Technical SEO
What Google's actually indexing and what your site wants indexed are rarely the same thing.
Most technical SEO starts with a crawl of the site as it's built. I start with the log files. What matters isn't the site's structure on paper, it's what Googlebot is actually doing on it - which URLs it's hitting, what response codes it's getting back, and whether that's changed since the last visit. I want to know what Google is actually indexing versus what the site wants indexed. The gap between the two is where revenue gets lost.
On one engagement, that diagnosis found only 2% of the URLs Google was visiting returned a 200. The rest were errors, redirects and dead ends - and no amount of new content was going to fix that until the crawl itself was clean. That's the job. Diagnosis first, then a plan built from what the data actually shows, not a generic checklist run against every site the same way.
Log files, not just a crawler
A standard crawler tells you what your site looks like. Log files tell you what Google's actually doing with it. My process isolates the URLs returning inconsistent response codes across visits - the ones flickering between 200 and error, or quietly redirecting somewhere they shouldn't - and re-crawls the exact URL list Googlebot hit, rather than crawling the site structure generically. That distinction matters more than it sounds. A generic crawl tells you what a bot would find if it started from your homepage and followed every link. It won't tell you what Googlebot is actually choosing to look at, which is a different and much more useful question. It's slower than running an off-the-shelf audit tool. It's also the only way to see what's actually happening, rather than what should be happening.
Crawl budget and indexation are two different problems
Wasted crawl budget means Google's spending its visits on pages that don't matter. Poor indexation strategy means the pages that do matter aren't getting indexed at all. Treating them as the same problem is how sites end up with both - a crawler burning time on parameter URLs and duplicate pages while the commercial pages sit unindexed. Fixing the budget side is usually mechanical: block the noise, consolidate the duplicates, tidy the sitemap. Fixing the indexation side means asking harder questions about why Google doesn't think a page is worth including in the first place. The fix for one usually isn't the fix for the other, which is why they need diagnosing separately rather than bundled into one generic recommendation.
Core Web Vitals as a programme, not a fix
Speed work doesn't hold if it's a single sprint. I ran a global haircare brand's Core Web Vitals programme for a year, on bi-weekly calls with the dev team, because performance regresses every time a new component ships unless someone's watching for it. A marketing team adds a widget, a new tracking script gets bolted on, an image isn't compressed before it goes live, and the scores drift back down. The only way to hold a CWV score is to keep checking it against every release, not audit it once and move on. Same approach on CWV work for another global haircare platform - it's ongoing, not a one-off.
a retained cybersecurity SaaS client - cleaning the web spam out of the crawl
After clearing web spam out of a retained cybersecurity SaaS client's crawl, 4xx crawl errors dropped 78% and Googlebot's homepage crawl frequency rose 38.7%. That's the pattern with a lot of technical work: the win isn't a new feature, it's stopping Google wasting its attention on junk.
When it's gone further - penalty recovery
Sometimes the technical problem is a manual action. I've removed two manual penalties from Google at National Tyres, one involving more than a million suspicious backlinks. When the first one lifted, traffic doubled overnight. Penalty recovery is rare work most SEOs never do - the process is slow, and it only works if the underlying problem is actually fixed rather than just disclaimed in a reconsideration request. Google wants to see the cause dealt with, not just an apology attached to a spreadsheet, and that means the same diagnostic discipline as any other technical work: find what's actually there, not what you'd like to tell Google is there.
The same faults, on well-known sites
Broken schema. Robots.txt blocks that quietly exclude whole sections. Redirects that misfire and send Google, and customers, somewhere wrong. I find these repeatedly, including on well-known brands' sites where you'd assume someone had already checked. Bigger brand, bigger site, doesn't mean fewer basic faults - often it means more people have touched the site over more years, and nobody owns the whole picture anymore. Nobody had looked at what Google was actually doing on the site. That's the recurring story, and it's why I start there rather than trusting that the fundamentals are already sound just because the brand is well known.
Content that isn't indexed doesn't rank, whatever it says. It's the question a technical SEO consultant is actually for. So the first question on any site I look at isn't "what does the content say", it's "does Google even see it, and does it see it the way you think it does". Get that wrong and everything built on top of it - content, links, conversion work - is standing on nothing.