Latest / Marketing Automation in 2026 / Multimodal Readiness: Images, Alt Text, and Diagrams that Models Can Cite

Transcript

Multimodal Readiness: Images, Alt Text, and Diagrams that Models Can Cite · Marketing Automation in 2026

• 0:00
  SPEAKER_00: Multimodal Readiness, images, alt text, and diagrams that models can cite.
• 0:05
  SPEAKER_00: By Lucas Richter, senior content automation strategist.
• 0:09
  SPEAKER_00: A diagram can explain in seconds what a page of text would take minutes to describe.
• 0:13
  SPEAKER_00: But if its meaning lives only inside an image file, many readers and many search systems may miss the point.
• 0:20
  SPEAKER_00: That is why multimodal readiness is not about adding more pictures.
• 0:24
  SPEAKER_00: It is about packaging each useful visual so its meaning is clear in the image, in the page text, and in the code that connects the two.
• 0:32
  SPEAKER_00: There is one important limit.
• 0:34
  SPEAKER_00: Good alt text may help a system understand an image, but it does not guarantee that an AI assistant will cite the page.
• 0:41
  SPEAKER_00: Understanding an image and choosing a web page as a source are different steps.
• 0:46
  SPEAKER_00: What the research does and does not show, captions can help with visual questions.
• 0:51
  SPEAKER_00: Some research finds that text paired with an image can help a model answer questions about it.
• 0:57
  SPEAKER_00: A 2019 study found that adding paragraph style captions to images improved performance on a visual question-answering task.
• 1:05
  SPEAKER_00: The captions supplied details that were easier to match to a question than image features alone.
• 1:11
  SPEAKER_00: But this was a model experiment, not a test of whether adding alt text to a live web page increases AI citations.
• 1:19
  SPEAKER_00: ACL anthology paper, charts, and diagrams also remain hard for models to read precisely.
• 1:26
  SPEAKER_00: In one chart question benchmark, researchers found weaknesses in tasks such as finding relationships between values.
• 1:34
  SPEAKER_00: Their results improved with a special text prompt, which points to the value of clear language, but does not prove that any one-page format will earn more citations.
• 1:45
  SPEAKER_00: Chart Insights benchmark.
• 1:46
  SPEAKER_00: The broader lesson is practical.
• 1:48
  SPEAKER_00: A visual can carry useful evidence, but models may still miss small labels, exact values, or the path between steps.
• 1:57
  SPEAKER_00: OpenAI's vision guidance, for example, warns about small text, charts, color-coded elements, and precise spatial reasoning.
• 2:05
  SPEAKER_00: OpenAI Vision Guide.
• 2:07
  SPEAKER_00: Better alt text is not a citation switch.
• 2:10
  SPEAKER_00: Search systems may use image descriptions and surrounding text to understand what a picture shows.
• 2:16
  SPEAKER_00: Google says it uses alt text, computer vision, and page content together when understanding images for search.
• 2:24
  SPEAKER_00: It also recommends placing images near relevant text.
• 2:27
  SPEAKER_00: At supports writing useful alt text, but it is not evidence that alt text by itself makes an AI assistant cite a page.
• 2:36
  SPEAKER_00: Google Image SEO Guidance.
• 2:38
  SPEAKER_00: The evidence reviewed for this article does not establish that changing alt text alone raises citation rates across current AI assistants.
• 2:48
  SPEAKER_00: Many systems do not explain exactly which page elements they read or why they choose one source over another.
• 2:56
  SPEAKER_00: Google says its AI search features have no special markup requirement and do not guarantee that an eligible page will be shown.
• 3:04
  SPEAKER_00: Google's AI features guidance.
• 3:07
  SPEAKER_00: So treat improved visual packaging as a reasonable, testable hypothesis, not a promise.
• 3:13
  SPEAKER_00: It can make a page easier to use, easier to parse, and more likely to contain text that directly answers a question.
• 3:21
  SPEAKER_00: Whether an assistant then cites it is a separate result to measure.
• 3:25
  SPEAKER_00: Package each important visual as evidence.
• 3:28
  SPEAKER_00: A strong visual package has four parts the image, concise alt text, a caption, and a nearby text explanation.
• 3:36
  SPEAKER_00: Each part has a different job.
• 3:39
  SPEAKER_00: Write alt text as a short replacement.
• 3:41
  SPEAKER_00: Alt text should give readers who cannot see an image the information they need to follow the page.
• 3:47
  SPEAKER_00: For a simple diagram, that may fit in one sentence.
• 3:50
  SPEAKER_00: For a complex chart or flow chart, alt text should identify the visual and its main purpose, not squeeze every detail into one long description.
• 4:00
  SPEAKER_00: For example, alt, checkout flow with a retry path after payment failure.
• 4:06
  SPEAKER_00: For a decorative image, use empty alt text so assistive tools can skip it, do not leave alt text out, and do not fill it with repeated search terms.
• 4:16
  SPEAKER_00: W3C's complex image guidance and Google's image guidance distinguish short text alternatives from fuller descriptions.
• 4:24
  SPEAKER_00: Use a caption to state the figure's point.
• 4:26
  SPEAKER_00: A caption should help a reader understand why the image matters.
• 4:31
  SPEAKER_00: A label such as architecture diagram is usually too vague.
• 4:35
  SPEAKER_00: A more useful caption is Figure 1.
• 4:38
  SPEAKER_00: A failed payment sends the shopper back to payment, the cart stays open.
• 4:43
  SPEAKER_00: Use the HTML, figure, and fig caption elements to connect the caption to its image.
• 4:48
  SPEAKER_00: A clear figure number or name also makes it easier to refer to the same visual in the article text.
• 4:55
  SPEAKER_00: Google's Documentation Style Guide gives examples of figures paired with alt text, captions, and longer descriptions.
• 5:03
  SPEAKER_00: All, put the full explanation in the page.
• 5:05
  SPEAKER_00: A complex diagram usually says more than a short alt description can.
• 5:09
  SPEAKER_00: Put the important meaning in ordinary page text too.
• 5:13
  SPEAKER_00: For a flowchart, describe the steps and what happens at each branch.
• 5:17
  SPEAKER_00: For a chart, include the time period, units, key values, and the trend.
• 5:22
  SPEAKER_00: For a technical diagram, name the parts and explain how they connect.
• 5:27
  SPEAKER_00: W3C recommends a short description, plus a fuller text description for complex images.
• 5:33
  SPEAKER_00: That fuller description can be in the article itself, near the image and available to all readers.
• 5:39
  SPEAKER_00: A data table can help when exact values matter.
• 5:43
  SPEAKER_00: W3C, WAI, Complex Images, keep names and labels consistent.
• 5:49
  SPEAKER_00: If your diagram calls something activation, use that same word in the caption and article.
• 5:54
  SPEAKER_00: If the figure shows a result that matters, state it in sentence nearby.
• 5:59
  SPEAKER_00: In Figure 2, most signups drop off before the first project is created.
• 6:04
  SPEAKER_00: This is not a magic citation trick, it is a way to make the page's evidence easier to follow and quote.
• 6:10
  SPEAKER_00: Google says, there is no special schema for AI overviews or AI mode.
• 6:15
  SPEAKER_00: The useful work is keeping important content in text and making the structured data match what people can see.
• 6:21
  SPEAKER_00: Oh, here is a simple pattern for a flowchart.
• 6:24
  SPEAKER_00: The short alt text identifies the diagram, the caption gives its main point.
• 6:29
  SPEAKER_00: The last sentence explains the path in plain text.
• 6:32
  SPEAKER_00: For a chart with exact values, add a table or list of those values to.
• 6:37
  SPEAKER_00: Make sure crawlers can fetch the image.
• 6:39
  SPEAKER_00: A well-described image still needs to be available.
• 6:42
  SPEAKER_00: Use a normal HTML image element with a real source URL.
• 6:47
  SPEAKER_00: Google says it can find images in an image element's SURS, but it does not index CSS background images in the same way.
• 6:55
  SPEAKER_00: Its image guidance also supports CDN-hosted images in image sitemaps.
• 7:00
  SPEAKER_00: Google Image SEO Guidance.
• 7:02
  SPEAKER_00: For a CDN, check the basics.
• 7:05
  SPEAKER_00: The image URL works without a login or an expiring access token.
• 7:09
  SPEAKER_00: The page's crawler permissions and the CDN's security rules both allow access.
• 7:14
  SPEAKER_00: The image URL in the rendered page is stable and returns the correct file.
• 7:19
  SPEAKER_00: Responsive images include a fallback source.
• 7:22
  SPEAKER_00: Important text is not hidden in an image when it could be real page text.
• 7:26
  SPEAKER_00: Below the fold images load when they become visible, not only after a person clicks or scrolls in a way a crawler cannot reproduce.
• 7:35
  SPEAKER_00: Google's Lazy Loading Guidance says image URLs should appear in the rendered HTML and load when they come into view.
• 7:42
  SPEAKER_00: It also advises against lazy loading images that are already visible when the page opens.
• 7:48
  SPEAKER_00: Google's Lazy Loading Guidance.
• 7:51
  SPEAKER_00: Check permissions by Platform 2.
• 7:53
  SPEAKER_00: OpenAI documents OAI SearchBot as its search crawler and says it is separate from GPT bot, which relates to training data.
• 8:01
  SPEAKER_00: Allowing one does not require allowing the other.
• 8:04
  SPEAKER_00: This matters if you want search discovery but have a different preference about training use.
• 8:09
  SPEAKER_00: OpenAI crawler documentation.
• 8:11
  SPEAKER_00: Test whether your pages get cited for visual questions.
• 8:15
  SPEAKER_00: Do not rely on a few casual chatbot prompts.
• 8:18
  SPEAKER_00: Run two tests because model comprehension and web citation are not the same thing.
• 8:24
  SPEAKER_00: Test 1.
• 8:25
  SPEAKER_00: Does the model understand the visual?
• 8:27
  SPEAKER_00: Choose a set of charts and diagrams.
• 8:29
  SPEAKER_00: Write questions whose answers depend on them, such as, which step follows payment?
• 8:35
  SPEAKER_00: What happens when payment fails?
• 8:36
  SPEAKER_00: Which month had the highest conversion rate?
• 8:39
  SPEAKER_00: What does the diagram say is the main reason users drop off.
• 8:43
  SPEAKER_00: Then compare model answers under controlled input conditions.
• 8:47
  SPEAKER_00: The image alone.
• 8:48
  SPEAKER_00: The image plus its alt text and caption.
• 8:51
  SPEAKER_00: The image plus those annotations and the page's text summary or data table.
• 8:56
  SPEAKER_00: Keep the model, question, and image the same.
• 8:59
  SPEAKER_00: Score answers for correctness, including exact values and relationships.
• 9:04
  SPEAKER_00: This helps show whether added text helps the model answer your questions.
• 9:09
  SPEAKER_00: It does not show whether a web search assistant will find or cite the page.
• 9:14
  SPEAKER_00: Test 2.
• 9:16
  SPEAKER_00: Does a search assistant cite the page?
• 9:19
  SPEAKER_00: For a small pilot, pair similar diagram-heavy pages.
• 9:23
  SPEAKER_00: Add the full visual package to one group, leave the comparison group at its current accessible baseline.
• 9:29
  SPEAKER_00: Do not remove essential alt text to create a control group.
• 9:33
  SPEAKER_00: Prepare two sets of questions, ones that require reading a figure, and ordinary questions that the page's text already answers.
• 9:42
  SPEAKER_00: Run the same prompts in the search assistance you care about.
• 9:45
  SPEAKER_00: Record the date, product, question, answer, and cited page.
• 9:51
  SPEAKER_00: Repeat the checks over time because results can change.
• 9:54
  SPEAKER_00: Wait until your crawler or search tools show that the updated pages have been recrawled.
• 9:59
  SPEAKER_00: Google notes that recrawling can take days or longer.
• 10:03
  SPEAKER_00: Track four outcomes.
• 10:05
  SPEAKER_00: Citation rate.
• 10:06
  SPEAKER_00: How often does the answer cite the page?
• 10:09
  SPEAKER_00: Answer accuracy.
• 10:10
  SPEAKER_00: Did the assistant get the visual fact right?
• 10:13
  SPEAKER_00: Citation support.
• 10:14
  SPEAKER_00: Does the cited page actually support the answer?
• 10:18
  SPEAKER_00: Business result.
• 10:19
  SPEAKER_00: Did the visit, referral, or lead help the business?
• 10:22
  SPEAKER_00: Compare visual question results with the ordinary question control set.
• 10:27
  SPEAKER_00: A simple before and after comparison can be misleading.
• 10:30
  SPEAKER_00: The assistant, its search index, and the wider web may all change at once.
• 10:36
  SPEAKER_00: Matched pages and repeated checks help, but a small test should still be reported as a pilot, not a universal rule.
• 10:43
  SPEAKER_00: Also be precise about what the result means.
• 10:46
  SPEAKER_00: If a page gets cited more often after you add a text summary, the test shows that the whole package helped.
• 10:53
  SPEAKER_00: It does not prove the assistant read the image itself.
• 10:56
  SPEAKER_00: Citation dashboards can show which URLs appear in AI answers, but not necessarily which part of a page was read.
• 11:04
  SPEAKER_00: Bing's AI Performance Report, for example, reports cited pages and query information.
• 11:10
  SPEAKER_00: Bing Webmaster Tools, AI Performance, build the work into your content system.
• 11:16
  SPEAKER_00: Founders should not have to rewrite a diagram by hand every time they publish an article.
• 11:21
  SPEAKER_00: But automation should not invent what the picture means.
• 11:24
  SPEAKER_00: A simple workflow is the creator provides the source data or diagram plus the main point they want the reader to take away.
• 11:32
  SPEAKER_00: A content workflow drafts the alt text, caption, and plain text summary from those source materials.
• 11:39
  SPEAKER_00: A person checks that the description is accurate and that labels, values, and routes match the visual.
• 11:46
  SPEAKER_00: The page publishes the image and its text together.
• 11:49
  SPEAKER_00: A repeatable check confirms that readers and crawlers can access both.
• 11:54
  SPEAKER_00: The creator's point is the part that should not be automated away.
• 11:58
  SPEAKER_00: For example, a founder may want to show not just that signups fall, but that the team chose to fix first project setup before buying more traffic.
• 12:07
  SPEAKER_00: That is an opinion and a product lesson, not a captioning task.
• 12:12
  SPEAKER_00: In an automated content engine, the creator, audience, and taste inputs can preserve that point across articles.
• 12:19
  SPEAKER_00: Autopod uses those inputs to guide each article while the workflow handles repeatable tasks like drafting captions and checking page structure.
• 12:28
  SPEAKER_00: The model can package the insight.
• 12:30
  SPEAKER_00: It should not make up the insight.
• 12:32
  SPEAKER_00: Conclusion: The strongest case for descriptive alt text, captions, figure labels, and nearby summaries is not a promised SEO lift.
• 12:41
  SPEAKER_00: It is that they make a visual easier to understand, easier to access, and easier to connect to the page's argument.
• 12:49
  SPEAKER_00: That may also help an assistant answer a visual question and cite the page.
• 12:54
  SPEAKER_00: But the lift is not guaranteed.
• 12:56
  SPEAKER_00: Package diagrams as evidence make the image fetchable and measure citations separately from comprehension.
• 13:03
  SPEAKER_00: That is a simple system with value for readers first and a testable upside for AI discovery.
• 13:10
  SPEAKER_00: All links to sources are available in the text version of this article.
• 13:14
  SPEAKER_00: You can find the full article at autopod.co.
• 13:18
  SPEAKER_00: Thanks for listening.