Key takeaways
- Start your Q4 review with your archive, where you can check what each article attracted and earned.

When a publisher wants more revenue before the year closes, the first thought is usually to pay a writer for something new, like a new series or another gift guide. I get it. New feels like progress.
But I think the best bet you can make this quarter is probably already sitting in your archive. You paid for it a year or three ago. It brought in readers, and some of them were worth a lot to advertisers. With that article, you don't have to guess. You have the receipts.
In short, a content refresh means taking an article that has already proven it can attract readers and earn revenue, improving it in a way a returning reader would notice, and putting it back in front of an audience. To choose which articles deserve that work, match at least a year of traffic data to ad revenue for the same dates, look past RPM to total earnings, read the search queries behind each page, and fund only the updates with a clear reason to exist.
Below is how I'd do it, step by step, using data sources I trust.
Your archive shows which articles have brought in readers and revenue. With the holiday advertising window ahead, that gives you a useful place to start deciding where to spend.
People have been measuring this for years. When HubSpot's Pamela Vaughan analyzed HubSpot's own blog, she found that 76% of monthly blog views and 92% of monthly blog leads came from posts published before the current month. Just 30 posts produced 46% of the monthly leads, out of a library of roughly 6,000. When the team updated old posts, organic search views of those posts rose by an average of 106%. That's old data from a B2B blog, not an ad-supported publisher, so I wouldn't borrow the numbers. I'd use those results as a reason to check whether a small number of pages are doing much of the work on your own site.
Now add the calendar. The ad management company Mediavine describes Q4 as consistently the highest-earning quarter, with CPMs and fill rates surging in November and December as advertisers race to spend the rest of their annual budgets. It also says the "real" Q4 is shorter than the calendar one. In the ad business, it's a period beginning on Black Friday and ending a few days before Christmas. So if you're reading this in mid-November, you haven't missed it. That holiday advertising window is still ahead, so there's a reason to get your strongest updates in front of readers now.
Look at 12 months at least, and further back if your tools let you. One recent month tells you what's moving now. It can't tell you which articles have lasting value.
This matters because two very different kinds of winners look the same in a 30-day report. A page can have an exceptional week because a news event made its subject relevant for a moment. Another page can quietly attract readers every autumn for three years. You'd treat those two pages very differently, and you can only tell them apart if you look back far enough.
Each tool only keeps so much history, so know the limits before you start.
Then comes the boring part. Take the article URLs from Analytics and match them with Mediavine revenue (or the page-level reporting from whichever ad platform you use) for the same dates. Keep the periods aligned. Traffic from one period and revenue from another won't tell you what those visits actually earned.
For each article, you want to know how much traffic it got, what it earned, where the visitors came from and how that's changed over time. Then, and this is the step people skip, open the article and read it. The numbers tell you where to look. You won't know what you've got until you read it.
Before you try to repeat a success, find out what caused it. Was it steady search traffic, or one big moment on Facebook? And were those readers worth much to advertisers?
I ask the same questions about every article on the list.
Was the article answering a question people keep searching for? Did one Facebook post send most of its visitors? Has it performed across several different periods, or does almost everything trace back to one viral run? And did the readers it attracted generate worthwhile ad revenue, or just pageviews?
For example, say a home publication has a guide to setting up a small guest room. It gets search traffic all year and has done well on social before the holidays in the past. So the publisher knows people still care about the subject, and knows a way of reaching them that has worked before.
Now compare an old story about a product recall that also earned a lot. If the issue has been resolved and nobody needs the information anymore, its past revenue doesn't make it a good Q4 assignment. The money was real, but nobody's looking for that story anymore.
So you can't just read the totals. Past success is a clue, and it doesn't promise a repeat. For an evergreen piece, I'm looking for a continuing problem, question or decision that the publication can still help with. The details may need updating, but the reason someone would read it is still there. That's a much better place to start than sharing the old link again and hoping the spike comes back.
It's a reason to look closer. On its own, it shouldn't decide anything. RPM tells you the rate a page earned on the traffic in that report. It says nothing on its own about how much money is at stake.
Quick definitions first. RPM means revenue per mille, or revenue per 1,000 units of traffic. Page RPM is earnings divided by pageviews, multiplied by 1,000, which is how Google AdSense defines it. Session RPM divides by sessions instead, and Mediavine uses sessions because, in the words of its ad tech glossary, they better capture the value of your traffic.
Which number you divide by matters more than you'd expect. On our own homepage, we publish the June 2026 numbers from one site we manage: $142,696.79 in ad revenue from 1,766,092 pageviews and 1,440,024 sessions. That works out to a session RPM of $99.09 and a page RPM of about $80.80. That's the same site and the same month's revenue, with an $18 gap depending on what you divide by. When you compare RPMs across reports, make sure they're the same kind of RPM.
This is where people get caught out. Take two articles over the same period.
Article B has double the rate. Article A made four and a half times the money and generated nine times the pageviews. Either one might deserve an update, but you can't pick sensibly by staring at the RPM column.
Mediavine says this pretty directly. Its page-level guide says it does not recommend sorting RPM highest to lowest, because pages with small amounts of traffic can produce inaccurate RPM calculations. The same guide has a better rule of thumb, and I like it a lot. It says posts with high CPM and high traffic are the most valuable. Mediavine's Q4 explainer also tells publishers to think about total revenue rather than RPM alone.
What I'd really want to know is whether plenty of people wanted the article, and whether you understand why it paid. A rate earned during one promotion is evidence of what happened then. You're deciding where to invest, and the number of readers you can realistically reach matters as much as the rate. If you want to raise the rate itself, we cover how in how to get better RPMs with programmatic ads.
They tell you what readers are still trying to figure out, and whether your article actually helps them do it. Read the query list as questions your readers are asking. It's more useful that way than as a keyword report.
Google Search Console's Performance report shows clicks, impressions, click-through rate and average position by query and page. Its advanced filtering lets you isolate one URL, compare this period with the same period last year, and use regular expressions to group related queries. Google also warns that filtering by query or URL can change the totals, because anonymized queries are left out. So the list you see is only part of the demand.
Back to the guest room guide. The queries might show people searching for ways to fit a bed into a small space, add storage, or make the room useful when no one's visiting. If the existing article is a list of decorating ideas, you've found the gap. The update might include measured layouts, original photographs and a comparison of practical storage options. Each addition answers something the audience asked for. That's a far better brief than "freshen up the wording."
Think about what a keyword is worth to your business in particular. A popular search can bring attention without much income, and a smaller, very specific subject can support an article that earns well and attracts exactly the readers you want. Look at search volume and page revenue side by side, and think about what the person searching actually wants.
This step should also tell you when to leave an article alone. If interest has faded, the subject no longer fits your publication, or a real update would cost more than the opportunity is worth, move on. Deciding what to skip is half the point of doing this review.
An update is worth it when a returning reader would notice the difference. If you can't name the specific improvement, you don't have an update yet. You have a date change.
Check the facts, recommendations, links and examples. What needs correcting? What useful detail was missing the first time? What have readers asked in comments or emails since it went up? Keep what still works. Then put your time into the parts that make it better, like a detail you've checked yourself, a clearer explanation, a demonstration or some original reporting.
Google's own guidance on helpful content is useful here because it's so specific. It asks whether a page provides original information, reporting, research or analysis. It specifically calls out "changing the date of pages to make them seem fresh when the content has not substantially changed" as a warning sign. And it says adding or removing lots of content just to make a site seem fresh won't help rankings. The same page describes how Google's quality raters assess effort and originality, including whether content offers information or perspectives that aren't already available elsewhere.
I use one test. If someone used the guide last year, what would they gain by reading it again now? The answer should be specific enough to become an assignment and clear enough to put in the social post that announces it.
If you're updating an article anyway, use that work to make the answers more specific and easier to check. The research below gives you ideas for the brief, although it doesn't establish that refreshing your page will earn a citation.
A quick definition. Generative engine optimization (GEO) is the practice of shaping content so that AI answer engines like Google AI Overviews, ChatGPT, Gemini, Perplexity and Microsoft Copilot are more likely to use it and cite it. The researchers behind the GEO paper, first shared in 2023 and accepted at KDD 2024, studied ways to improve visibility in generative answers.
The click data shows why that matters for your archive. In a Pew Research Center study of 900 U.S. adults' browsing in March 2025, Google users who saw an AI summary clicked a traditional result in 8% of visits, compared with 15% for those who didn't. Only 1% of visits with an AI summary produced a click on a link inside the summary. For your own business, the next question is whether those citations send readers back to your site, and what those visits earn.
Three studies give a decent idea of what to add.
Look at what the freshness research actually found. Ahrefs analyzed nearly 17 million citations across AI platforms and found that the pages AI assistants cited were, on average, about 2.9 years old, compared with 3.9 years for pages in Google's organic results. ChatGPT leaned furthest toward newer pages, while Google AI Overviews behaved much like regular search. The study is correlational, and cited pages were still almost three years old on average. It doesn't tell us what would happen if you refreshed your own article, so use it as context rather than a reason to chase dates.
Specific claims stood out in one small study of cited passages. In an August 2026 passage-level study by Advanced Web Ranking, passages that Google AI Overviews and Bing Copilot cited showed a visible 2025 or 2026 date 80% of the time, versus 53% for passages that were surfaced but not cited. Cited passages named their subject at first mention 96% of the time, versus 82%. And none of the 17 uncited passages contained a hard number or a new claim. It's a small study (112 coded passages from 40 queries), and the author says so, but the takeaway is easy to use.
Sources and numbers help. The original GEO paper found that adding citations, quotations from credible sources and statistics could raise a source's visibility in AI-generated answers by up to 40%, while old-school keyword stuffing didn't help.
So when you update that guest room guide, the AI version of the brief is the same as the reader version, just more exact. Put the subject's name in the first sentence under each heading. Add the measurements, the storage comparison and the numbers you checked yourself, and cite where they came from. Show an honest "updated" date that reflects real changes. We go through the sentence-level detail in writing for AI citations.
Build new posts based on what worked last time. Don't just repost the old one. Look at how the old posts did, then write fresh ones that point to the updated article.
Go back to the social activity that previously sent traffic. Look at the post, the creative, the timing and the angle that hooked people. Then compare that response with the website visits and revenue you can actually trace. A post that spread widely is useful evidence, but you also want to know whether those readers made the visit valuable. That's how you know what to do again and what to drop.
Meta cares a lot about originality right now. In March 2026, Meta updated its original content guidelines and reported that views and time spent watching original Reels on Facebook roughly doubled in the second half of 2025 compared with the same period in 2024. It also said duplicative content, or minor edits to someone else's post, will be treated as unoriginal and deprioritized, and that accounts posting mostly unoriginal content may be demonetized. Your own photos, how-to demos and explanations are the kind of thing Meta says it wants to reward.
For the guest room guide, that might mean a photo of a practical layout with a caption explaining the space it saves, a second post on storage, and a third showing how the room works between visits. Each one gives a reader something useful on its own and points to the full article for the rest. Make sure the post matches what's in the article. If someone clicks for a comparison, the comparison should be easy to find.
Then decide when the subject is most relevant, how often it should come back, and what response would justify another round. Let the results tell you when to share it again. Don't just put the link on a loop.
Compare the additional revenue you can reasonably attribute to the new work against what the new work cost, and keep the article's lifetime numbers alongside it. You need both to make a good call next time.
Take an illustrative example (not a forecast). An article has earned $500 against $300 in original production costs. You spend another $100 updating and recirculating it, and you can identify $250 in additional revenue associated with that activity.
The first number tells you whether this round of work was worth it. The second tells you what the article is worth to your business overall, so keep track of both.
Be careful about what you credit to the update. Some readers would have come back anyway. Ad demand rises on its own in Q4, as Mediavine's seasonality data shows. Ad rates and visitor mix change. A better month doesn't prove the edit caused the whole improvement.
So record a baseline, the changes you made and the date of each promotion. Use UTM campaign parameters in Google Analytics to identify the new traffic, and follow Google's own advice to keep them consistent. Always set utm_source, utm_medium and utm_campaign, and stick to lowercase, because Google Analytics treats "Meta" and "meta" as different values. If your ad reports can't separate out the revenue from one campaign, make a note of that instead of pretending you know.
Then weigh the result against the time involved. A short, useful update is a different investment from a full rebuild, even if both articles performed well historically. Pick the one that makes more sense for the time and money you have left this year.
Short, and easy to explain. Every article on it should have proven demand, a worthwhile earning history, a specific improvement to make and a practical route back to readers.
If only eight articles meet that standard, work on those eight. Give each one a clear assignment, an owner, a budget and a distribution plan. The rest can wait until there's a better reason to invest in them. I'd much rather see a publisher do eight updates properly than thirty cosmetic ones.
Some of the preparation can be automated. n8n, for example, has documented support for scheduled workflows and extracting data from CSV and spreadsheet files, which you can use to pull exports on a schedule and join article URLs across reports. The matching still needs checking, and a person still needs to read the pages and make the call. That's where publishing automation is worth it. Your team spends less time building reports and more time deciding what to do with them.
If you'd like help with that call, look at Publisher Consulting and talk with us about what your team needs. And if you want to know how AI engines currently see your site before you decide what to refresh, the GEO Authority System tests exactly that and gives you a plan, most important fixes first.
A content refresh is a real update to an existing article that has already shown it can attract readers and earn revenue. A real refresh adds something a returning reader would notice, such as corrected facts, new examples, original photos or a clearer explanation, and is followed by a new round of distribution. Changing the date without changing the content isn't a refresh, and Google's guidance specifically warns against it.
Use at least 12 months of data and go further back where your tools allow. Google Search Console keeps 16 months of performance data, which is enough for a year-over-year comparison of the current season, and Mediavine's page-level data goes back to December 22, 2019. A longer window separates steady demand and recurring seasonal interest from one-off spikes.
Not automatically. Read RPM alongside total revenue, traffic volume and the reporting conditions, because a high rate on a small sample can be misleading. Mediavine advises against sorting pages by RPM from highest to lowest and suggests that pages with both high CPM and high traffic are the most valuable.
Use an update to improve the specific answers and sources in your article. Ahrefs found that AI assistants tend to cite newer pages than Google's organic results do, and an Advanced Web Ranking study found that cited passages more often showed a recent date, named their subject clearly and contained specific claims. Those observations don't establish that refreshing an existing page causes more citations. The 2024 GEO paper found that adding sources and statistics raised visibility in AI answers by up to 40%.
That's useful evidence. Review what interested the audience, whether the posts brought website visitors, and what those visits earned. Then judge whether the subject is still relevant and create fresh, original posts around the improved article, since Meta deprioritizes duplicated or lightly edited content.
Let the analysis decide. If eight articles have proven demand, a solid earning history and a specific improvement to make, update those eight. There's no reason to add weaker candidates just to fill the calendar, and it's better to stick to the editing and promotion you can actually pay for.
Compare the additional revenue you can reasonably attribute to the update with what the update cost, and track the article's lifetime revenue and costs separately. Use consistent UTM tags for each promotion, record a baseline before you change anything, and account for seasonality and changing ad rates when you interpret the result.
See more from Publisher in a Box in your Google results
Facebook Monetization Suite
The publishing infrastructure behind 300M+ followers, handed over and calibrated to your page. Seven deliverables, one purchase. $499.
Get the Suite →Free newsletter
The publishing playbook, every weekday
Strategies, case studies, and algorithm updates from the team behind 300M+ followers.