The SEO Clock
Most SEO timelines are published as if every activity could start on Monday. They can't. Half the work on this board is gated — it produces nothing until something else is true first. So this isn't a list of durations. It's the order, and then the duration.
Set your situation and the board re-sequences. Treat every number as a basic estimate — competitive landscape, technical prowess, existing authority and internal resourcing decide the real one.
The earliest point the change is visible above noise in Search Console or analytics — not the point it pays for itself.
The point the curve flattens — roughly 90% of the organic gain this activity delivers on its own, with nothing else changing.
Start now
week 1Nothing has to be true first. If you only fund one column, fund this one.Removing an accidental noindex or robots block▼Technical3d → 3w3w
You can deploy a template change, and you have Search Console access to confirm the pages are being re-requested.
The tag comes off but nothing else changes — no sitemap resubmission, no internal links to the recovered pages — so Google takes weeks to notice instead of days.
The fastest thing in SEO, and the hardest gate on this board: while it is in place, every content activity below produces exactly nothing. If pages previously ranked, most traffic returns within a crawl cycle. If they never did, you are not recovering — you are starting, and the rest of the sequence applies.
Fixing analytics and conversion tracking▼Revenue & Measurement4d → 3w3w
Access to the tag manager or codebase, and agreement on what counts as a conversion.
Fixing the tracking and then comparing the new numbers to the old ones as if the trend were real. Annotate the change date or you will spend a quarter explaining a fake drop.
Zero ranking impact and it still goes first, because it gates everything you would want to prove later. Note that authority and competition do not move this bar at all — no amount of domain strength makes a tag container fire faster.
XML sitemap and indexing hygiene▼Technical1w → 4w4w
A reliable list of the URLs you actually want indexed, and a way to keep the file generated automatically.
Treating the sitemap as a ranking lever. It is a discovery aid. A clean sitemap full of thin pages gets them found faster, not ranked.
Discovery speeds up almost immediately for new URLs. What it will not do is rescue pages Google has already decided to ignore — those need a quality or linking fix, not a sitemap entry.
Google Business Profile optimisation▼Local1w → 7w7w
A verified profile and a genuine physical presence or service area.
Filling the profile out once and abandoning it. Freshness signals — posts, photos, answered questions, new reviews — carry more weight here than anywhere else in SEO.
The fastest local wins on the board, and almost entirely independent of your website — which is why it is not gated by any of the technical work. Category selection alone can move map pack visibility within a week or two.
llms.txt and opening access to AI crawlers▼AI Search2w → 2mo2mo
A decision on whether you want your content used for training and retrieval at all. That is a business call, not an SEO one.
Publishing llms.txt and treating it as an AI search strategy. Adoption is partial and unproven; blocking the crawlers that do cite you is the more common and more expensive mistake.
Low effort, low certainty. I recommend it as cheap insurance rather than a growth lever — and I would rather you spent the quarter on the citation work at the bottom of this board.
Fixing broken internal links and 404s▼Technical2w → 2mo2mo
A full crawl and, ideally, server logs to see which broken URLs are actually being requested.
Blanket-redirecting every 404 to the homepage. Google treats irrelevant redirects as soft 404s and you lose the signal anyway.
Mostly a hygiene and user-experience fix. The exception is a 404 that used to hold external links — redirect that one properly and the recovered authority can be immediately measurable.
Site migration or replatform, done properly▼Technical2w → 6mo6mo
A complete URL inventory, a mapped redirect plan signed off before launch, and a staging environment that is genuinely blocked from indexing.
Treating redirects as a launch-week task. Also: changing URLs, templates, content and platform in one release, so when traffic drops nobody can tell which change caused it.
Not a growth activity — a risk-management one, and the only item here that pauses the rest of the board. Success looks like flat. Expect a dip of a few weeks even on a clean migration, and do not stack other changes on top of it for a full quarter.
Reclaiming unlinked mentions and broken backlinks▼Links & Authority2w → 2mo2mo
A brand people already write about, and a list of 404ing URLs that used to hold links.
Sending the same template to 200 editors. Reclamation works because the ask is small and obviously fair — mass-mailing it destroys that.
The most overlooked link tactic. You are not asking for a favour, you are pointing out an incomplete citation. Conversion rates run far higher than cold outreach — but the whole tactic scales with how often you are already mentioned, which is why it moves so much with authority.
Canonical and duplicate content cleanup▼Technical3w → 3mo3mo
A crawl that shows which URL variants exist and which one you want to win.
Canonicalising to a page that is itself canonicalised elsewhere, or contradicting your canonicals with your sitemap and internal links. Google resolves the conflict its own way.
Rarely produces a headline win on its own. What it does is stop you wasting authority across near-identical URLs, which is why it sits before the content work rather than alongside it.
Fixing JavaScript rendering issues▼Technical3w → 3mo3mo
Engineering time, and a rendering strategy decision — server-side, static, or hydration — rather than a patch.
Testing with the URL Inspection tool alone and assuming what renders there is what gets indexed at scale. Rendering is queued and budgeted; a page that renders in the tester may still be indexed empty.
When content is genuinely invisible to crawlers this is a gate, not an improvement — publishing into it wastes the whole content budget. When rendering is merely slow, the gain is smaller than the engineering cost suggests. Diagnose which one you have before committing a quarter.
Digital PR campaign▼Links & Authority3w → 5mo5mo
A genuine story, data, or angle a journalist would run without being paid. Everything else is advertising.
Measuring coverage instead of links. Half of press mentions carry no link at all, and a campaign that trends on social but earns nothing followed has cost you a quarter.
The highest-variance activity here, and it needs no site prerequisites — a story is a story. A campaign that lands moves authority faster than anything else on the board; one that does not returns nothing measurable. Budget for two or three attempts, never one.
Site speed and Core Web Vitals work▼Technical4w → 4mo4mo
Real-user field data, not lab scores. And an engineering team able to touch the render path, not just add a caching plugin.
Optimising for a Lighthouse number. The ranking signal uses field data, the revenue impact comes from conversion rate, and neither correlates neatly with the score you screenshot for the deck.
The honest position: as a ranking factor this is a tiebreaker, and it is the activity most often oversold. As a conversion and crawl-efficiency factor it can be genuinely valuable. Fund it on the second argument, not the first — and note it barely moves with your authority, only with how fast your competitors already are.
Guest posts and niche placements▼Links & Authority6w → 6mo6mo
A shortlist of sites with genuine traffic and editorial standards, and something worth writing.
Buying volume from link vendors. Cheap placements cluster on the same networks, and the pattern is visible from the outside.
Slow and linear — you get roughly what you pay in effort, with no compounding. I treat it as a floor-raiser while PR and content do the real work, not as a strategy in itself.
Becoming the source others cite▼AI Search2mo → 10mo10mo
Access to data nobody else has, or a willingness to run the study nobody else will. Plus the patience to promote it for a year.
Publishing the research as a gated PDF. If it cannot be crawled, quoted and linked, you have built a lead magnet, not a citation asset.
The longest play on this board, and one of the few that a brand-new site can legitimately start on day one — original data does not require existing authority, it creates it. Each citation feeds the next, across journalists, competitors and language models alike.
Brand building and demand generation▼Links & Authority4mo → 1.2y1.2y
A product worth talking about and a marketing function that does more than SEO.
Not measuring it, then cutting it. Branded search volume is the cleanest proxy you have — track it monthly from the start or you will never be able to defend the spend.
Barely feels like SEO and outperforms most of it over three years. Start it now precisely because it is slow: rising branded search lifts click-through on every non-branded query you rank for, and it is the one advantage a competitor cannot buy past in a quarter.
Once you can measure
~week 3Pointless before the tracking is trustworthy — you would be optimising against numbers you cannot defend.CRO on traffic you already have▼Revenue & Measurementstarts ~3w · 5w → 3mo3mo
Working conversion tracking, and enough traffic on the target template to reach significance — roughly 300+ conversions a month, or you are reading noise.
Testing button colours on a page whose offer is wrong. Also: calling a winner at 80% confidence because the deadline arrived.
The only activity here that raises revenue without raising rankings, and completely insensitive to your authority or competition — it is your traffic and your funnel. On a site with real volume it is usually the fastest path to money, which is why it should run in parallel with the slow work rather than after it.
Content pruning and consolidation▼Contentstarts ~3w · 7w → 5mo5mo
At least a year of trustworthy performance data per URL, so you are cutting genuine dead weight rather than seasonal pages or pages your broken tracking under-reported.
Deleting pages that get no clicks but carry links or internal-linking value. Check the backlink profile of every URL before it goes.
Results are indirect and slower than the effort feels like it deserves — you are improving the average, not adding to the total. Expect the reporting dip before the recovery, and warn stakeholders in advance.
Once the site can be crawled and indexed
~week 4Every activity here produces literally nothing while an indexation block is in place.On-page optimisation of existing ranking pages▼Contentstarts ~4w · 5w → 3mo3mo
Pages that are indexable and already ranking somewhere in positions 5-30, so there is existing relevance to sharpen.
Rewriting titles for keywords instead of for the intent behind them. Position goes up, click-through goes down, and net traffic falls.
The best effort-to-outcome ratio in SEO and the most under-done. Pages on page two already have Google’s trust; you are adjusting the match, not earning it. This is where I would start the moment the technical gate clears.
Structured data and schema markup▼Technicalstarts ~4w · 5w → 2mo2mo
Indexable content that genuinely matches a supported rich result type. Marking up something you do not have is a penalty risk, not a shortcut.
Adding schema for its own sake and expecting rankings. It affects how you appear, not usually where. Invalid markup can also remove eligibility entirely.
Rich result eligibility can appear within days of a recrawl, but Google decides whether to show it and can withdraw the feature for your whole category overnight. Treat any CTR uplift as borrowed, not earned — which is why the bar fades rather than holding.
Resolving keyword cannibalisation▼Contentstarts ~4w · 6w → 3mo3mo
Query-level data showing two or more of your indexable URLs trading places for the same intent.
Deleting the weaker page instead of consolidating it. You throw away links and content that should have been merged into the winner.
Consolidation nearly always beats deletion. Pick the page with the strongest link profile, fold the rest into it with 301s, and let the signals stack instead of splitting.
Refreshing decaying content▼Contentstarts ~4w · 6w → 3mo3mo
Indexable pages with a clear downward trend in impressions or position over two or more quarters.
Changing the date and calling it a refresh. Also: refreshing pages that are declining because the query intent moved on — those need replacing, not updating.
Reliable and repeatable, but understand what you are buying: a refresh resets the clock, it does not stop it. Build the cadence into the calendar or you will be doing this again next year on the same URLs.
Internal linking restructure▼Technicalstarts ~4w · 7w → 4mo4mo
Pages that can actually be indexed, a clear view of which ones matter commercially, and a CMS that lets you place links contextually rather than only in navigation.
Adding links everywhere. A hundred sitewide footer links to a money page dilute the signal; six contextual links from topically related pages move it.
The highest-leverage technical work on most established sites — you are redistributing authority you already own. It sits after the indexation gate for a blunt reason: links pointing at pages Google will not index do nothing.
New content on an established site▼Contentstarts ~4w · 8w → 5mo5mo
A crawlable, indexable site with existing topical authority in the area, and internal links from relevant pages on day one.
Publishing and moving on. A new page with no internal links and no promotion is an orphan with a sitemap entry.
On a site Google already trusts, new pages get crawled in days and often enter the index with a provisional position that then settles. The first two weeks are a bad read on eventual performance in either direction.
Building a topic cluster from scratch▼Contentstarts ~4w · 3mo → 9mo9mo
An indexable site, a topic you can credibly own, and the resourcing to publish the whole cluster rather than the pillar plus two posts.
Abandoning it at 60% because the early numbers look flat. Clusters are non-linear — most of the value arrives once the internal linking between the pieces is complete.
The slowest activity most clients should still do. Half-built clusters are the most common wasted investment in SEO; finished ones are the thing still paying two years later.
New content on a brand-new domain▼Contentstarts ~4w · 4mo → 10mo10mo
Patience, and a budget that survives two quarters of nearly flat charts without anyone panicking.
Judging the strategy at month three. Also: publishing a hundred thin pages to look busy, which sets the site’s quality baseline before it has any authority to spend.
The number the industry lies about most, and the row most sensitive to the two controls above — move authority to “brand new” and “brutal” and watch what actually happens to the estimate. Ten genuinely excellent pages will outrun a hundred mediocre ones, and neither looks impressive before month four.
Once on-site foundations are sound
~month 3Scale multiplies whatever your templates already do. Fix the template first, then multiply.Location landing pages▼Localstarts ~3mo · 4mo → 9mo9mo
Sound on-site foundations, a real reason to serve each location, and content that differs beyond a swapped city name.
Templated pages with the town name find-and-replaced. These are the textbook thin-content pattern and they now routinely fail to index at all.
Works when each page reflects an actual local presence — staff, cases, reviews, service specifics. Launching these onto a site with duplicate-content or crawl problems is how you end up with two hundred unindexed URLs.
Programmatic pages at scale▼Contentstarts ~3mo · 5mo → 10mo10mo
A genuinely useful dataset, unique value per page, clean templates, and crawl budget to spare. Missing any one of the four and this fails.
Shipping 40,000 URLs and discovering three months later that 6% are indexed. Launch a few hundred, verify indexation, then scale.
Either the best ROI on this board or a quarter spent manufacturing a crawl-budget problem. It is gated on foundations for a reason: scale multiplies whatever your template already does, including its mistakes.
Once you have authority to spend
~month 8These need a reputation that already exists. Starting here first is the most expensive mistake on the board.Getting cited in AI answers▼AI Searchstarts ~8mo · 10mo → 1.3y1.3y
Content that is already retrievable and quotable — clear claims, clean structure — and enough earned authority that a model has reason to trust you over the incumbent.
Optimising for one assistant’s output this month. Retrieval sets shift weekly; chasing a single answer box is chasing weather, not climate.
The most requested and least predictable work I do, and the most heavily gated — models cite sources that are already cited elsewhere, so this is downstream of authority you have to earn first. What helps is being the clearest source on a specific claim. What does not help is prompt-level tinkering.
Drawn from industry consensus and twelve years of my own client work, stated as such — not from a published study. Where the two disagreed I used what I've actually seen. If your data contradicts a row here, I'd genuinely like to hear it.
The hard question isn't how long. It's what you cut.
Nobody gets to run 28 activities. You have a headcount, a runway and a board asking about next quarter — and the right answer is usually four of these, in a specific order, with the rest explicitly abandoned. Tell me your constraints and I'll tell you what I'd drop.