volume or negativity past a threshold, your team hears about it.
Where accuracy breaks
Reading pictures and audio is harder than reading text. These are the failure modes to watch, with a fix for each.
What to do with visual mentions
- User-generated content. Unboxings, reviews and meals with your product, found even when the caption is empty. Use them to find advocates and ask permission to reshare.
- Sponsorship tracking. OCR reads your name on banners, jerseys and on-screen text. Speech-to-text catches creators and commentators who say it out loud. Count them alongside press coverage.
- Counterfeits and scams. Fake giveaways and discount codes put brand names on images with no matching text. Reading the words in the image helps you find and report them faster.
- Crisis images. Many crises start as a screenshot of a rude reply or a video of a store incident. Pair visual mentions with spike alerts, and use our crisis monitoring playbook for the response.
Checklist for your trial
Run every tool through the same test, on your own brand, not the vendor's demo.
- Collect 10 recent posts where your brand appears only in an image or only in speech, and count how many each tool finds.
- Include a story, a reel, a meme, a screenshot and, if your audience uses messengers, a Telegram video note.
- Open a found mention and read the extracted text. Can you see why it matched?
- Test every language you report on, plus mixed-script posts if your market uses them.
- Read 50 visual-only matches and count how many are really about you.
- Check that a spoken complaint is scored negative, not neutral.
- Note when a test post went live and when the tool alerted you.
- Ask whether visual monitoring is on every plan and whether it counts against your mention cap.
How Scano reads images and video
Scano is an AI media monitoring and social listening platform. It tracks news sites, social networks, messengers, blogs, forums and reviews across 1M+ sources and 20+ languages. It is strongest where English-first tools go thin: Türkiye, Central Asia, the Caucasus and the Gulf, with coverage of Telegram channels and VK as well as Instagram, TikTok and YouTube.
- OCR reads text in images and video frames: story stickers, memes, screenshots and on-screen captions.
- Speech-to-text transcribes what people say in videos and stories, including Telegram video notes.
- One search. Extracted text runs through the same queries, exclusions and smart tags as every other mention.
- AI sentiment with dedicated models for English, Turkish, Arabic, Russian, Kazakh, Persian, Hebrew and French.
- Spike signals and a daily digest to email and Telegram, when volume or negativity jumps.
- AI Analyst. Ask questions about your mentions in plain language and get answers that link to the source posts.
Learn more on the image and video monitoring page, or read our Telegram monitoring guide for how video notes and channels fit in.
What Scano does not do: it does not monitor print, TV or radio. Prices are public on the pricing page, from $417 a month billed annually, with unlimited users on every plan.
Book a Scano demo and we'll set up a trial workspace on your own brand after a short demo, so you can see which stories, reels and video notes your current tool is missing.
FAQs
What is image and video monitoring?
It is media monitoring that looks inside images and videos, not only at their captions. It uses OCR to read written text, speech-to-text to transcribe spoken words, and in some tools visual recognition to spot marks with no words.
Can monitoring tools read text in Instagram stories?
Some can. A tool needs to download the story before it expires and run OCR on its frames. Scano reads text inside stories with OCR and transcribes spoken words with speech-to-text.
What is the difference between OCR and visual recognition?
OCR reads letters and words, so it finds your brand when the name is written. Visual recognition looks for shapes, marks and objects, so it can match an image with no text. They solve different problems, so test each one separately.
How accurate is speech-to-text for brand monitoring?
It depends on the language, the audio quality and the background music. Accuracy is usually lower for less-resourced languages and noisy reels. Test each tool on your own videos and keep a review queue for visual-only matches.
Do visual mentions count toward my mention limit?
That depends on the vendor, so ask during the trial. Some tools reserve visual features for their top plan or price them separately. Check both before you compare prices.