A page that brought three hundred clicks a month two years ago brings sixty today. Nobody deleted it, nobody broke it, there was no drop you could point at in a chart. It simply got old, slowly and quietly, while three new articles were written above it that have the same fate ahead of them.
That is the most expensive reflex in content marketing: writing new things while the existing stock decays. This piece sorts the three ways content ages (they need completely different treatments), shows how to pull the candidates out of Search Console rather than out of a hunch, and explains why changing the date does not just fail to help but destroys your ability to measure.
Why a rewrite is cheaper than a new page
A new page starts at zero: no history, no internal links, no assessment, no position. Weeks pass before Google sorts it anywhere serious, and most new pages land where pages land when they contribute nothing of their own.
An existing page already has all of it. It sits at position 9, it has impressions, it is linked internally, it has been crawled and assessed. The distance from position 9 to position 4 is shorter than the distance from nothing to position 9, and it is far cheaper. The same arithmetic in numbers is in increasing website traffic: the lever on pages that already rank is the cheapest one there is.
Three kinds of decay, three different jobs
This is the part the usual content refresh guides mash together, which is why they always arrive at the same action. There are really three causes, and only one of them is treated with new paragraphs.
Factual decay. Prices, deadlines, grant amounts, tax rates, version numbers, screenshots of interfaces that no longer look like that. The page is still built correctly, it just says things that are partly untrue. This is the easiest kind: swap it out and you are done. It is also the most urgent, because one wrong figure costs you the trust of everyone who checks it.
Competitive decay. Your page is exactly as good as it was two years ago, and three others got better. Swapping things out does not help here. You have to look at the results page and answer what the pages above you deliver that you do not. Usually the answer is uncomfortably specific: prices, real photographs, a calculator, one detailed question actually answered.
Intent decay. The query stayed the same, the expectation behind it did not. "What is a heat pump grant" has become "apply for a heat pump grant", because the topic moved from the information phase into the action phase. Your explanatory page keeps collecting impressions and loses clicks. This is the case where a rewrite is not enough: the page has to take on a different job, or it needs a second page beside it.
Before you touch a line, decide which of the three you are looking at. The wrong treatment is wasted work, not merely half-effective work.
The candidate list comes from the report, not from the gut
Four buckets, all from the Search Console performance report. Give it an hour once and write the results into a spreadsheet.
Bucket one: falling clicks at steady demand. Compare the last three months against the same period last year, not against last month. Pages whose clicks fall while impressions hold are losing the click decision. Pages where both fall are losing visibility. How to tell the patterns apart is in lost rankings.
Bucket two: pages at position four to fifteen with plenty of impressions. These are the pages where a little work moves a lot, because they are nearly there. Sort by impressions, not by clicks, or you will only find what already works.
Bucket three: pages ranking for queries they do not answer. Filter to a single page and read its queries. This is the most productive half hour in the whole exercise. You reliably find questions you never considered that Google shows you for anyway. Each one is a missing section.
Bucket four: two pages taking turns on the same query. Filter to one query and see which pages appear for it. If two alternate week by week, they are competing with each other and neither gets far up the page. That is a merge case, not a rewrite case.
Changing the date is not an update
The most widespread practice is also the most useless: set the date on the post to today, reshuffle two sentences, done. Behind it sits the idea that Google rewards freshness as a property. It does not, not as an end in itself; it assesses whether the page answers the question better than the competition. A date is not an answer.
Worse, it costs you two things. First trust, because a reader who sees "updated on" and then a screenshot of an interface from three versions ago will not believe your next figure either. Second, measurability. If you spin the date on twenty pages and genuinely work on three of them, you cannot say afterwards which change did anything.
And the opposite direction, which goes wrong just as often: longer is not better. If a page runs to three thousand words because a word count was mandated back then, cutting is the improvement. What you delete improves the page as much as what you add.
What a rewrite actually involves
In this order, because the early steps carry most of the effect:
- Read the page's queries and add a section for every question left unanswered. With a subheading, so it can be found.
- Replace outdated facts: numbers, deadlines, names, screenshots. If you cannot source a figure, remove it rather than estimate it.
- Compress the first screen. Answer at the top, introduction cut back. Anyone who has to run three hundred words before the answer arrives has already left. What users actually do at that point is in analyzing user behavior.
- Set internal links in both directions. From the revised page to related ones, and more importantly from your strong pages to this one. The second part is almost always forgotten and is the more effective one.
- Sharpen the title and snippet once the content stands. The method is in title and meta description.
- Leave the URL alone. A new address for the same content means a redirect, broken comparison periods in your reports and extra risk, for nothing. If the URL is genuinely wrong, that is its own project, not a side effect of editing copy.
Merge rather than multiply
If bucket four turned up two pages competing for the same query, the job is not to improve both. It is to keep one.
The sequence: take the stronger URL, meaning the one with more clicks and more external links. Work the best of the weaker one into it. Redirect the weaker one permanently to the stronger. And then the step that usually gets skipped: change the internal links that pointed at the old page so they point straight at the new one, not through the redirect. The details on redirects and canonicals are in canonical tags done right.
Two mediocre pages become one good one. Five thin location pages with the town name swapped out become one service page with an honest paragraph about the area covered, and that is usually a gain rather than a loss.
When deleting is the right answer
Deleting is allowed, but it is rarely the lever it gets sold as. The idea that a site gets assessed better overall once weak pages are removed is, as a mass measure, a bet. What deleting reliably delivers is less maintenance, less confusion, and fewer pages competing for the same queries.
The four conditions that should hold together: no clicks, barely any impressions, no external links, no job in the sales process. If one of them is missing, rewrite or merge instead of deleting. And if you do delete, decide deliberately what the address answers afterwards: a redirect to the matching page if one exists, otherwise a clear error. What Search Console reports about that is in page not indexed.
Measuring without fooling yourself
Three rules, or you will know nothing afterwards.
Write down the date of every revision. With no date there is no before and no after.
Wait four to eight weeks. Google has to recrawl and reassess, and the first two weeks are noise. Google states no deadline for it.
Compare against the same period last year, not last month, and look at position and CTR rather than clicks alone. Otherwise you will write the autumn's seasonal rise into your notes as the success of your editing. And work in groups of two to four pages instead of twenty at once, so attribution stays possible at all.
A rhythm that survives
Two to four revised pages a month beats the annual stocktake that stops happening after year three. Put it in the calendar like maintenance, and work the list in order of contribution: the pages collecting the most impressions today go first.
Traffalyzer takes the mechanical part off your hands: assembling the candidates from the four buckets and flagging a page when it starts to decay. Which of the three kinds of decay it is, and what the page ought to say instead, is still decided by whoever knows the business.
Where to start
- Open the performance report, last three months against the same period last year.
- Note the ten pages with the largest click loss.
- For each one decide: factual, competitive or intent decay?
- On one of them, read the queries and write down the unanswered questions.
- Revise that single page, note the date, leave the URL untouched.
- Wait six weeks, then compare position and CTR.
- Only then move to the next page.
Frequently asked questions
How often should I update my content?
Not by calendar but by trigger. A trigger is: the facts no longer hold, clicks are falling at steady demand, or the queries show questions the page does not answer. As a working rhythm, two to four pages a month is realistic and more effective than one big campaign a year.
Does changing only the date improve rankings?
No. Google assesses whether the page answers the question better than others, not whether a number at the top of the article is new. A changed date with unchanged content costs you credibility with readers and the ability to tell later which change actually did anything.
Should I delete old blog posts?
Only when four things hold together: no clicks, barely any impressions, no external links, no job in the sales process. Otherwise rewriting or merging is the better call. Bulk deletion as a ranking measure is a bet lost more often than won, and it cannot be undone.
How do I spot two of my pages getting in each other's way?
Filter the performance report to a single query and look at which pages get shown for it. If two alternate over the weeks, they are competing for the same intent. Then the job is not to improve both but to keep the stronger and redirect the weaker to it permanently.
Write a new page or revise the old one?
Revise, as long as the old page serves the same intent. A new page is right when the intent has split: an informational question and a buying intent belong on two pages, because one page cannot serve both well. If you are unsure, read the existing page's queries: two clearly separate kinds of question mean you need two pages.
How long does an update take to have an effect?
Google has to recrawl and reassess. That takes days to weeks, and a reliable read on the effect arrives after four to eight weeks. Google states no deadline. Checking the chart after five days shows noise, and noise invites wrong conclusions.
Do I need to change the URL if I rewrite the content heavily?
No, and you should not. The old address carries history, links and assessment. A new address means a redirect, broken comparison periods in your reports and extra risk with nothing in return. Only if the topic of the page shifts completely is a new address even worth considering.
How much do I have to change for it to count?
There is no threshold in percent or words, and any figure quoted for it is invented. The right question is not how much you changed but whether the page now answers something it previously left open. One added section that closes a real gap does more than a thousand reworded ones.
Do longer texts help?
Only when the length comes out of substance. Length is a side effect of answering a question completely, not a target. On pages written years ago to hit a word count, cutting is nearly always the improvement: answer to the top, run-up gone, filler paragraphs out.