Author
A chart image alone can hide the scale, units and qualifications a reader needs to interpret it. This review separates a short identifier from a longer textual account, records axes, scale and source coverage before writing, and keeps the description inside what the chart actually displays. A complete description does not establish crawling, indexing, citation or correct AI interpretation.
When a business page provides a chart only as an image, we check which explanatory details remain available outside it. The axes, the units, the legend and the note about which period or population the data covers may all sit inside that image, and a reader who receives only the surrounding prose can miss them. The review question is therefore not how to make the picture look better, but which information disappears when only the image remains.
The accessibility tutorial treats charts as complex images and distinguishes a brief identifying description from a fuller text account of their essential content. We use that distinction to separate identification from explanation, within the tutorial’s accessibility scope. That framing is about accessibility, not about eligibility for any AI feature, and it should be read within that scope.
Applied to a chart, the distinction gives the review its first boundary. The short identifier says what the chart is. The longer account says what a reader must know to interpret it. Neither part is a place to add a conclusion the chart does not support, and neither part substitutes for the other.
Before writing the longer account, record what the approved chart actually contains. The table below is a proposed review record, not an observation about any published chart. Its fields are recommendations, and every verification cell is left as a state to be filled from approved material rather than a finding.
| Field | What it records | Verification state |
|---|---|---|
| Chart identity | The title or subject that distinguishes this chart from others on the page | To be confirmed against the approved source |
| Axis labels | The label on each axis, including the category or measure named | To be confirmed |
| Scale and units | The scale used and the unit each value is expressed in | To be confirmed |
| Series and legend | Which series each visual mark belongs to and how the legend maps them | To be confirmed |
| Source coverage | The period, population or condition the chart covers | To be confirmed |
| Displayed relationship | The relationship the chart visibly shows, stated without cause | To be confirmed |
| Unsupported inference | Any conclusion a reader might draw that the chart does not establish | Open until resolved |
The scale and units row is not administrative detail. The tutorial’s example of a long description points to scales, values, relationships and trends as the kind of information a chart encodes visually, which is why a description that omits the scale can leave a reader with a shape but no way to read it. That example belongs to the tutorial’s own documentation; it is not evidence about any chart on a business site.
The last row matters as much as the first. Writing down the inference the chart does not support keeps the description from drifting into explanation during drafting, when a plausible cause is easiest to add.
With the record in place, we decide where readers can find the essential information. We use the tutorial’s short/long distinction to keep the image identifiable and its fuller explanation available as reachable text.
The cited search guidance points in the same direction for a different reason: it lists making sure that important content is available in textual form among its best practices. That guidance is written for the cited search service and does not describe how every AI system parses a page, so it should not be generalized into a universal rule.
In practice, the connection between the two parts should be explicit to a reader. A short identifier that names the chart, followed by a visible heading or sentence that introduces the longer account, lets someone move from the picture to the explanation without guessing where it is. The tutorial discusses several ways to associate the two, including a text link near the image and a structural grouping of image and description; the choice is an implementation decision, not a fixed requirement.
We would resist turning this into an alt-length rule. The tutorial does not set a universal character limit for either part, and a description that is long enough to carry the scale but short enough to identify the chart is a judgment about this chart, not a standard to apply everywhere.
Related reading: Image links and alternative text
An unsupported added conclusion is a risk we check in a chart description. A line that rises can invite a sentence about why it rose, and a bar that is taller can invite a claim about superiority. Neither is in the chart unless the approved source says so.
The review step is to check each statement in the longer account against the record built earlier. Does the description name the scale the chart uses? Does it keep each value with the category and condition the chart assigns it? Does it describe the relationship the chart displays without asserting a cause? The tutorial’s example of a long description covers scales, values, relationships and trends, which is a reminder of what the description is meant to represent rather than a licence to explain them.
Where the chart does not settle a question, the description should say so rather than fill the gap. A missing value, an unreadable axis or an ambiguous category is an unresolved state to record, not an invitation to estimate from the picture. This is a recommendation about drafting discipline; it does not claim that the tutorial proves any statistical point.
Charts get revised. A series is added, a period is extended, an axis is rescaled, or the image is replaced while the surrounding text stays as it was. When that happens, the image and the explanation can describe different versions of the same chart, and a reader has no way to tell which one is current.
The cited search guidance asks for structured data to match the visible text on the page. The same consistency logic applies to a chart and its description: the two should be read against the same source version, and a change to one should trigger a check of the other.
We would treat three outcomes as separate unresolved states rather than one failure. A missing source means the approved chart cannot be found to compare against. An unreadable axis means the chart is present but its scale or labels cannot be confirmed. An unsupported conclusion means the description asserts something the chart does not show. Each points to a different next action, and collapsing them into a single "needs review" label hides which one applies.
A complete chart description establishes something narrow and useful: the essential information in the image is now available as text, with its scale, units and coverage stated and its unsupported inferences left out. That is a real improvement for readers who cannot see the image and for anyone reading the page without it.
It does not establish that a search engine or AI system will crawl, index or cite the page. The cited search guidance is explicit that meeting its requirements, best practices and policies does not mean the cited service will crawl, index or serve the content. Accessibility and readability are distinct from AI interpretation and citation, and a good description should not be presented as a visibility tactic.
If the underlying page structure cannot carry a stable, readable chart and its explanation, that is a publishing problem worth fixing first. SHMLANG describes a CMS-ready publishing structure intended to let future SEO and GEO work build on a stable base rather than restart each time. That is a published capability description, not a delivered result.
For teams that want to check how their pages are actually described, SHMLANG’s published GEO scope connects question baselines, approved business facts, useful pages and source research. Applied to charts, that would mean starting from the approved chart and its source, writing the description against it, and reviewing the result before treating the page as ready.
A practical next step is to pick one chart on a priority page and run the record above against it. If the scale, units or coverage cannot be confirmed from approved material, the description is not ready to publish, and the open question belongs with the owner of the chart rather than in the prose.
No comments yet. Be the first!