Why technical always comes first
Businesses routinely commission twenty blog posts for a site that Google is struggling to crawl, that takes seven seconds to become usable on a mid-range Android phone, and whose most valuable pages are buried five clicks from the homepage.
The posts get published. Nothing happens. The conclusion drawn is that SEO does not work.
SEO worked fine. The site was the problem. Technical SEO is not a competing discipline to content, it is the precondition for content being worth paying for. Which is why we will not sell you a content retainer on a site that has not passed a technical audit. It would be taking money for work we know cannot produce a return.
Crawl and indexation
Start with the only question that matters: does Google actually have your pages?
Search Console’s index coverage report is routinely alarming when someone looks at it properly for the first time. Pages excluded by a stray noindex left over from a staging environment. Canonical tags pointing at the wrong URL, so Google consolidates your good page into a worse one. Parameter URLs multiplying a 200-page site into 4,000 near-duplicates and exhausting the crawl budget before the important pages are reached. An entire section blocked in robots.txt by someone, at some point, for a reason nobody remembers.
These are unglamorous, they are extremely common, and they are frequently the single largest constraint on a site’s performance. They are also usually fixable in days.
Core Web Vitals, the field data, not the lab score
There is a specific trap here that catches a lot of people. A developer runs PageSpeed Insights, sees a green score, and declares performance solved. Then Search Console reports that Core Web Vitals are failing.
Both are correct. They are measuring different things. The lab score is a simulation on a specified device. Field data is what actually happened to real people, on the phones they own, on the networks they are using, in the places they live. That is the data Google uses, and if it disagrees with your lab score, your lab score is describing a device your audience does not have.
The usual culprits are consistent: images served far larger than they render, third-party scripts (chat widgets, heat maps, tag managers loading tags nobody remembers adding), fonts that block rendering, and layout that shifts as things load in. Each is fixable. The chat widget nobody uses is the most satisfying to remove.
Schema, done honestly
Structured data helps Google understand what a page is, and can earn you richer presentation in results. It is worth doing.
It is also routinely abused, and the abuse is punished. FAQ schema on a page with no FAQ. Review schema with a five-star aggregate rating and no reviews anywhere on the site. LocalBusiness schema for a location that does not exist. This is not a clever edge; it is a structured data manual action with your name on it.
Our rule is deliberately dull: we only mark up what is genuinely on the page. If you want FAQ rich results, we will write a real FAQ, put it on the page, and then mark it up. In that order.
On-page and on-site optimisation
The fundamentals still decide a great deal, and they are still neglected. Title tags that describe the page rather than stuffing it. One H1 per page, and a heading hierarchy that is a genuine outline rather than a set of font-size choices.
Internal linking is the most underrated item on this list. It is how authority moves around your site and how Google understands which of your pages matters. Most sites link almost everything from the navigation and nowhere else, leaving their most commercially valuable pages orphaned at the bottom of the hierarchy. Fixing internal linking costs nothing but thought, and it regularly moves rankings on its own.
Penalty recovery
First, establish which of two very different things has happened to you.
A manual action means a human at Google reviewed your site and applied a penalty. You will see it in Search Console. There is a defined route back: fix the cause honestly, then submit a reconsideration request that describes what happened, what you have changed, and what will stop it recurring. Requests that minimise or obscure the cause tend to be rejected, and the second attempt is harder than the first.
An algorithmic drop, typically after a core update, is not a penalty and there is nothing to appeal. No message arrives, because no human looked. The only route back is genuinely improving the site’s quality and waiting for the next assessment. It is slow and there is no shortcut, and any agency claiming a special relationship that can accelerate it is lying to you.
If you have been hit, we will tell you which one it is and give you an honest assessment of the odds, including when the honest assessment is that a fresh domain would be faster than recovering this one.
Audits, one-off work and fractional support
Not every site needs a monthly technical retainer, and we will not pretend otherwise to secure recurring revenue. A great many sites need one thorough audit, a focused fix sprint, and then to be left alone.
Ongoing technical support genuinely earns its cost when the site is large, publishing constantly, or carrying an e-commerce catalogue where URLs appear and disappear every week. Outside those cases, a one-time engagement is usually the honest recommendation, and it is the one we will make.
Planning a replatform or a domain change? That is its own discipline with its own failure modes, and it deserves more than a section here.