Switching Marketing Agency?
Start Here Before You Touch the Backlinks

Assume Nothing About What the Last Agency Did

The instinct when you take over from another agency is to get moving quickly — show results, justify the switch, build momentum. That's understandable. But the worst thing you can do is skip the audit phase and start building on a foundation you haven't checked.

In our experience taking over accounts, the previous agency's work splits roughly into three categories: they did a reasonable job and it just wasn't enough; they did some good work early and then got lazy; or they quietly made a mess that nobody noticed because rankings held up for a while on historical authority. That third category is the dangerous one, because the consequences are often delayed.

The client assumes things are fine because nothing catastrophic has happened yet. The new agency assumes things are fine because the client isn't screaming. And quietly, pages that should be performing have been degrading for months.

Read the Pages Before You Optimise Them

This sounds so obvious it barely feels worth writing. And yet it's the step that gets skipped more often than any other. People run crawls, pull keyword reports, check domain authority — and never actually sit down and read what's on the pages they're about to optimise.

Read the pages. Not scan them — read them, the way a visitor would. Check the key service pages, the main blog posts, the location pages if there are any. Ask yourself honestly: does this copy make sense? Does it read naturally? Would someone who landed here think it was a credible business?

You'll find things you wouldn't find in any tool:

• Copy that trails off mid-paragraph because a content writer ran out of ideas (or word count)
• Service descriptions that describe a competitor's process rather than this client's
• Testimonials or case study references that are clearly fabricated
• Stats and claims that are years out of date ("as of 2019...")
• Thin pages with 150 words dressed up as articles
• And the one we'll cover in the next section: formatting and encoding issues that make the page look broken

The goal isn't to judge — it's to see what you're actually working with before you build a strategy on top of it.

The Encoding and Formatting Gremlins

This is one of the most common and underappreciated issues in inherited SEO work: content that was originally written in one system and then copy-pasted into another, carrying invisible formatting with it.

The symptoms are familiar once you know to look for them. Em-dashes that become â€". Apostrophes that turn into ’. Bullet points that render as squares. Non-breaking spaces that push words off-alignment. Smart quotes turned into question marks or boxes. These are character encoding problems — typically UTF-8 content being interpreted as ISO-8859-1, or content pasted from Microsoft Word carrying Windows-1252 encoding into a UTF-8 CMS.

Why does this matter for SEO? Because Google reads the page the same way a browser does — it sees the garbled characters, not the intended text. A sentence that was supposed to read clearly now contains noise. The readability score drops. The semantic context the page was meant to communicate gets diluted. And a visitor who lands on a page with visibly broken characters leaves immediately, sending a bounce signal that further suppresses the page.

The fix is usually quick once you've identified it — find and replace the garbled strings in the CMS, or re-paste the content through a plain-text intermediary like Notepad. But you have to go looking for it. Nobody will tell you it's there.

Dig Into the Analytics History

Before you look at where the traffic is now, look at where it used to be. Pull up Google Analytics, go back as far as the data goes, and look at organic traffic trends by landing page. You're specifically looking for pages that were once bringing in meaningful traffic and have since dropped.

This is different from looking at the current state of the site. Current traffic tells you what's working now. Historical traffic tells you what was working before something changed — and identifying that "something" is often worth more than any amount of new content creation.

What you're looking for in the history:

• A sudden cliff drop: traffic to a specific page falls off sharply at a particular point in time. This usually means a technical change — a redirect was added or removed, the page was accidentally noindexed, a canonical tag was pointed somewhere wrong, or the URL changed without a proper 301.
• A slow fade: traffic declines gradually over 6–18 months. This usually means the page went stale — content didn't keep up with how the search intent evolved, or competitors published better answers and the page started losing position incrementally.
• A post-update collapse: traffic drops that correlate with Google core update dates (easily cross-referenced with Moz's Google algorithm change history). These need a different kind of fix than technical issues.

Find the Content Kings That Dropped Off

Every site has them — pages that at one point drove a disproportionate share of organic traffic. Sometimes they're blog posts that went semi-viral. Sometimes they're service pages that ranked for a high-volume term before a site redesign shuffled the URL structure. Sometimes they're resource pages that dominated a niche query for years until they stopped being updated.

In Google Analytics, filter your organic landing pages by date range: set the comparison window to look at performance from two or three years ago versus now. Sort by the biggest declines in sessions. The pages at the top of that list are your priority.

For each one, before you do anything else, ask: why did this page used to work?

• Was it ranking for a specific keyword it's no longer targeting clearly in the H1 or title?
• Did the page get shorter or have sections removed during a redesign?
• Are there broken internal links pointing to it that were never fixed?
• Did the meta title get changed to something less clickable?
• Is the content still accurate, or has the industry moved on and left the page behind?

In many cases, the fix is genuinely simple. Re-add the section that was accidentally removed. Restore the H1 to include the primary keyword. Fix the canonical that's pointing to the wrong URL. Update a statistic that's four years out of date and add two paragraphs covering what's changed since then.

The reason this works so well as a first step is that these pages already have age and historical authority behind them. Google has seen them before, indexed them before, and ranked them before. You're not building from scratch — you're reactivating something that was already trusted. In most cases, even modest improvements produce ranking movement within a few weeks.

Meta Description Testing: Theory First, Then Data

Meta descriptions don't directly affect rankings — Google has been clear about that for years. But they directly affect click-through rate, and click-through rate is one of the most underutilised levers in SEO. If you're ranking fifth and your meta description convinces more people to click than the pages ranked above you, you'll eventually rank higher. And even if you don't, the traffic uplift from improving a 2% CTR to 4% on a high-impression page compounds significantly over time.

The mistake people make is treating meta descriptions as a one-time setup task. You write something reasonable during the initial optimisation, publish it, and never look at it again. Meanwhile, Google Search Console is quietly accumulating impression and click data that tells you exactly how persuasive that description is to real users — and almost nobody goes back to act on it.

The right approach:

1. Form a theory first. Before you write the description, think about the intent of the person searching for that term. Are they comparing options? Looking for a specific answer? Ready to buy, or still in research mode? Your description should speak to that intent directly — not describe the page, but answer the implicit question the searcher is asking.
2. Write two or three variants. Different angles: one that leads with the outcome, one that leads with the process, one that includes a specific number or fact. These become your test candidates.
3. Publish the one you think is strongest and note the date. Check Search Console after 4–6 weeks. Look at CTR for that specific page and compare it to what it was before. If it improved, keep it. If it didn't, try the next variant.
4. Don't rotate too fast. The data takes time to accumulate, especially on lower-traffic pages. A change that looks like a failure after two weeks might look like a win after eight. Give each variant a fair run before concluding anything.

This process isn't perfectly scientific — there are too many variables (seasonal shifts, position fluctuations, SERP feature changes) to call it a controlled experiment. But directionally, if your CTR goes up after a change and holds, something is working. If it goes down, revert and try something else. The discipline of testing is more valuable than any individual result.

Use Google Ads CTR to Validate Your Copy Before It Goes Organic

Waiting 4–6 weeks for Search Console to give you CTR feedback on a meta description change is workable, but there's a faster way to test copy — and if the client is running Google Ads, you probably have access to it already.

RSA descriptions in Google Ads get real click-through rate data at the asset level within days of going live. If you're trying to decide between three different angles for a meta description, you can put all three into an RSA description slot, let Google rotate them against real search traffic, and look at asset-level CTR after a few weeks. The description with the highest CTR is almost always the right one to deploy as the organic meta description too.

The audiences aren't identical — paid search and organic search attract slightly different intent states — but they're similar enough that directional signals transfer. If description variant A beats variant B by a meaningful margin in Google Ads, it's a safe bet to run A in the organic SERP and see if the pattern holds.

This approach has an added benefit: it forces the SEO and paid search work to stay connected. When the same copy is being tested across both channels, the two sets of data reinforce each other, and you're not building separate messaging strategies for organic and paid that slowly drift apart over time.

If you want to go deeper on this, we wrote a dedicated guide on using RSA descriptions as a copy testing lab for SEO meta tags — including the exact slot setup and how to read the asset-level CTR report.

The Right Order of Operations

To summarise: when you take over an account, the instinct to jump to backlinks, keyword research, and content creation is understandable, but it's the wrong sequence. The correct order is:

The reason this matters practically is that steps 1–4 typically produce visible results within the first 60–90 days. That's important in an agency handover context — clients who've just switched need to see early evidence that the change was worth it. Fixing a broken page that used to rank well and watching it recover is a concrete, explainable win. It builds trust that the more patient work (new content, backlinks) is worth waiting for.

None of this is complicated. That's the point. It's the kind of work that gets skipped because it feels too simple to be worth doing when there are more interesting things to optimise. The previous agency skipped it. Don't make the same mistake.