What to fix first after an audit comes back red and yellow

An audit report with forty findings is not a to-do list. It is raw material for one. Hand a client the raw version and they will either fix the wrong thing first or fix nothing, because everything is marked urgent.

Color coding does not sequence work. Red and yellow tell you severity, not order. A finding can be severe and still belong in month three. A finding can look minor and belong in week one. You need a second axis: not "how bad is this" but "what does fixing it unlock."

The test that actually sorts findings

Ask one question per finding: can this business be considered at all for the thing it wants, regardless of how good it is?

If the answer is no, you have an eligibility blocker — something that keeps the business out of consideration entirely, independent of quality. If the answer is yes but the business loses to competitors who do it better, you have a competitiveness gap. If the answer is yes and fixing it only tidies the edges, you have a nice-to-have.

This is a triage test, not a quality score. A business can have excellent content and still be functionally invisible because of one blocker. Fix the blocker first, every time — a stronger competitive position is worthless if the business isn't in the running at all.

Eligibility blockers: week one, no exceptions

These share one trait: nothing else on the list matters until they're fixed, because they remove the business from consideration before quality is even evaluated.

• A crawler you need is disallowed. If robots.txt blocks Googlebot, or blocks a specific AI crawler whose product you're trying to appear in, the pages behind that block are not being evaluated at all. This is not a ranking penalty — it's non-participation.
• Name, address, or phone conflicts across sources that still get cited. If the website says one phone number and the Google Business Profile says another, or a major directory has a defunct address, you're giving ranking and answer-generation systems contradictory evidence about which entity you even are. Fix the disagreement before worrying about how well the entity performs.
• The Google Business Profile is missing, unclaimed, suspended, or has no primary category set. A profile that doesn't exist, or exists without a category, cannot compete for local pack placement for the relevant queries. This is table stakes, not an optimization.
• The site is not indexable. Blocked by a stray noindex, unreachable sitemap, or broken canonical pointing somewhere else — these mean pages are absent from the index they need to be evaluated in, not weak within it.

None of these fixes make a business rank better than a strong competitor. They make the business eligible to be compared to that competitor in the first place. That's why they come first — not because they're the most severe-looking line item, but because everything downstream depends on them.

Competitiveness gaps: real work, not week one

Once the business can actually be considered, the next question is whether it wins the comparison. These findings are usually not fixed in an afternoon, and that's fine — they belong in the four-to-twelve-week window, not the first sprint.

• Thin service or location pages. The page exists and is indexable, but it says less than what a competitor's equivalent page says, or doesn't address the specific service or area at all. This is a competitiveness problem, not an eligibility one — the page is in the race, it's just behind.
• Low AI-engine citation or mention rate on category and comparison prompts. If the business is eligible (not blocked, NAP is consistent, content exists) but answer engines routinely name competitors instead, that's a signal about relevance and corroboration, not access. Read it by intent and by engine before deciding what to fix — a low rate on comparison prompts and a healthy rate on branded prompts point at different work.
• Weak review velocity or volume relative to the local competitive set. Reviews aren't binary — a business with twelve reviews isn't ineligible, it's just less competitive against one with two hundred.
• Missing schema that describes the business but isn't required for basic participation. LocalBusiness markup that's absent or incomplete falls here — it helps systems understand the entity with more confidence, but its absence doesn't disqualify the page the way a blocked crawler does.

Sequence these by leverage, not by list order. A category page rewrite for the service that actually drives revenue usually outranks a schema cleanup on a page nobody queries.

Nice-to-haves: don't let these steal week one

These are legitimate findings. They are also the ones most likely to eat a sprint that should have gone to a blocker, because they're fast and satisfying to close.

• FAQ content or markup polish on pages where the content already exists and is accurate
• Photo count and freshness on an already-complete Google Business Profile
• Minor alt-text or heading-structure cleanup on pages that are already indexable and reasonably competitive
• Posting cadence on a profile that already meets category and completeness basics

If a nice-to-have and a blocker are both on the list, the nice-to-have waits. Not because it's unimportant forever — because doing it first doesn't change whether the business can be found.

Putting it on a calendar

A rough shape that holds up across most audits:

1. Week one: every eligibility blocker. Crawler access, NAP conflicts, GBP claim/category/completeness, indexability.
2. Weeks two through eight: the competitiveness gaps ranked by which query intent or revenue line they affect most.
3. Ongoing, lower priority: nice-to-haves, folded into normal maintenance rather than a dedicated sprint.

Re-run the audit after week one specifically to confirm the blockers actually cleared — a robots.txt edit that didn't deploy, or a GBP category change still pending review, will quietly stall everything scheduled after it.

None of this promises a ranking outcome. It promises that the business is competing on the same field as everyone else being evaluated, which is a precondition for the rest of the work to mean anything.