Information Quality: What Can Go Wrong?

Information does not have to be completely false to cause a bad decision. A statement can be based on a real event, contain accurate numbers or come from a genuine source and still create a misleading impression when important context is missing.

Understanding how information becomes unreliable is useful when reading websites, news articles, social-media posts, research summaries, product descriptions and AI-generated answers.

1. Outdated information

Some information has an expiration date. Laws change, software receives updates, products are discontinued, prices move, organizations change policies and scientific understanding develops.

An old statement can therefore remain factually accurate about the past while being incorrect as a description of the current situation.

This is particularly important when an answer contains current recommendations, regulations, product specifications, eligibility requirements or rapidly changing technology.

2. Missing context

A statement can be technically true but misleading when important conditions are left out.

Consider:

That statement does not tell you what users saved, how the saving was calculated, what they were compared with, how many users were studied or how long the measurement lasted.

Context determines how a reader should interpret a fact.

3. Selective information

Information can become misleading when only the evidence supporting one conclusion is presented while relevant contrary evidence is omitted.

This does not necessarily mean that the information is fabricated. The problem can instead be incomplete selection.

When a claim matters, ask whether there are relevant facts, limitations or results that have not been included.

4. Unsupported certainty

The language used to describe information can create a stronger impression of certainty than the available evidence justifies.

Compare:

The second statement may require substantially stronger evidence because it makes a broader claim.

Words such as “always,” “never,” “proves,” “guarantees” and “everyone” should be considered in relation to the evidence behind them.

5. Source drift

Information can move through the internet without preserving its original context. A statement may appear on several websites, each repeating information from another website, which ultimately came from one original source.

The number of websites repeating a claim does not necessarily represent the number of independent sources supporting it.

For important claims, trace the information backward toward the original document, study, dataset, statement or other primary material.

6. Misleading statistics

Statistics can be technically correct while being presented in a way that makes them difficult to interpret.

Potential problems include:

• very small sample sizes;
• missing comparison groups;
• relative percentages without absolute numbers;
• cherry-picked time periods;
• averages that hide large differences;
• correlation presented as causation; and
• measurements that do not match the conclusion being drawn.

When you see an impressive number, ask what the number actually measures.

7. Correlation versus causation

Two things can occur together without one causing the other.

For example, if sales increased after a company introduced a new marketing campaign, the timing alone does not prove that the campaign caused the entire increase. Other factors may have changed at the same time.

Causal claims usually require stronger evidence than observations that two events occurred together.

8. Definitions can change the conclusion

Words that sound straightforward can have technical or context-specific meanings.

Terms such as “user,” “customer,” “success,” “income,” “active,” “risk,” “effective” or “growth” may be defined differently by different sources.

Before comparing two statistics, make sure the underlying definitions are actually comparable.

9. Small samples and broad conclusions

An observation from a small or unusual group may not represent a larger population.

When a source makes a broad statement based on limited evidence, ask:

• How many observations were involved?
• Who or what was studied?
• Was the sample representative?
• What population is the conclusion supposed to describe?
• Are there limitations on generalizing the result?

10. Context collapse online

A statement can be removed from the situation in which it originally appeared and then acquire a different meaning.

This can happen when:

• a quotation is shortened;
• a video clip removes surrounding discussion;
• a chart is shared without its explanation;
• a headline is read without the article;
• a research result is summarized without its limitations; or
• an old statement is presented as current.

When something seems surprising, finding the original context can be more useful than simply searching for additional repetitions of the same claim.

11. Authority can be mistaken for evidence

A statement can come from a famous person, large organization or professional-looking website and still require examination.

Authority can be relevant, but the important question remains: What evidence supports the particular claim?

Likewise, an unfamiliar source should not automatically be considered wrong simply because it is unfamiliar. Its evidence and methodology should be examined according to the subject.

12. AI can amplify information-quality problems

AI systems can make information-quality problems harder to notice because they can produce fluent, organized answers very quickly.

An AI answer may:

• combine information from different periods;
• summarize a source incorrectly;
• omit an important qualification;
• present an interpretation as a fact;
• provide an incorrect citation;
• make a prediction sound certain; or
• fill a gap in available information with a plausible-sounding statement.

The polished appearance of an answer should therefore be separated from the quality of the evidence behind it.

For an AI-specific verification process, see How to Verify an AI Answer.

13. Repetition is not independent confirmation

One of the easiest mistakes to make online is treating repetition as confirmation.

If ten websites copy the same statistic from one press release, you still have one underlying source rather than ten independent pieces of evidence.

Look for genuinely independent evidence where independence matters.

14. What makes information easier to investigate?

Good information is not necessarily information that contains every possible detail. However, important claims become easier to examine when they provide useful anchors.

Where Supience fits

The Supience Information Check is designed around several of these observable characteristics.

It can help identify signals involving evidence, specificity, uncertainty and verifiability. This can give a reader a structured starting point for deciding what to investigate.

It does not independently determine whether the information is correct.

Check Information with Supience

A practical way to think about information quality

When information matters, consider four questions:

The goal is not to distrust everything

Information verification is not about assuming that every online statement is false. It is about matching confidence to evidence.

A simple claim supported by clear primary evidence may require little additional investigation. A surprising claim with significant consequences may deserve much more.

The most useful habit is to recognize when information deserves a closer look and then investigate the underlying evidence.

For a reusable process, see the Information Verification Checklist. For an explanation of how Supience evaluates submitted text, see the Methodology.