Content Analysis Heuristics

Fred Leise developed eleven heuristics for qualitatively evaluating a content collection: collocation (are similar things grouped together where people would expect to find them), differentiation (can people tell dissimilar collections apart), completeness (are there gaps compared to what competitors offer, and does referenced content actually exist), information scent (do labels and headings use the audience's own language), bounded horizons (does the structure make the scope of the space clear), accessibility, multiple access paths, appropriate structure (does it match how the audience actually thinks and searches), consistency (do similar page types share a consistent structure), audience-relevance, and currency (is the content kept up to date).

Each heuristic can be rated on a five-point scale, from strongly deviating to strongly conforming, which turns a qualitative impression into something you can compare across a whole content collection and prioritize.

Eleven heuristics can feel like a lot to hold in your head at once, but in practice, most real content problems trace back to just two or three of them showing up together.

Exercise

The scenario: A technology company recently merged two separate product support sites into one. The merged site has duplicate troubleshooting articles under different titles, several articles reference a "legacy dashboard" that was retired two years ago, and articles from the two original sites use inconsistent structures, one site always led with a summary and steps, the other buried the actual fix in the middle of long narrative paragraphs.

1. Which heuristic is most directly violated by the duplicate troubleshooting articles under different titles?
2. Which heuristic is most directly violated by articles referencing a dashboard retired two years ago?
3. Which heuristic is most directly violated by the two original sites' inconsistent article structures?
Try It With Your Data: Quality Criteria Definition

Fill in your own details below; the prompt updates as you type. When it's ready, copy it into Claude or whatever AI tool you use.

What this prompt is made of (RICE breakdown)
R

Role

Casts the AI as: You are a content strategist helping define quality criteria for a content audit.

I

Instructions

Names 5 specific outputs to produce, so the response comes back structured rather than a general summary.

C

Context

My documentation covers, Primary audience and Key user goals tell the AI what your specific situation is, not a generic one.

E

Expected format

For each criterion, provide: a name, a plain-language definition, how to measure it, and an example of a page that fails it.

One or two sentences
e.g. "Backend developers, intermediate level"
What users need to accomplish
Whole number, e.g. 200
Assembled prompt