An AEO strategy can look busy and still go nowhere. The tell is not effort. It is whether AI answers name you more often this quarter than last. These are the signs that an AEO strategy is not working, and what each one points you toward fixing.
What Failure Actually Looks Like
The clearest sign is a flat Share of Model. If the share of AI answers that name you is not moving across a sixty to ninety day window, the work is not landing, no matter how much content shipped. A second sign is citations that appear and then vanish, which usually means pages are going stale faster than you refresh them.
A third sign is that you cannot say why an answer changed. If your reporting is a single score with no cited sources behind it, you are guessing. A free AI visibility scan that opens the cached answer turns guesswork into a specific list of prompts, engines, and sources to fix.
“The tell is not effort. It is whether AI answers name you more often this quarter than last.”
The real signal
The Reasons Behind the Symptoms
Most stalled programs share a few root causes. Measuring by keyword instead of by prompt, so you optimize for the wrong unit. Publishing content that buries the answer instead of leading with it, so engines cannot lift a clean passage. And ignoring off-site trust, which is where competitive answers are usually won or lost.
Another quiet killer is measuring once. Citation data is volatile month to month, so a single reading tells you almost nothing. A strategy you measure on a loop catches drift early. A strategy you measure once is a snapshot that ages badly.
How to Turn It Around
Re-baseline by prompt and engine, then fix the highest-intent answers first. Lead each target page with a direct answer, add the schema engines reward, and refresh your most-cited pages on a cadence. The background on how AI engines cite explains why these specific moves change which source a model quotes.
Then close the loop. Reading the gap does not change the answer, so the turnaround comes from shipping the fix and earning the off-site mentions, then scanning again to confirm the answer moved. A program that only reads will keep showing the same symptoms.
Measure the Change, Not the Activity
A working AEO strategy shows up as a moving Share of Model and citations that stick. If you are not seeing that, the problem is rarely effort. It is usually the unit you measure, the structure of your pages, or a missing action step. Compare your approach to a clear AEO strategy and check which part is missing.
Judge the program by the change in the answers, not by the volume of work. If the answers are not moving, change the inputs, not the output count.
“Judge an AEO program by the change in the answers, not by the volume of work.”
The honest metric
The signs that an AEO strategy is not working are easy to read once you measure the answers instead of the activity. Find the stalled prompts, fix the inputs, then scan again to prove the change.
Authoritative references: Ahrefs research on search, Search Engine Land search coverage, and Search Engine Journal guidance.