- Rankings decay quietly. Maintenance is what stops a good year turning into a slow decline.
- Refreshing existing pages usually outperforms publishing new ones in year two.
- Watch for regressions after every site change - they are the most common cause of sudden drops.
- A maintenance programme is smaller and cheaper than the build. It should not cost the same.
The first year builds the asset. The second decides whether it keeps producing. Most of the visibility a site loses in year two is not lost to a competitor or an algorithm: it is lost to neglect, one aged page and one unnoticed regression at a time. This is what a sensible maintenance programme contains once the heavy lifting is done, what it should cost, and how to tell when it has stopped working.
The quarterly refresh cycle
List the pages producing enquiries and the pages ranking between three and ten. Every quarter, take the top few and update them: current prices, new examples, questions customers started asking, anything that aged. Update the visible date only when the content genuinely changed - a date bumped on an unchanged page fools nobody and erodes trust with readers who return.
This is usually the single highest-return activity available in year two, and it is the one most often dropped in favour of publishing something new. A page already ranking eighth has proved it can rank; moving it to third is a smaller job than starting a page from nothing.
Not every page deserves the same attention. Sorting them by how fast their facts age gives you the cadence:
| Page type | Refresh cadence | What actually changes | Overdue when |
|---|---|---|---|
| Prices and packages | Whenever the facts change | Figures, what is included, minimums | The page contradicts your quote |
| Core service pages | Twice a year | Examples, proof, objections, internal links | Ranking slipped out of the top five |
| Local and district pages | Twice a year | Coverage, opening hours, local proof | Business Profile no longer matches |
| Guides and explainers | Once a year | Screenshots, tool names, regulation | It names a tool that no longer exists |
| Case studies | When a better one exists | Replace, do not accumulate | The most recent one is two years old |
| Seasonal pages | Before the peak, not during | Dates, stock, offers | It is live and still says last year |
Monthly hygiene
The recurring checks are dull, quick, and the reason most sites never develop the problems that take a quarter to unwind. Half a day a month covers all of it for a small site.
| Check | Time | What you are looking for |
|---|---|---|
| Business Profile | 30 min | Hours, photographs, posts, and a reply to every review |
| Search Console coverage | 20 min | New not-indexed URLs, especially a whole template at once |
| Query movement | 30 min | Terms that lost clicks month on month, before they lose rankings |
| Broken links and 404s | 20 min | Internal links pointing at pages that moved or were deleted |
| Template speed | 20 min | Scripts and plugins accumulating on the main templates |
| Backlink additions | 15 min | New links worth thanking for, and anything that needs watching |
None of these produce a headline for a monthly report, which is precisely why they get dropped from retainers that are judged on how impressive the report looks.
A twelve-month maintenance calendar
Spreading the heavier tasks across the year keeps the workload flat and stops everything landing in December. This is the shape most sites settle into:
| Quarter | The heavier task | Alongside the monthly hygiene |
|---|---|---|
| Q1 | Full technical crawl and a redirect audit | Refresh the two pages closest to page one |
| Q2 | Content inventory: what to keep, merge, or retire | Update prices and any regulated claims |
| Q3 | Competitor comparison on your ten commercial queries | Prepare seasonal pages before the peak |
| Q4 | Internal linking review and analytics sanity check | Refresh case studies and proof |
After every site change
A new plugin, a theme update, a developer deployment or a marketing tag can break indexing, slow a template or remove a canonical. Check the affected pages immediately rather than waiting for the monthly report. Most catastrophic drops are regressions that went unnoticed for weeks, and almost all of them were reversible on the day they shipped.
Three account for the majority of them:
The single most expensive two words in SEO. It removes pages from the index within days and takes weeks to recover. Check the robots meta tag on a live URL after every deployment.
New templates, new slugs, no redirects. The rankings belonged to the URLs, not to the design. Map old to new before launch, not after the traffic falls.
Nothing is wrong with the site at all, but every report says otherwise for a month. Compare Search Console clicks against analytics sessions before investigating anything else.
“Almost every catastrophic traffic drop we are called in to diagnose turns out to be something the site did to itself, two weeks earlier, on a Thursday afternoon.”
What maintenance should cost
Maintenance is genuinely less work than the build, and the fee should reflect that. A retainer that stays identical the month after launch deserves a question. Roughly what the work amounts to:
| Site | Hours a month | What that buys |
|---|---|---|
| Local business, under 30 pages | 3-5 | Hygiene, one refreshed page a quarter, Business Profile |
| Services company, 30-100 pages | 8-15 | The above plus a quarterly technical pass and content inventory |
| Multi-location or ecommerce | 20-40 | The above plus per-location or per-category work and release checks |
What matters more than the number is what arrives. Ask for the change log: which pages were edited, what shipped, what was fixed. If it is empty two months running, the retainer is buying reports rather than work.
What still needs new work
New services, new districts, new questions the market started asking, and seasonal pages that need to be live before the peak. Beyond that, the emphasis shifts from volume to depth: fewer pages, better maintained. A site of forty pages that are all current will out-perform one of two hundred where half contradict each other.
This is also the point at which consolidation pays. Pages published in year one that target the same query compete with each other; merging them into one stronger page and redirecting the rest usually recovers more ground than anything new you could publish that quarter.
Signals your maintenance is not working
Clicks can move for many reasons, but impressions for your target cluster are the leading signal. Flat for six months means nothing is being built.
If nobody can list the pages edited last month with URLs and dates, no work was done regardless of what the dashboard shows.
A report that repeats "improve internal linking" quarter after quarter is describing work, not doing it.
A healthy site loses pages as well as gaining them. A page count that only ever rises means duplication is accumulating.
Frequently asked questions
Can we reduce the retainer after year one?
Usually yes, and honest providers will suggest it. Maintenance is genuinely less work than the build; a fee that never changes deserves a question.
How often should pages be refreshed?
Important pages twice a year, price and regulation pages whenever the facts change, everything else when it stops performing.
What happens if we stop entirely?
Decline is gradual rather than immediate: competitors update, facts age, technical debt accumulates. Most sites hold for months and then slide.
Should we keep publishing new articles every month?
Only where there is genuinely something new to say. A fixed publishing quota is how sites end up with pages competing against each other. In year two, one properly refreshed page usually beats two new thin ones.
Can we do the maintenance in-house?
The monthly hygiene, yes - it is a checklist and it takes an afternoon. The quarterly technical pass and the decisions about what to merge or retire are where outside judgement earns its fee.
How quickly does a refreshed page move?
A page already in the top ten typically responds within two to six weeks. A page on the third page of results is not a refresh problem; it needs rebuilding or merging into something stronger.
Move from build to maintenance
We define what still needs doing monthly, and what you can stop paying for.
Talk to Madrid SEORelated: SEO services.