By 
Shreyas M V · Published · 14 min read
Content Refresh Playbook: Update, Merge, Redirect or Delete?
Decide each old page by asking six questions in order: is it still relevant, does another page of yours answer the same query, does it still match what searchers want, is it accurate, is it showing a ranking signal, and is it earning anything at all. The answers send it to keep, update, overhaul, merge, redirect or delete. This page gives that decision tree and a refresh log to record every call. The tree is Kanso's reading of four sources, and none of them measures whether these decisions raise rankings.
What is the short answer?
Run the six questions in the decision tree in order and stop at the first one that gives an outcome. The rest of the page explains each outcome and gives you a log to keep the decisions in.
- Surfer frames the decision as keep, update, overhaul or remove. Powered by Search adds merging as a fifth move, and Kanso puts both into one tree.
- Pages that already earn impressions and sit just off page one are the first candidates. Benchmark Email also names pages not updated in 12 months or more.
- Merge overlapping pages into the strongest one and 301 redirect the rest. Expect a dip of a few weeks, according to Powered by Search.
- Check facts against their sources, not against the wording. A new date on an unchanged page is not a refresh.
- Log the baseline before you change anything. No source read here measures the effect of a refresh, so your own log is the only evidence you will have.
What is a content refresh, and what is it not?
A content refresh is a deliberate update to an existing page that improves its accuracy, relevance, depth or search performance while keeping its URL.
Benchmark Email defines it in nearly those terms and separates it from a full rewrite, which treats the piece as essentially new, and from minor copyedits. Surfer frames the same idea as updating a page's facts, links, structure, on-page SEO and media so it matches what searchers want today.
Benchmark Email is blunt about what does not count: a cosmetic reskin, a minor word swap, or a paragraph of filler added to make a page look longer. It recommends updating the publish date or adding an Updated note once the refresh is done, not instead of it.
Most guides stop at the update. This page treats the update as one of several outcomes, because the pages that need a decision are often not the pages that need an edit. Content pruning, the practice of removing pages that no longer earn their place, is the same decision made in the other direction.
What is the decision tree for an old page?
Ask six questions in order and stop at the first one that gives you an outcome. The order is deliberate: relevance and overlap come before accuracy, because there is no point fact-checking a page you are about to merge away, and the performance check comes last, because a well-ranked page that is wrong still has to be fixed.
Kanso built the tree from four documents read on 5 November 2026. Surfer supplies the four actions (keep, update, overhaul, remove), Powered by Search adds merging, Benchmark Email supplies the candidate criteria, and Google's help page covers what happens when a deleted page still shows in results. The order of the questions and every rule marked as Kanso's are our judgement and have not been tested.
- Q1. Does the page still serve a topic your business and audience care about? If not, go to Remove. If yes, continue.
- Q2. Does another page on your site answer the same query for the same reader? If yes, go to Merge. If not, continue.
- Q3. Does the page still match what the results for its main query now show? If not, Overhaul it: change the angle or the format, or combine it with similar pages. If yes, continue.
- Q4. Is every date, statistic, tool name, price and link still right? If not, Update first and fix the facts before anything else. If yes, continue.
- Q5. Does it show a ranking signal: positions 11 to 20, high impressions with few clicks, or a decline? If yes, Update, with changes matched to the signal. If not, continue.
- Q6. Does it earn any impressions, clicks or links? If yes, Keep it and recheck at the next review. If not, Park it, which is Kanso's rule: log it, recheck at the next review, then merge or remove it if nothing has changed.
The table below puts the outcomes in one place, with the source for each. Two columns, “choose it when” and “what to do”, are taken from the sources named in the last column, rewritten here. The tree that routes pages to them is ours.
| Outcome | Choose it when | What to do | Source |
|---|---|---|---|
| Keep | The page is accurate, still matches the query and is performing | Leave it alone and recheck it at the next review | Surfer |
| Update | The topic and intent are right, but details are old, sections are thin, or a ranking signal says it is close | Refresh facts, links, examples and on-page elements; rewrite the title and description if clicks are low | Surfer, Benchmark Email |
| Overhaul | The search intent has changed, or the page's angle no longer fits what readers want | Rewrite it, change its format, or combine it with similar pages | Surfer, Benchmark Email |
| Merge | Two or three of your pages answer the same question | Fold them into the strongest page and 301 redirect the others to it | Powered by Search |
| Remove | The page is outdated or irrelevant and no longer serves your audience | Delete it with a 301 redirect, or use noindex or a canonical tag | Surfer |
Which old pages should you check first?
Check pages that already have search impressions and decent rankings but have not been updated in 12 months or more, because Benchmark Email names those the highest-ROI refresh targets. Surfer adds three groups that usually pay off first: pages ranking just off page one, pages with many impressions and few clicks, and pages whose traffic or rankings are falling.
| Signal | Why it matters | Source |
|---|---|---|
| Positions 11 to 20 | Ranks just off page one, so a focused update may be enough | Surfer, Benchmark Email |
| High impressions, low click-through rate | Shown often but rarely chosen, which points at the title and description first | Surfer |
| Declining traffic, engagement or rankings | Used to perform, so something about the page or the results has changed | Surfer |
| Impressions and decent rankings, not updated in 12 months or more | Benchmark Email calls these the highest-ROI refresh targets | Benchmark Email |
| Statistics, tools or events more than 18 months old | Dated references likely need updating | Benchmark Email |
| Core business topics with sustained search interest | A moderate-traffic page on a core topic beats a high-traffic page on a side topic | Benchmark Email |
Both guides call positions 11 to 20 the “almost page one” group, and neither gives a measured rate for how often a focused update gets a page onto page one. Treat the group as a place to look first, not as a promise. For the Search Console steps, and for the pages that should be left alone, Kanso's striking-distance query workflow goes through them in order.
To find the candidates you need an inventory. Surfer suggests crawling the site for URLs, page titles, meta descriptions, word counts and last-modified dates, then joining them with Google Analytics and Search Console figures: views, engagement, impressions, clicks, click-through rate, average position and top queries. Benchmark Email keeps it smaller: URL, publish date, last updated date, primary topic and current search position.
Powered by Search also says to look beyond search. A page that was widely shared on social media but has decayed may be worth refreshing, because a timeless topic can reach people who missed it the first time.
When should you update a page instead of rewriting it?
Update when the page's topic and intent are still right and the problems are details: old facts, broken links, thin sections or a weak title and description. Benchmark Email says to reserve a rewrite for a page whose angle or structure is fundamentally misaligned with current search intent, and Surfer puts that case under overhaul.
The work that sources attach to an update falls into five groups.
- Facts: replace old statistics, update examples and case studies, remove information that no longer reflects reality (Surfer, Benchmark Email).
- Links: fix broken ones and add internal links to pages published since the original (Surfer, Benchmark Email).
- Coverage: add sections for questions the page does not answer and remove sections that are no longer accurate (Benchmark Email).
- Opening and structure: rewrite the introduction to match current intent, and add a key-takeaways section or an FAQ if the page lacks one (Benchmark Email).
- Snippet: rewrite the title tag and meta description when impressions are high and clicks are low (Surfer, Powered by Search).
Powered by Search describes title and description changes as a test with a control and one challenger, and warns against running more than one new version at a time because it risks confusing Google on each crawl. It reports that it has seen rankings improve when more people clicked a new version, which is a practitioner observation without figures.
One caution is Kanso's own. Powered by Search suggests mentioning the current year in a title, and a year in a title is a claim that the page was checked for that year, so add it only when the content really was.
When should you merge two pages into one?
Merge when two or three of your own pages answer the same question for the same reader, so that one stronger page can replace them. Powered by Search's method is to pick the most authoritative of the group as the main page, combine the information from the others into it, and 301 redirect the URLs of the pages you did not choose.
The same guide says to expect a dip in organic traffic after a merge, because the redirected pages lose their rankings for their old keywords and Google needs time to adjust. It says traffic can take a few weeks to recover and gives no figures, so check the dates in your own log rather than assuming a number.
Powered by Search also argues that merging gathers more backlinks and ends cannibalization, which is when your own pages compete for one search result. Those are its claims, offered without data. If you want to read how Kanso handles cannibalization as a weekly check, the weekly growth checklist has an entry on it.
Telling real overlap from pages that merely share a topic takes three tests. They are Kanso's, and none of them comes from the sources read.
- Do both pages show for the same top queries in Search Console?
- Would a reader who landed on either page have the same need met?
- Could one page cover both without becoming a list of unrelated sections?
If the answers are yes, yes and yes, merge. If a reader of one page would want something different from the other, leave both and sharpen how they differ.
When should you redirect a page and when should you delete it?
Redirect a page when a close replacement exists and delete it when none does; that rule is Kanso's, because the guides read for this page list the options without ranking them. Surfer's “remove” action says to delete the page with a 301 redirect, or to use noindex or a canonical tag, for content that is outdated or irrelevant and no longer serves the audience.
| Route | Use it when (Kanso's rule) | Where it comes from |
|---|---|---|
| 301 redirect | A close replacement exists, so readers and links should land on it | Surfer and Powered by Search both name it |
| Noindex | The page should stay reachable for readers but stay out of search results | Surfer names it |
| Canonical tag | A near-duplicate has to stay live and one version should count as the main one | Surfer names it |
| Delete with no redirect | No page on the site serves the same need, so a redirect would send readers somewhere unrelated | Not covered by the sources read; Kanso's rule |
Powered by Search gives a concrete case: product-announcement posts go stale, and instead of leaving them live their content can be folded into an existing product roadmap page. That is a merge followed by a redirect, which is why the tree sends pages to Merge before it sends them to Remove.
A deleted page can keep showing in Google for a while. Google's help page says that if content was deleted but still comes up in results, the description or cache may be outdated, and it offers a Request refresh form in the tool for that case. The same page says you do not need to use the form if you have already updated a live page, and it gives no timeframe.
On the evidence question: none of the pages read for this guide measures what pruning does to rankings. Prune because a page is wrong, off-topic or duplicated, and write that reason in the log, not because you expect a lift.
How do you check that a page is still accurate?
Check each time-sensitive claim against its source rather than against the sentence, because a reworded sentence can repeat an old fact. Benchmark Email treats any reference to a year, tool, statistic or event more than 18 months old as a likely update, and it warns that dated content erodes credibility.
Reworking text at volume makes this harder, not easier. A page rewritten quickly can read as new while keeping a figure from three years ago, so check the claims most likely to have changed first.
| Check | Why it comes up | What to do |
|---|---|---|
| Statistics and year references | Benchmark Email says anything more than 18 months old likely needs updating | Open the source, confirm the figure and its date, replace or remove it |
| Tools, features and screenshots | Powered by Search gives outdated screenshots and broken links as the example of a page that needs light cleanup | Open the product's own page, confirm the name and what it does today |
| Examples and case studies | Surfer says to refresh examples so they reflect current trends or recent events | Check each example still exists and still shows what the text says it shows |
| External and internal links | Surfer says to repair or replace broken links | Click each one, and add links to pages published since the original |
| Claims about what a rule or policy says | Kanso's check: a rewritten sentence can repeat an old fact | Go back to the primary document and read it again |
When you replace a figure, put the month and the source in the same sentence, so the next reviewer can see how old it is. Kanso's claim-by-claim source verification workflow sets out how to extract the claims, open each source and log the result, and it applies to human edits as much as to AI drafts.
What should a content refresh log record?
A refresh log should record, for each page, what you decided, which question decided it, what you changed, when it shipped and the baseline you will compare against. The columns below are Kanso's design; the check dates and the next-review interval are derived from the figures Benchmark Email and Surfer give.
| Column | What to enter | Why it is there |
|---|---|---|
| URL and primary query | The page's address and the one query it should win | Keeps the decision tied to a page and a purpose |
| Review date | The day you made the call | Dates the evidence you used |
| Last updated (before) | The page's own last-updated date before any change | Benchmark Email's inventory tracks this; it feeds the 12-month and 18-month checks |
| Signal | Positions 11 to 20, low click-through rate, declining, stale facts, overlap, or none | Records why the page came up |
| Question that decided it | Q1 to Q6 from the tree | Lets you audit the rule later |
| Decision | Keep, update, overhaul, merge, redirect, delete or park | The outcome |
| Changes made | What was edited, added, removed or redirected | Needed to read any result honestly |
| Redirect or merge target | The URL that now answers the query | Needed for every merge and redirect |
| Baseline | Clicks, impressions and average position for a fixed window before the change | Without it there is nothing to compare with |
| Shipped date | The day the change went live | Starts the clock for the checks |
| Check dates | 4 weeks, 8 weeks and 90 days after shipping | Derived from Benchmark Email's 4 to 8 weeks and 60 to 90 days |
| Result and notes | The same three numbers at each check, and anything else that changed | Records what you saw, not what you hoped |
| Next review | 6 to 12 months out | Surfer's review interval |
Copy the line below into a spreadsheet header row.
URL | primary query | review date | last updated | signal | question | decision | changes | target | baseline | shipped | checks | result | next review
Fill the baseline with the same window every time, and record the change before you publish it. If you cannot say afterwards what you changed, you cannot say what a movement means.
Five hypothetical pages through the tree
The pages below are invented to show the routing and contain no data. They are not Kanso pages and not results.
| Page | Question that decides it | Decision | Log entry |
|---|---|---|---|
| A tool roundup from several years ago; the topic still matters, but two named products have changed | Q4: facts are no longer right | Update | Replace the product details, confirm each from the product's own page, fix links, record a baseline |
| Two posts on the same how-to, written a year apart | Q2: another page answers the same query | Merge | Pick the stronger page, fold in what the other adds, 301 redirect the weaker one |
| A launch post for a feature that no longer exists | Q1: no longer relevant | Redirect or delete | 301 redirect to the current feature page if one exists, otherwise delete |
| An accurate, relevant post that ranks around position 14 | Q5: positions 11 to 20 | Update | Targeted changes to the sections and title tied to the queries it shows for |
| An accurate, relevant post with no impressions and no links | Q6: earns nothing yet | Park | Log it, recheck at the next review, then merge or remove if nothing has changed |
How long does a refresh take compared with a new post?
Benchmark Email says refreshing an existing 1,200-word post typically takes 20 to 40 percent of the time required to create a new piece on the same topic. It gives no method for the figure, so use it for planning and not as a promise.
Here is the arithmetic with an assumption stated up front. Suppose a new 1,200-word post takes 10 hours. That is an assumption for this example, not a sourced figure.
- Low end: 20 percent of 10 hours is 0.20 x 10 = 2 hours.
- High end: 40 percent of 10 hours is 0.40 x 10 = 4 hours.
- So each refresh costs 2 to 4 hours under that assumption.
Benchmark Email's cadence adds more numbers. A monthly review of 20 to 30 minutes is 12 x 20 = 240 minutes at the low end and 12 x 30 = 360 minutes at the high end, which is 4 to 6 hours of review a year. Flagging 2 to 4 pieces a month is 2 x 12 = 24 to 4 x 12 = 48 flagged pieces a year.
If every flagged piece were refreshed at the assumed cost, that is 24 x 2 = 48 hours at the low end and 48 x 4 = 192 hours at the high end. The spread is the point: decide how many hours a month the refresh gets, take that many pages from the top of the priority list, and park the rest. No source read gives a time for a merge or an overhaul, so log your own.
How often should you review old content, and when can you judge a refresh?
Surfer recommends reviewing blog content every 6 to 12 months, and Benchmark Email suggests a monthly 20 to 30 minute review that flags pieces as they come due. For email sequences, Benchmark Email suggests a quarterly review. Neither interval comes with a controlled basis, so pick the one your team can keep.
Judging comes later than shipping. Benchmark Email says search engines typically take 4 to 8 weeks to re-crawl and re-evaluate an updated page and recommends watching position, traffic and engagement for 60 to 90 days. Powered by Search says a merge dips first and recovers over a few weeks.
These are vendor statements, and Google's help page gives no ranking timeline. That is why the log sets check dates at 4 weeks, 8 weeks and 90 days and records all three, instead of one verdict at a time you picked after the fact.
A before-and-after comparison has a limit that no log removes: it cannot tell your edit from seasonality, a search update or new links. If you can leave a few comparable pages untouched for the same weeks, you get a comparison group. Kanso's pre-registered content refresh experiment design shows how, and it states that it has no results yet.
What is documented and what is Kanso's analysis?
The documented items are statements the four sources make, attributed to them; the analysis items are Kanso's own reading and have not been tested. All three practitioner guides come from companies that sell search or content tools or services, which is a reason to treat their claims as opinion.
Documented, as of November 2026, per the source named:
- Surfer: the four actions keep, update, overhaul and remove; removal by 301 redirect, noindex or canonical; priority groups (positions 11 to 20, high impressions with low click-through rate, declining pages); review every 6 to 12 months.
- Powered by Search: five levels of refresh from title and description changes to a redesign; merging by choosing a main page and 301 redirecting the others; a traffic dip of a few weeks after a merge.
- Benchmark Email (17 June 2026): highest-ROI pages have impressions and decent rankings but no update in 12 months or more; a refresh takes 20 to 40 percent of the time of a new post; a monthly 20 to 30 minute review; 4 to 8 weeks to re-crawl; 60 to 90 days of monitoring.
- Google Search Help: a Request refresh form exists for deleted content that still appears in results, and no request is needed if a live page has already been updated.
Kanso's analysis, untested:
- The order of the six questions, and the choice to check overlap before accuracy.
- The redirect-or-delete rule and the park rule for pages that earn nothing.
- The three overlap tests and the refresh log columns, including the 4-week, 8-week and 90-day check dates.
- The five hypothetical pages and the 10-hour assumption in the time arithmetic.
What none of this covers: how often a refresh, a merge or a prune actually moves rankings. The pages read describe the actions and their effects in general terms and report no controlled result, and the pages that appeared only as search-result titles were not read.
Where does the SEO Agent fit into a content refresh?
The SEO Agent can supply evidence for the first pass, because it audits your site and Search Console data and turns what it finds into recommendations and tracked tasks for your approval. That covers finding candidates; it does not make the keep, update, merge or delete decision for you.
Kanso does not publish to a CMS and does not promise rankings, so the edits, the redirects and the log stay with your team.
Content refresh: frequently asked questions
What is a content refresh?
A content refresh is a deliberate update to an existing page that improves its accuracy, relevance, depth or search performance. Benchmark Email separates it from a full rewrite, which treats the piece as new, and from minor copyedits, which do not change the page's value.
What is content pruning?
Content pruning is removing, merging or redirecting pages that no longer earn their place on a site. In the guides read for this page it is the remove branch of a content audit: Surfer lists deleting a page with a 301 redirect, or using noindex or a canonical tag, as the options, and none of the pages read measures what pruning does to rankings.
How do you update old blog posts?
Check the page against its current search results, then fix what is wrong: replace old statistics, repair links, refresh examples, add missing sections and tighten the title and description. Benchmark Email adds updating internal links to content published since the original, and recording a new date or an Updated note once the work is real.
How often should you refresh blog content?
Surfer recommends reviewing blog content every 6 to 12 months, and Benchmark Email suggests a monthly 20 to 30 minute review of the content inventory that flags 2 to 4 pieces. Neither gives a controlled basis for the interval, so treat both as practitioner guidance.
Which old posts should you refresh first?
Start with pages that already have search impressions and decent rankings but have not been updated in 12 months or more, which Benchmark Email calls the highest-ROI targets. Surfer adds pages ranking in positions 11 to 20, pages with many impressions and few clicks, and pages whose performance is declining.
When should you merge two posts instead of updating one?
Merge when two or three of your own pages answer the same question for the same reader. Powered by Search recommends making the most authoritative page the main one, combining the others into it and 301 redirecting the weaker URLs, and warns that traffic usually dips for a few weeks afterwards.
Does Google prioritize fresh content?
The sources read for this page assert it without quoting a Google statement. Benchmark Email says search engines factor freshness into rankings, particularly where recency matters, and Powered by Search points to Query Deserves Freshness, so treat freshness as a reason to keep facts current rather than a ranking lever.
How long does it take for Google to refresh outdated content?
Google's help page gives no time. It says that if you have already updated a live page you do not need to submit a request, and that a request is for a deleted page still showing outdated text in results. Benchmark Email says search engines typically take 4 to 8 weeks to re-crawl and re-evaluate an updated page, which is a vendor estimate.
How do you measure whether a content refresh worked?
Record clicks, impressions and average position before you change the page, then check again at fixed dates. Surfer says to track rankings, traffic and click-through rate, and Benchmark Email says to watch position, traffic and engagement for 60 to 90 days, but a before-and-after comparison cannot separate your edit from seasonality or a search update.
What are common mistakes to avoid when refreshing content?
The mistakes the sources name are cosmetic changes, such as a date change or a word swap, padding that makes a page longer without improving it, and testing more than one new title at a time. Kanso adds two more that its tree is built to prevent: refreshing pages nobody searches for, and merging without redirecting the old URLs.
How do you check factual accuracy when refreshing content at scale?
Check each dated or numeric claim against its source, not against the sentence, and log what you checked. At volume the failure is a page that reads newer but repeats an old fact, so sample the claims most likely to have changed first: figures, tool names, prices and policies.
Can an AI agent refresh old content on a schedule?
Not in Kanso. The SEO Agent produces recommendations and tracked tasks for your approval, Kanso does not publish to a CMS, and scheduled agent runs are not available. An AI tool can help build the candidate list, but the keep, update, merge or delete decision and the fact check stay with a person.
Sources
- 5-Step Content Refresh Guide for High-Performing Blogs, Surfer
- Content Refresh: The Powerful Yet Simple Solution for Improving Your Search Traffic, Powered by Search
- Stop Letting Good Content Fade: How to Make Content Refreshing Part of Your Strategy, Benchmark Email, published 17 June 2026
- Refresh outdated content, Google Search Help
Note: Facts are attributed to the page that states them and were read on 5 November 2026. Surfer, Powered by Search and Benchmark Email sell tools or services, so their claims are treated as practitioner opinion. The decision tree, the rules marked as Kanso's and the refresh log are untested. Kanso has no affiliation with Google or the three companies.
Playbooks