Why is my citation rate low?

A low rate has a small number of causes and they fail in a fixed order. Work down them in that order, because fixing content when the problem is access wastes a month.

What you'll learn

• The five causes, in the order to check them
• How to tell which one you have
• What actually moves the number

First, make sure it is low

Two things look like a low rate and are not:

• Exclusions counted as zeroes. An excluded surface has no measurement. Read the rows, not the average.
• A thin window. A rate over very few checks moves violently. Check the window in the footer.

Cause 1: Crawlers cannot reach you

The crawlers that fetch pages in order to cite them are not the same as the ones that gather training data, and blocking the first group removes you from answers entirely.

This is two lines in a file, it is invisible in every other analytics tool you own, and it is the most common single cause. Check Access in Readiness first, every time.

Cause 2: Your content cannot be extracted

Engines lift passages, not pages. The usual failures:

• The content only exists after JavaScript runs. Most AI crawlers do not execute it.
• The answer is implied across a page rather than stated in a sentence.
• The page is a brochure: adjectives, no facts. There is nothing to quote.

Check Structure in Readiness.

Cause 3: No reason to trust you

Trust is the tiebreaker between similar businesses. Machine-readable contact details, links tying your entity to its other profiles, review markup, freshness, and a page saying who is behind the business.

Check Authority in Readiness.

Cause 4: The prompts are wrong

A rate is only as meaningful as the questions behind it:

• Too broad. "Best software" produces an answer that names nobody.
• Wrong place. A local prompt without your market in it measures somebody else's city.
• Wrong intent. A set of informational prompts gives you a rate that says little about revenue. See intent classification.

Cause 5: Somebody else answers better

Once one to four are sound, this is the real work. Read two answers naming a rival in Competitors, then look at the page feeding them in Sources.

Usually they have done one thing you have not: stated who the thing is for and who it is not for, in a sentence, near the top.

What moves the number

In rough order of impact per hour spent:

1. Unblocking a crawler
2. Making an existing page's answer extractable
3. Getting onto a third-party page engines already lean on
4. Adding the qualifier: who it is for, who it is not for
5. Writing new pages

Most teams start at five.

Do it through the ledger

Approve each change in the platform before you ship it, so a control is captured and you find out whether it worked. See ship and prove a change.

Try this in Outercite

Ask the assistant "Is my website ready for AI answers?" and fix the first real failure it names before touching anything else.

Related