What's the highest-leverage SEO action I can take this week?
Short answer
Fix the highest-revenue page that's currently broken, and if nothing is broken, improve the page ranking in positions 4 to 10 for your most commercial query. Leverage is revenue at stake divided by hours to fix, and the winner is nearly always an existing asset you're losing money on rather than something new. Publishing a new article is almost never the answer to a one-week question.
Score candidates properly
Rank every candidate on one number: estimated revenue at stake divided by hours to implement.
Estimating revenue at stake takes four inputs you already have. Sessions the page could earn, conversion rate for its intent bucket, close rate, and average deal value. Rough numbers are fine, because you're ranking options rather than forecasting.
Two rules make the ranking honest.
Weight by certainty. A technical fix with a known outcome beats a content bet with the same expected value, because you'll actually realise it.
Count the full cost. Hours means brief, write, review, and ship, not just drafting time. A "two-hour" rewrite that needs legal review is not a two-hour task.
Rank the candidates by leverage
| Candidate | Typical effort | Outcome certainty | Why it ranks here |
|---|---|---|---|
| Commercial page not indexed | 1 to 2 hours | Very high | Zero traffic on a page that should convert |
| Broken redirect on a page that earned | Minutes | Very high | Signals exist and you aren't receiving them |
| Cannibalisation between commercial pages | 1 day | High | Two pages split signals, neither ranks |
| Position 4 to 10 on a commercial query | 1 to 3 days | Moderate | Moving 6 to 3 roughly doubles clicks |
| High-traffic page, weak CTA | Half a day | Moderate | No ranking dependency, ships fast |
| Cited by AI, converting badly | 1 day | Moderate | High-intent traffic landing wrong |
| Publishing something new | Days | Low, delayed | Won't rank for months |
The bottom row is where most teams instinctively go, which is the entire reason to write the list down. A new article cannot answer a one-week question, however good it is.
The first three rows share a property worth naming: the outcome is known before you start. That certainty is what makes them beat higher-ceiling work when the horizon is seven days.
Position 4 to 10 is the sweet spot, and here's why
Pages in the second half of page one have already demonstrated relevance and authority. The engine believes they belong on the page and ranks them below three better answers.
Closing that gap is usually a smaller job than it feels: matching the format the SERP now returns, adding the two subtopics your competitors cover and you don't, or improving the first hundred words so the answer is extractable.
Compare that to a page at position 25, where you're asking for a fundamentally different judgement, or a new page where you're starting from nothing. Same effort, very different odds.
Check the SERP before you commit the week
One check prevents the most common wasted week: confirm the clicks are still available.
If an AI Overview now answers the query, moving from position 6 to 3 may return far less than the click curve implies. Pew Research found that when an AI summary appears, users click a source link in only about 1% of visits.
The corollary is that becoming the cited source can be worth more than the ranking. Our own state of AI search research, covering more than 22,000 answers, found presence is path-dependent: once a brand is mentioned for a prompt, next-observation mention probability is 83.2% on ChatGPT, 83.3% on Perplexity, and 84.2% on Google AI Mode, against roughly 10% after a miss.
So on a query with an Overview, the highest-leverage move is often making the page extractable rather than pushing it up three places.
What to skip this week
A full site audit. It produces a hundred findings and no shipped work. Run one, then act on the top three, but that's a quarterly exercise.
Chasing an algorithm update. You can't act on it usefully in a week.
Anything with a payback longer than a quarter, unless nothing urgent exists.
Reporting improvements. Valuable, and it's not a leverage action.
Generate the ranked list rather than debating it
The reason this question gets answered by instinct is that assembling the candidates properly takes longer than the week allows.
Every input is a node. GSC Indexing Status finds the unindexed commercial pages. Keyword-Page Breakdown returns every page a query sends clicks to, which surfaces cannibalisation without manual checking. Page Performance Summary finds the positions 4 to 10 opportunities. fresh-win-opportunities and quick-wins surface the near-miss set directly, and Citation Watched Pages flags pages cited without converting.
The Google Search node then checks whether the clicks are still there before you commit.

A Monday-morning agent:
Start (schedule, Monday 7am) → GSC Indexing Status → GSC Page Performance Summary (positions 4 to 10, sorted by impressions) → GSC Keyword-Page Breakdown (detect cannibalisation on commercial queries) → fresh-win-opportunities recipe → Citation Watched Pages → HubSpot Get CRM Objects (revenue per landing page, to weight the candidates) → Loop / For Each over the top 15 candidates → Google Search node (confirm clicks are still available) → Code (compute revenue at stake divided by estimated hours, rank, and apply the certainty weighting) → Prompt LLM (return the top three with the specific action named for each) → Send Email to the SEO lead.
The output is three items with a named action each, arriving before the week starts. That replaces the Monday planning conversation with a decision, and the revenue weighting from CRM data is what stops the list optimising for traffic that doesn't convert.
FAQ
Related answers
- When should I rewrite, merge, redirect, or retire an underperforming page?
- Is this page underperforming because of content or because it isn't indexed?
- Did a competitor take my traffic by publishing a better answer?
Want three ranked actions in your inbox every Monday? Start a free Analyze AI trial and connect Search Console and your CRM.
