Most audits do not die because the diagnosis was wrong. They die in the week after delivery. The PDF is admired, a few easy items get fixed, and the rest wait for a mythical SEO sprint. Roadmapping is the bridge between diagnosis and backlog. Without it, you bought a story about your site, not a change programme.
This process assumes you already have a findings register with evidence. If you have findings but no order, pair this article with prioritising findings when dev time is limited. The goal here is the operating plan: who does what, in what order, and how you know it worked.
Translate findings into work objects
A finding is not a ticket. A ticket has scope, owner type, dependencies, and a definition of done. During roadmap drafting, rewrite every selected finding into a work object: engineering change, content decision, redirect map, analytics fix, or investigation spike.
- Engineering: template, robots, canonical, performance, schema rendering.
- Content: refresh, merge, kill, brief for net-new pages.
- Ops: sitemap generation, CMS fields, publishing QA checks.
- Investigation: time-boxed research when cause is unclear.
Group by root cause before you schedule heroes
Twenty duplicate-title warnings may be one facet policy bug. Fifteen thin URLs may be one archive template that should be noindexed. Roadmaps that schedule symptoms create motion without progress. Roadmaps that schedule causes free capacity.
Create cause groups with a blast-radius note: how many URLs or templates a fix touches. Stakeholders understand “fix category filter indexation affecting thousands of URLs” faster than a wall of minor severity labels. Put the cause group name on every ticket so engineers see the system, not a random URL from a crawl export.
Also separate reversible work from hard-to-undo work. A noindex on a filter class is usually reversible. A mass URL rename without a complete redirect map is not. Roadmaps that ignore reversibility push risky tickets into the same lane as safe ones and create avoidable launch fear.
Build a Now, Next, Later board with real capacity
Write capacity in plain numbers before you fill Now. How many engineering days exist this quarter for SEO? How many editorial days? Which freezes or launches are already locked? Then fill Now until capacity is honestly full. Everything else is Next or Later, even if a crawler painted it red.
| Lane | What belongs here | Owner pattern | Exit criteria |
|---|---|---|---|
| Now | High impact, unblocked, fits capacity | Named person + sprint ID | Shipped + verified in GSC/analytics |
| Next | High impact, waiting on dependency or capacity | Named team, target month | Dependency cleared or capacity freed |
| Later | Lower impact, polish, or post-migration items | Backlog owner only | Re-score trigger hit |
| Parked | Needs product decision or redesign | Stakeholder decision date | Decision recorded |
Sequence dependencies so tickets do not fight
Index access before sitemap expansion. Redirect maps before URL redesigns. Template fixes before bulk on-page copy on dying templates. Content merges before rewiring internal links to URLs you are about to remove. Publish the sequence so non-SEO managers stop asking why meta rewrites are not ticket one.
- Crawl and index eligibility blockers.
- URL and redirect policy stability.
- Money template repairs.
- Content keep/refresh/kill waves.
- Internal linking and supporting IA.
- Polish: bulk meta, non-critical schema, minor nits.
One-page roadmap entry format
### NOW-03 Fix filter URL indexation
Cause group: Facet index policy
Impact: High (category templates)
Effort: M (1-2 eng days)
Dependencies: None
Owner: Web platform
Ship window: Sprint 14
Done when:
- Filter param URLs emit noindex,follow
- Canonical points to clean category
- Sitemap excludes filter URLs
- 5 URL Inspection samples confirm
Risk if slipped: Continued index bloat on /shop/Make verification part of the roadmap, not a hope
Every Now item needs a check you can run after release. URL Inspection samples, robots header checks, analytics annotations, impression trends for a template, or a crawl diff. Without verification, “done” means merged code, not improved eligibility or clarity.
Schedule a short readout two to four weeks after each Now wave. Keep it boring: what shipped, what verified, what moved to Next, what new regressions appeared. Boring readouts build trust. Slide theatres do not. For scoring detail behind the order, see impact versus effort scoring.
Handle politics, freezes, and split ownership
When engineering and content must move in parallel
Do not let content wait behind engineering if the work is independent. Run a content lane for keep/refresh/kill decisions while platform fixes indexation. The roadmap should show parallel swimlanes, not a single bottleneck narrative that excuses inaction.
When a redesign will discard work
If a rebuild is coming, put continuity items in Now and park cosmetic front-end SEO on the dying templates. Migrations need their own gate; see site migration audit expectations so the roadmap does not fight the launch calendar.
When leadership wants everything Now
Answer with capacity math and cause grouping. “We can ship three cause-level fixes that remove fourteen symptoms, or fourteen symptom tickets that leave the cause.” Show the defer list publicly so Later does not feel like denial. If you want the operating philosophy behind the deliverable, how we work describes the same diagnosis-to-roadmap stance.
Document decision rights once. Who can approve kills and redirects? Who can change robots rules in production? Who signs off that a Now item is verified? Ambiguous authority is why roadmaps stall after the kickoff meeting. Write the names next to the lanes.
A roadmap is a capacity contract. If everything is Now, nothing is managed.