An SEO roadmap that starts with Search Console, not a tool estimate
An SEO roadmap is only as good as the numbers it plans against. This one starts from your own Google Analytics and Search Console exports, checks every issue you already suspect, finds the problem behind them, and ends in a document, a 10-row summary and, if you want it, a page your team can open on a call.
It is built in Claude Code with a workspace of 24 Python helpers, and it has been used across 15 client workspaces so far, producing 13 roadmaps and 9 published client pages.
Why most roadmaps plan against the wrong numbers
Third-party tools estimate traffic. Search Console records it. On one site a tool estimated about 7,747 sessions a month, while Search Console showed 5,234 clicks in 28 days on more than 2 million impressions, and 801 pages with impressions but no clicks at all.
A plan built on the estimate would have been confident and wrong. So the first request is always the two exports, and every figure in the roadmap is labelled with where it came from.
Five analyses before a single recommendation
The analytics pass runs before any crawl or keyword work, and five of its outputs have earned their place on every roadmap since:
Analysis | What it found on one client |
|---|---|
CTR upside | What today's impressions would earn at a realistic click-through rate |
Clicks by section | A glossary section with 94,850 impressions and 12 clicks, which justified retiring 23 pages |
CTR-leak list | 65 pages holding 858,072 impressions, ranking well and under-clicked |
Clicks against conversions | The blog earned 51% of clicks and 2% of conversions; gated guides earned 13% and 86% |
Twin pages | One page with 121,634 impressions and a near-copy of it competing for the same searches |
A page makes the CTR-leak list when it has at least 3,000 impressions, an average position of 12 or better, and a click-through rate below the site's own average. Those are the cheapest wins on the site, because the ranking already exists.
Checking your own list before adding to it
Clients usually arrive with a list of problems. Each one is tested with requests against the live site before it goes in the roadmap, and the result is written up as check, result and evidence.
On one site, 21 reported issues were checked and 4 turned out to be wrong in the detail. The more useful finding was underneath: 7 of the 21 were the same bug. The site's SEO layer never ran, so all 10,843 URLs in its sitemap returned an identical empty shell. Fixing that one thing closed a third of the list, which is why every roadmap looks for the root cause before it lists fixes.
What the roadmap document contains
The roadmap is written to you, in plain English, and runs up to ten pages. Every problem it names comes with a fix you can copy, like the actual title and meta text or the schema block. The same outline works as an SEO roadmap template for any site:
- A 60-second summary: where you stand, the opportunity, what we'll do and what to expect
- One section per question you asked, plus one or two clearly labelled extra recommendations
- Technical SEO and schema, the content plan, internal links, links from other sites, and conversion
- Measurement, a 12-month projection, and a 90-day plan week by week
- Next steps split into what we need from you and what we'll do
Unknowns stay visible as placeholders and questions rather than being filled with guesses. And the document always says what is already working, because a roadmap that only lists problems gets argued with instead of used.
Two plans: the pages you have and the pages you don't
A site that already earns impressions gets two content plans. Plan A fixes existing pages in the first month: new titles and meta descriptions for the top leaks, merges and redirects ranked by impressions at risk, and a bridge from the articles that get clicks to the guides that convert.
Plan B is new articles, in the format the searches already reward. On one site that meant listicles, because the client appeared in none of the 48 listicle results checked while a competitor appeared in 11. New articles come from keyword research automation, and what was filtered out of the keyword list is always stated.
The summary table and the projection
Alongside the document sits a 10-row table: action, description, timeline, priority, outcomes and deliverables, one sentence each. It fits on one screen, which is the point.
The 12-month projection follows strict rules, because it is the part most likely to be quoted back later. It is built from the scope of work, anchored to Search Console clicks, shown as ranges, and no cell may sit below today's figure. When the first draft is optimistic, outcomes are halved and deliverables are not, since deliverables are what you actually control.
Every number re-checked before it ships
In August 2026 a client's question exposed 11 wrong numbers in a roadmap that had already been delivered, plus three places where the document and its page disagreed. Since then, every roadmap gets a verification step before it goes out.
A per-client script re-derives each headline figure from the raw exports and prints what the document claims beside what the data says. Another compares the document, the page and the tables against each other for figures that exist in one and not the others. They catch the mistakes a careful read misses, like a tracking variant of a URL counted as a separate page, or "joint best" position that was really fourth.
The page version
When a roadmap is published as a page, it is one self-contained file styled to your own brand: your colors, fonts and headline language. Charts are limited to three series and checked for color-blind separation, since a fourth color failed that check on one client.
Before calling it done, the page is checked live, screenshotted and searched for anything that shouldn't be there. One check caught a serif font rendering "107" as "IO7" in a headline figure.
When an SEO roadmap is worth commissioning
Commission one when you have at least a few months of Search Console data and a sense that traffic isn't turning into much. It is also worth it before hiring or replacing an SEO provider, so the brief is based on your data rather than a pitch.
It is less useful for a brand-new site with no search data yet, where there's nothing to reconcile and a technical checklist does more. Larger sites with many sections get the most from it, for the same reasons an enterprise SEO audit does.
Buying the roadmap
The closest thing to buy is a Discovery sprint: $2,500 over two weeks, ending in a prioritized plan that's yours to keep either way. Where search is the channel that matters most, this roadmap is the core of that plan.
Start with two exports
Send your last 28 days of page-level data from Google Analytics and Search Console, and the list of problems you already suspect. On a call we'll show you the first analysis on your own numbers.
Related automations
SEO workflow management
Let the content plan's status column run the work: each automation picks up rows at its stage, writes the next status, and stops where a person approves.
Read the build →
Topical content map
Group a keyword universe into clusters, map every page you already have onto them, and phase each gap by evidence before anything gets written.
Read the build →
AI Overview tracking report
A weekly report joining Search Console and GA4 with visibility across ChatGPT, Gemini, Perplexity, AI Overviews and AI Mode, checked before it's written.
Read the build →