Discussion starter: a content SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Content SEO
On a mature blog with 2,000+ posts, what split of effort between updating old posts and publishing new ones has worked best?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
No ratio has evidence behind it as a general rule, and “updating versus publishing” leaves out the third job a 2,000-post archive needs: consolidating or retiring posts that overlap or no longer earn their keep. The split that holds up is derived, not picked. First reserve a maintenance share sized by how quickly your topics go out of date, because an archive this size is a liability as well as an asset. Then send the remaining hours wherever your own measured return per hour is higher: recovering posts that lost ground, merging overlaps, or covering gaps with real demand.
As a planning assumption rather than a finding, I would budget more than half of the first quarter’s hours for existing posts on a blog that has never been systematically triaged, because the first pass through an untouched archive is where cheap wins are likeliest. A blog with a stable, evergreen archive and plenty of uncovered demand could reasonably invert that. Four numbers from your own data show which of those you are, and the split should move each quarter as measured returns come in.
Why updates and new posts can’t be judged at the same checkpoint
The two kinds of work pay back on different curves, so comparing them on one date biases the answer. Think of renovating a rented flat versus building a new one. The renovation earns sooner, because the tenant, the address and the rent history already exist, but its ceiling is the flat you already had. The new build earns nothing for a while, then adds capacity the old stock never could.
In search terms, an updated post already has an indexed URL, links and a ranking history. Once Google recrawls and reassesses it, any effect shows up comparatively quickly, and it is bounded by the queries that URL can plausibly rank for. A new post has to be discovered, indexed and earn links and a history before its return arrives, and for some posts it never does. Judge both at six weeks and the comparison is tilted towards updates. Judge a new post at twelve months against an update measured at six weeks and the comparison tilts the other way. Compare them at matched ages instead.
The four numbers that set the split
Number Where to get it Points towards existing posts when Points towards new posts when Share of organic clicks and conversions landing on posts more than a year old Search Console page export joined with publish dates; GA4 landing-page key events The archive carries most of the value Recent posts already carry much of it Recoverable decay: posts that lost clicks for reasons an edit can fix Search Console year on year by page, reading position and impressions as well as clicks The pool is large The pool is small, or the losses came from the results page changing Topic volatility: how quickly the facts in your posts go out of date Sample 50 posts and count how many contain something now wrong Many samples are out of date Few are Uncovered demand: questions with search demand and no dedicated post Queries with impressions but no matching page, keyword research, sales and support questions Little remains A lot remains The second number needs care. A post whose clicks fell while its position and impressions held has not lost rankings. The likelier cause is the results page around it changing, whether through AI Overviews, other features, a layout shift or a rewritten title, and rewriting the post may not bring those clicks back. As of September 2026, Search Console’s Search generative AI performance report is available to all eligible sites and shows AI-feature impressions by page (impressions only, no clicks or queries), which helps flag posts whose queries now trigger AI features. Count those as a SERP problem, not as recoverable decay.
Size the maintenance floor before arguing about ratios
The part of the split that is not optional is keeping posts that earn traffic, links or conversions accurate. Size it from the archive rather than from a ratio: how many posts matter, how often each needs checking given its topic, and how many checks turn into real updates.
Hypothetical example:
A 2,000-post blog has 160 editorial hours a month, and 500 posts earn meaningful clicks, links or conversions. Of those, 150 cover volatile topics (software, pricing, regulation, statistics) and need a check every six months: 25 checks a month. The other 350 need a check every two years: about 15 a month. Forty checks at 45 minutes each is 30 hours. If one check in four becomes a four-hour substantive update, that adds 40 hours. The maintenance floor is 70 hours, about 44% of capacity, before any recovery, consolidation or new post.
Change one assumption. If only 50 of the 500 are volatile, checks fall to about 27 a month and the floor to about 47 hours, roughly 29% of capacity. Same archive size, very different split. Volatility, not post count, drives the floor.
Two consequences follow. If the floor alone would consume most of your capacity, the answer is not a better ratio but a smaller maintenance surface, which is what consolidation and retirement buy. And the 1,500 posts outside the review schedule are not a to-do list. They are candidates for merging, retiring or leaving alone.
Sort the archive into jobs, not a queue
Bucket Signal Job Effort Earning and accurate Stable clicks or conversions; facts current Scheduled check only Low Earning but dated Facts, prices, screenshots or recommendations now wrong Substantive update; change the visible date only when the content materially changed Medium Lost ground to better pages Positions falling on the post’s main queries; competing pages now cover more Rewrite against what ranks now, adding what you know that they do not Medium to high Lost clicks, not positions Position and impressions steady, CTR down Do not rewrite for rankings; check AI-feature impressions; update only if dated Low Overlapping posts Several posts share queries; the ranking URL keeps switching Merge into the strongest URL and 301 the rest Medium No value No clicks, links or conversions for a year or more, and no user purpose Retire or leave; do not expect a sitewide lift from removal Low Gap Demand with no dedicated post, or only an off-intent one New post High Google’s documentation constrains both sides of the split. Its ranking systems guide describes “various ‘query deserves freshness’ systems designed to show fresher content for queries where it would be expected”. That implies maintenance earns more on topics where readers expect current information than on evergreen explainers, where an update helps only if it makes the page better. Its helpful content guidance lists, among its warning questions, “Are you changing the date of pages to make them seem fresh when the content has not substantially changed?” and “Are you adding a lot of new content or removing a lot of older content primarily because you believe it will help your search rankings overall by somehow making your site seem ‘fresh?’ (No, it won’t)”. For new posts, the same page asks: “Are you producing lots of content on many different topics in hopes that some of it might perform well in search results?” Volume is not a strategy on either side of the split.
Measure return per hour without fooling yourself
- Log every work item. URL, job type, hours and ship date, in one sheet. Mark batches with Search Console’s custom annotations so the dates sit on the charts.
- Randomise the queue. You cannot update 300 eligible posts at once anyway, so randomise the order. Posts still waiting act as the comparison group for posts already done, which absorbs core updates, seasonality and SERP changes that hit both groups.
- Read each job type on its own clock. Read an update batch once most of its posts show a recrawl after the edit (URL Inspection shows the last crawl date), then again a couple of months later to see whether the change held. Read new posts quarterly through their first year, because their return arrives later. Compare job types at matched ages, not on the same calendar date.
- Count what the blog is for. Use clicks and conversions (GA4 landing-page key events) per editorial hour, not traffic alone. A post that feeds sign-ups or product pages deserves maintenance priority over a bigger post that feeds nothing.
- Measure topics as well as URLs. A new post can take clicks from an older one on the same subject. Group the posts and queries for each topic and check whether the topic grew or just reshuffled between URLs.
Signals to rebalance the split
- Updated posts clearly beat the waiting queue, and new posts lag at matched ages: move hours towards maintenance, up to the size of the recoverable pool.
- Update gains are small or fade: the fixable decay is used up, or the losses were SERP-driven. Move hours to consolidation or new coverage.
- New posts cannibalise older ones (topic totals flat while URLs swap): stop publishing on that topic and consolidate.
- The maintenance floor consumes most of the capacity: shrink the surface through consolidation and retirement before adding writers.
The position I would take
If this were my blog, I would not pick a ratio. I would size the maintenance floor from the archive, sort the rest into jobs, and weight the first quarter towards existing posts because that is where cheap wins are likeliest. From the second quarter, the split follows the measured return per hour of each job type at matched ages, and the archive gets smaller wherever maintaining a post costs more than it earns.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.