Guide
How to improve your odds of being recommended by AI
You cannot tune the model, but you can change what it reads. This guide turns a disappointing AI answer into specific, evidence-led work: find the gap, locate the proof that would close it, improve the public evidence, and re-read to see whether the answer moves.
In short
To improve your odds of being recommended by AI, start from a specific weak answer rather than a generic checklist. Identify why you lost — a missing use case, an unclear claim, thin third-party proof, or a competitor with a stronger source — then improve the exact page, claim, or source behind it. Re-read the same question afterward to confirm the answer is moving. The work is ordinary marketing craft made measurable, not model trickery.
Who this guide is for
Teams who have already seen AI under-recommend them and want to do something about it without resorting to gimmicks. If you can name one question where a competitor wins, you have a starting point.
Start from a gap, not a checklist
Generic 'optimize for AI' advice produces busywork. Grounded work starts from evidence: the exact question, the answer that missed you, the source that shaped it, and the page or claim that needs to change.
Why this compounds
Clearer pages, better proof, and stronger sources help every future answer, not just one. Improve the evidence once and you raise your odds across the whole question set — and your own working model of the brand gets sharper too.
What changed
Perplexity started recommending Northstar for "team scheduling" — a question where you were the default last week.
Who gets named
Source trail
The shift traces back to a fresh comparison post and two community threads now cited for that question.
The loop, step by step
- 01
Pick a question you lose
Find a high-intent prompt where you are absent, misunderstood, or weaker than a competitor.
- 02
Diagnose the reason
Decide whether the gap is a missing use case, an unclear claim, thin proof, or a competitor's stronger source.
- 03
Locate the proof that helps
Identify the specific page, claim, customer evidence, or external source that would close it.
- 04
Improve the public evidence
Sharpen the page, publish the proof, or earn the third-party mention — make the brand easier to understand and verify.
- 05
Re-read and confirm
Run the same question again over the following reads to see whether the answer starts naming you.
What the loop produces
- The question you're losing
- The reason behind it
- The proof that closes it
- The page or source to change
- Follow-up movement
Improving AI recommendations, answered
- Can I make AI recommend me faster with tricks?
- Manipulation tends to be fragile and short-lived, and it does nothing for human buyers. Durable gains come from being genuinely clearer and better-evidenced — the same work that helps people choose you also helps assistants recommend you.
- How long until an answer changes after I fix something?
- It varies by assistant and by how the source is picked up. Treat it as a loop, not a switch: improve the evidence, keep reading the same question, and watch for movement over subsequent reads rather than expecting an instant flip.