The Keep-Kill-Combine Content Audit Framework: A Complete Guide

Why most content audits are a waste of time

I've inherited a lot of content audits over the years. They almost always look the same: a massive spreadsheet with hundreds of URLs, a few columns of metrics, and no clear next step for any of them.

The team that produced the audit spent weeks pulling data. They color-coded things. They added a "priority" column. And then nothing happened, because nobody could look at 600 rows and figure out what to actually do on Monday morning.

That's not an audit. That's a census.

A useful content audit ends with decisions, not data. Every URL on your site should come out the other side with a clear verdict and a clear owner. If your audit doesn't produce that, it's just an expensive way to confirm what you already suspected: you have a lot of content, and some of it isn't great.

I built the Keep-Kill-Combine framework because I needed a repeatable system that forced a decision on every piece of content. Not "maybe revisit in Q3." Not "needs further analysis." A decision. Keep it, kill it, or combine it with something else. Three options. No hiding.

The Keep-Kill-Combine framework

The core idea is simple. Every URL on your site gets sorted into one of three buckets:

• Keep: the page is performing, it's relevant, and it deserves to stay as-is (or with minor updates).
• Kill: the page is dead weight. It's not ranking, not converting, not serving any strategic purpose. Remove it or noindex it.
• Combine: the page has some value, but it overlaps with one or more other pages. Merge the best parts into a single, stronger piece.

That's it. Three buckets. The power of the framework isn't in its complexity. It's in the fact that it eliminates the most common audit failure mode: indecision.

I've used this on sites with 50 pages and sites with 4,000 pages. The scale changes. The process doesn't. You pull the data, you score each URL against a set of criteria, and you sort it into a bucket. Then you execute.

Let's break down each bucket.

Keep: what stays and why

A page earns a "Keep" when it's doing real work for your business. That doesn't mean it has to be your top performer. It means it's contributing something measurable and it's not duplicating effort elsewhere on the site.

Here's what I look at when deciding whether something stays:

• Organic traffic: Is Google sending people to this page? Even modest, steady traffic counts. I pull this from Google Search Console, not GA4, because GSC gives me impression data too.
• Conversions or micro-conversions: Does this page generate leads, sign-ups, purchases, or meaningful engagement? Check GA4 for events tied to the URL.
• Backlink profile: Does this page have external links pointing to it? If so, you don't just delete it. You redirect it. I use Ahrefs or Semrush to check this.
• Strategic relevance: Does this page support a topic cluster you care about? Sometimes a page doesn't rank yet but it's essential to your content strategy architecture.
• Freshness: Is the information still accurate? A page can have great metrics and still need a refresh if the content is outdated.

A Keep doesn't mean "never touch again." Most Keeps get a note about minor improvements. Update the stats, refresh the examples, tighten the intro. But the page stays at its current URL and retains its current purpose.

On a healthy site, I'd expect 40 to 60 percent of pages to land in the Keep bucket. If you're seeing less than 30 percent, that's a sign the site has been publishing without a strategy for a while. If you're seeing more than 80 percent, you're probably being too generous with yourself.

Kill: what goes and why

This is where most teams get stuck. Killing content feels wrong. Someone spent time writing it. Maybe a stakeholder approved it. Maybe it was part of a campaign that had a budget attached to it.

None of that matters if the page isn't serving a purpose today.

Here are the signals that tell me a page should be killed:

• Zero or near-zero organic traffic for six months or more. I check Google Search Console for both clicks and impressions. If Google isn't even showing the page in results, that's a strong kill signal.
• No backlinks worth preserving. If Ahrefs shows zero referring domains, there's nothing to redirect.
• Thin content with no unique value. If the page is 200 words of generic advice that exists on a thousand other sites, it's not helping your SEO.
• Outdated information that can't be refreshed. A post about a 2021 algorithm update isn't going to rank again no matter what you do to it.
• Duplicate or near-duplicate of another page that's performing better. If you have two pages targeting the same keyword and one is clearly winning, the loser gets killed (or combined, which we'll get to).
• Orphaned pages with no internal links pointing to them. If your own site doesn't link to a page, you're telling Google it doesn't matter.

"Kill" doesn't always mean hitting the delete button. The execution depends on the situation:

1. 1If the page has backlinks, set up a 301 redirect to the most relevant surviving page.
2. 2If the page has zero backlinks and zero traffic, you can simply remove it or noindex it. I usually remove it entirely to keep the site clean.
3. 3If the page is tied to a campaign that might return, archive it as a draft rather than deleting it.

I worked with one B2B company that had been publishing two blog posts a week for three years. That's over 300 posts. When we ran the audit, 140 of them had zero clicks in the previous six months. Not low traffic. Zero. Those pages weren't just failing to help. They were actively diluting the site's crawl budget and topical authority. We killed 120 of them and redirected the remaining 20 to stronger pages. Organic traffic to the surviving content increased 35 percent in 90 days.

Combine: the most underused move

If Keep and Kill are straightforward, Combine is where the real leverage lives. And it's the step that most content audits completely ignore.

Here's the pattern I see constantly. A company publishes a blog post about "email marketing best practices" in 2022. Then in 2023, they publish "email marketing tips for small businesses." Then in 2024, "how to improve your email marketing." Each post is slightly different, but they're all chasing the same intent and the same keyword cluster. Google doesn't know which one to rank. So it ranks none of them well.

This is keyword cannibalization, and it's one of the most common problems I find during audits. The fix isn't to kill the weaker pages and keep the stronger one. The fix is to combine them into a single, comprehensive piece that's better than any of the originals.

Here's how I run a combine:

1. 1Identify the cluster. Use Google Search Console to find URLs that share impressions for the same queries. Ahrefs' Content Gap tool and Semrush's Cannibalization report are useful here too.
2. 2Pick the winner. Choose the URL with the best combination of existing traffic, backlinks, and topical coverage. This becomes the surviving URL.
3. 3Pull the best content from each page. Go through every page in the cluster and extract anything unique, valuable, or well-written. Don't just keep the winner's content and throw away the rest.
4. 4Rewrite into a single, stronger piece. This is the part people skip. You can't just paste three articles together. You need to restructure, eliminate redundancy, and create something that genuinely covers the topic better than any individual piece did.
5. 5Redirect all losing URLs to the winner with 301 redirects. This passes link equity and prevents 404 errors.
6. 6Update internal links across the site to point to the new consolidated URL.

The combine move is powerful because it's additive. You're not just removing bad content. You're creating something better than what existed before. You're consolidating link equity, reducing cannibalization, and giving Google one clear page to rank instead of three mediocre ones.

On content-heavy sites, the Combine bucket often represents the single biggest opportunity. I've seen cases where combining five thin posts into one comprehensive guide took a keyword from page three to position four in under two months. The content already existed. It was just scattered across URLs that were competing with each other.

Running the audit step by step

Here's the exact process I follow when I run a Keep-Kill-Combine audit. I've refined this over dozens of engagements and it works whether you have 50 pages or 5,000.

Step 1: Crawl the site

I start with Screaming Frog. Crawl the entire site and export the full URL list. This gives me every indexable page along with metadata like word count, response codes, canonical tags, and internal link counts. If the site is large, I'll scope the audit to a specific section (like the blog or the resource center) rather than trying to boil the ocean.

Step 2: Layer in performance data

Next I pull data from three sources and merge it into a single spreadsheet:

• Google Search Console: clicks, impressions, average position, and top queries for each URL. I pull 12 to 16 months of data to account for seasonality.
• GA4: sessions, engaged sessions, conversions, and any custom events tied to specific URLs. This tells me which pages are driving business outcomes, not just traffic.
• Ahrefs or Semrush: referring domains, Domain Rating of linking sites, and keyword rankings. This tells me which pages have earned external authority.

If you want to go deeper, I also pull analytics data on scroll depth and time on page. These engagement signals help distinguish between pages people actually read and pages they bounce from immediately.

Step 3: Score each URL

I score every URL on a simple 1 to 5 scale across four dimensions:

1. 1Traffic (organic clicks over the past 6 months)
2. 2Authority (number and quality of backlinks)
3. 3Conversions (leads, sales, or meaningful engagement events)
4. 4Relevance (does the page align with your current business goals and target topics?)

The first three are data-driven. The fourth requires judgment. A page might have decent traffic but cover a topic you no longer serve. That affects the score.

I sum the four scores for a composite rating between 4 and 20. Pages scoring 14 or above are almost always Keeps. Pages scoring 8 or below are strong Kill candidates. Everything in between needs a closer look, and that middle band is where most Combines live.

Step 4: Identify cannibalization clusters

Before I finalize any decisions, I run a cannibalization check. I export all queries from GSC and look for cases where multiple URLs are getting impressions for the same keyword. I sort by query, then look at which URLs appear. If two or more pages are splitting impressions for the same term, they're candidates for a combine.

Semrush has a Cannibalization report that automates a lot of this. It's not perfect, but it catches the obvious cases quickly. I always verify manually before making a final call.

Step 5: Make the call

This is where I assign every URL to Keep, Kill, or Combine. For Combines, I also note which pages should be merged together and which URL should survive. For Kills, I note whether a redirect is needed. For Keeps, I note whether a refresh is warranted.

The output is a single spreadsheet with one row per URL and three columns that matter: the bucket (K/K/C), the action, and the owner. Everything else is supporting evidence.

How I use AI to accelerate content audits

I'd be dishonest if I told you I still do all of this manually. AI tools have changed the speed of content audits dramatically. They haven't replaced the judgment calls, but they've eliminated hours of grunt work.

Here's where I use AI in the process:

• Content summarization. When I'm auditing a 400-page blog that's been publishing weekly for three years with no audit, I don't read every post word for word. I use AI to generate a one-paragraph summary of each page, including the primary topic, target keyword (if apparent), and any unique data or examples. This cuts the review time from days to hours.
• Cannibalization detection. I feed my GSC query data into a prompt that groups URLs by overlapping keyword intent. It's not as precise as doing it manually, but it catches 80 percent of the clusters in a fraction of the time.
• Content gap identification. After sorting pages into buckets, I use AI to compare my surviving content map against the target keyword list. It highlights topics where I have no coverage, which feeds the editorial calendar that comes after the audit.
• Rewrite assistance for combines. When I'm merging three posts into one, I use AI to generate a first draft that pulls from all three sources. I rewrite heavily from there, but starting from a structured draft instead of a blank page saves time.

What AI can't do: make the Keep/Kill/Combine decision for you. I've tested this extensively. If you give a language model a spreadsheet of metrics and ask it to categorize each URL, it will give you reasonable-sounding answers that fall apart on the edge cases. It doesn't know your business priorities. It doesn't know that your CEO is about to launch a new service line that changes which topics matter. It doesn't know that the page with mediocre traffic is actually your best sales enablement asset because the sales team sends it to every prospect.

AI accelerates the mechanical parts. The strategic judgment is still human work. I'm comfortable with that division of labor.

What happens after the audit

The audit is only useful if you execute on it. I've seen too many audits end up as a Google Sheet that gets opened twice and then forgotten. Here's how I structure the execution phase to make sure that doesn't happen.

Week 1-2: Kill pages

I start with the Kills because they're the easiest to execute and they generate immediate results. Remove or noindex the dead content. Set up redirects for anything that has backlinks. Verify with Screaming Frog that no internal links point to the removed URLs. This phase typically takes one to two weeks depending on the number of pages involved.

Week 2-4: Combines

Combines are the most labor-intensive step. Each one requires rewriting content, setting up redirects, and updating internal links. I prioritize combines by opportunity size. If three pages cannibalizing each other are all on page two for a high-value keyword, that combine happens first. On a typical engagement, I'll plan for two to four combines per week, depending on how extensive the rewrites need to be.

Week 3-6: Keep and refresh

The Keep-and-refresh pages get updated on a rolling basis. I prioritize by a combination of current traffic and estimated improvement potential. A page that's ranking position 6-10 for a valuable keyword and just needs a content refresh is a higher priority than a page ranking position 40 that needs a full rewrite.

Ongoing: Monitor and measure

I track results at 30, 60, and 90 days post-execution. The metrics I watch are organic traffic to surviving pages, keyword rankings for target terms, crawl stats in GSC (are we being crawled more efficiently?), and conversion rates on updated content. I don't hand you a list of problems and walk away. The audit is the beginning of the work, not the end of it.

Real numbers

I'll share a few anonymized results from Keep-Kill-Combine audits I've run in the last two years. These aren't best-case cherry picks. They're representative of what happens when you actually follow through on the execution.

A SaaS company with 380 blog posts. We killed 160, combined 45 into 15, and kept the remaining 175 (90 of which got refreshes). Result: 52 percent increase in organic blog traffic over four months. The average position of their target keywords improved by 6.2 positions.

A B2B services firm with 120 pages across their site. We killed 25, combined 18 into 6, and refreshed 30. Result: organic leads from content pages increased 40 percent in 90 days. Their Google crawl rate doubled, which meant new content was getting indexed in days instead of weeks.

An ecommerce brand with 2,200 product and category pages plus a 200-post blog. We focused the audit on the blog. Killed 80 posts, combined 35 into 12, refreshed 50. Result: the blog went from generating 8 percent of organic revenue to 14 percent in six months. The product pages also saw a lift because we cleaned up internal linking and stopped diluting the site's topical authority with irrelevant blog content.

The pattern across all of these is consistent. Removing and consolidating weak content doesn't hurt your traffic. It concentrates your authority on the pages that deserve it. Search engines reward focus.

The goal of a content audit isn't to have less content. It's to have the right content. Every page on your site should earn its place.

If your site has been publishing consistently for more than a year and you've never run an audit, you almost certainly have pages that are holding back the rest of your content. The Keep-Kill-Combine framework gives you a structured way to find them and fix the problem.

If you want help running an audit like this, take a look at our content strategy services. I'll walk through your site, pull the data, and give you a clear action plan with priorities and timelines.