Text and visual consistency
Keep the visual and the text within the same claim.
Distinguish evidence from illustration. Check labels, scope and dominant impression, then repair any visual that implies an unsupported result.
Free learning from Semantec SEO.
Write the claim before choosing the visual.
Text and imagery should describe the same entity, scope and evidence. Start with the direct answer and its supporting record. Then give the visual a distinct explanatory job: show a relationship, sequence, comparison or spatial arrangement that the text also explains. If the visual adds an outcome the words and source cannot support, repair the visual rather than relying on a quiet disclaimer.
A photograph may document a particular object or event when its provenance supports that use. A generated illustration is conceptual, not proof that a project happened. A chart is a quantitative claim whose scales, units, population and source need inspection. Those truth states must stay distinct even when all assets share the same design palette.
This page owns agreement between the visual and verbal claims. File delivery, crawl access and image indexing have their own technical review. A perfectly delivered image can still misrepresent the evidence, while an accurate explanation still needs a reliable delivery path.
Inspect what the visual invites the reader to infer.
Read beyond explicit labels. An upward arrow can imply improvement, a before-and-after photograph can imply a causal result, and a scale can imply a measured comparison. Cropping can hide an exception; different axis ranges can exaggerate a change. Check that the dominant impression agrees with the bounded claim, not merely that a note somewhere contradicts it.
For a chart, identify the measured variable, unit, denominator, population, time period, source and uncertainty. Inspect the baseline and comparison method. A count of completed editorial fields does not establish ranking, revenue or customer experience. Correlation over time does not by itself identify the cause of a change.
For an illustration, state its conceptual role where confusion is likely. Do not describe a generated office, person or screen as documentary proof of an organization, employee or deployed product. Visual polish should not change the evidence class.
Repair a chart that invents a commercial outcome.
This fictional training scenario supplies a finite record: ten draft briefs were reviewed before an editorial repair, and ten after it; the same four checks defined completeness at each stage. Three briefs met all four checks before review, and eight after. These numbers are invented solely for this lesson. No traffic, customer, conversion or revenue measurements exist.
The faulty chart headline is “Our method multiplies leads.” Its vertical axis says “Leads,” and bars use 3 and 8. The accompanying paragraph only says that the brief review found more complete records. Relabeling editorial counts as leads changes the variable and claims an unsupported business effect. A footnote saying “results may vary” would not fix that mismatch.
The repair changes the headline, unit, legend and surrounding explanation to describe brief completeness only. It retains the stated denominators and finite scope. It removes the lead and causal claims. The corrected native-text chart below uses the fictional counts without presenting them as business results.
Fictional brief completeness review
Complete briefs among ten reviewed at each stage; illustrative training data.
Both bars count briefs meeting all four stated checks. Neither bar represents traffic, leads or revenue.
| Stage or measure | Complete briefs | Reviewed briefs | Evidence status |
|---|---|---|---|
| Before editorial review | 3 | 10 | Invented training record; all four checks required |
| After editorial review | 8 | 10 | Invented training record; same stated checks |
| Traffic | Not measured | Not applicable | No source; no effect claim |
| Leads or revenue | Not measured | Not applicable | No source; commercial inference rejected |
The four completeness checks in this fictional record are named reader task, assigned page owner, attached source and written acceptance condition. A brief counts as complete only when all four are present. This measures those recorded fields, not the quality of every answer or an improvement caused by a specific method. No uncertainty estimate, real study or generalizable effect is supplied.
Keep captions, legends and explanations in their proper roles.
A caption identifies the visual and can name its scope or source. A legend explains symbols, series or encodings. Neither should carry a qualification that the main visual impression contradicts. Keep units and denominators visible beside the relevant comparison, with the same definitions used in the paragraph and data table.
In the repaired example, the title identifies brief completeness, the legend says both bars count briefs passing the same four checks, and the prose states that the data is fictional. The native table exposes missing traffic and commercial evidence. Color is decorative: the stage labels, counts and common scale communicate the comparison without requiring color recognition.
For ordinary conceptual illustrations, a visible caption may add nothing useful. An accurate short alt description can identify the pictured objects while the explanatory meaning stays in nearby text. A complex chart or diagram needs a fuller equivalent rather than an overloaded alt sentence. W3C’s complex-image guidance describes that distinction.
Use a pairing record to bind the two layers.
This populated static record retains the useful decision fields from the image-and-text pairing method. It describes the fictional repair, not a measured Semantec result.
| Field | Accepted content | Blocked inference |
|---|---|---|
| Question and answer | How many briefs met the four recorded checks at each stage? Three of ten, then eight of ten. | No answer about leads or traffic |
| Visual job | Compare two counts under one denominator and definition | No growth projection or causal attribution |
| Shared terms | Complete brief, four checks, ten reviewed, before/after review | No silent substitution of performance for completeness |
| Source and scope | Invented training record, not a real customer study | No documentary or external proof claim |
| Legend and equivalent | Native stage labels, exact counts, full table and explanation | No reliance on color or a small disclaimer |
| Owner and review trigger | Editor approves claim; designer matches encoding; reviewer checks delivered pairing | Reopen if source, denominator, population or meaning changes |
When the source cannot support a proposed visual, choose among removing the claim, replacing the asset, qualifying it accurately or acquiring the missing evidence. Do not keep a misleading centerpiece simply because it has already been produced. A visual with no explanatory job can be omitted.
Check the pairing after crop, layout and reuse.
Inspect the final asset in its real placement. A narrow crop can remove a scale, label or limiting condition. A legend moved far from the comparison can change interpretation. Reuse can detach a fictional example from the paragraph that identifies it as fictional. Preserve those relationships wherever the asset appears.
Ask a reader what the visual claims before showing the source record. If the answer differs from the supported claim, investigate the encoding and emphasis. Record the observation and repair, without presenting a small review as universal proof. Check the text equivalent, meaningful order and non-color cues separately.
W3C’s color criterion provides one access boundary. It does not certify the truth of a chart. Google’s editorial usefulness questions also do not establish a chart’s data provenance or promise a search display.Maintain the evidence and the visual as one record.
Keep the source, variable definitions, scope, version, owner, truth status and review trigger with the accepted text and visual. A revised paragraph may require a new chart, and an updated chart may require different explanatory language. Review both when facts, units, dates, population or intended use changes.
Preserve unresolved evidence as a hold, not as confident decorative copy. Before reuse, confirm that the new placement retains the source and conditions. A visually consistent illustration series can help the theme while leaving every substantive assertion in ordinary readable text.
Return to the visual meaning and usable pages hub for hierarchy and accessible explanation. The learning overview connects this review to source context and evidence.