How We Actually Do SEO (Not How Most Agencies Do It.

Most people asking about SEO want to know what happens after they sign. Here is the working method, in the order it happens.

Week one: what we look at before touching anything

The first week is reading, not changing. We open the site and the accounts and work through the same list in the same order, because a problem found at the top can make everything below it irrelevant.

• Search Console: which queries bring impressions the site never converts, and which pages sit indexed but never receive an impression.
• Indexation: what is submitted, what is indexed, and what is missing from one but present in the other. Orphan pages and accidental noindex tags surface here.
• Crawl access: robots.txt, the XML sitemap, and whether anything in the render path is blocked.
• Analytics: whether tracking is installed once or three times, whether filters exist, and whether automated visits are counted alongside people.
• Conversion tracking: what is recorded as a lead, and whether it matches how the business receives enquiries. Call handling belongs here too - who answers, and what happens to a missed call.
• Local presence: the Google Business Profile, categories, service areas, and whether the website address matches the listing.
• Speed and mobile: the rendered page on a mid-range phone. Most visitors arrive on one.
• Current position: which pages already earn traffic, so we know what must not be disturbed. We keep a written list of pages we will not touch without a reason.
• What a customer is worth, and what the last five enquiries actually were.

That last item changes the rest of the work more than anything technical. If those five enquiries came from a term nobody had considered, the plan should say so.

The audit and what comes out of it

The audit is a working document, not a deliverable we present and file. It stays open for the life of the engagement.

What we check

Everything we are licensed to look at: crawl data as Googlebot sees it, index status, internal link counts per page, duplicate content, redirect chains, broken links, structured data, title and heading patterns, image weight, and the repeating templates. We compare the crawl against Search Console, because the two rarely agree.

What we do with it

We sort every finding into three buckets. Blockers stop the site being crawled, indexed or measured properly, and come first because nothing else works while they are present. Costs leak value quietly - thin pages diluting a section, internal links pointing at the wrong page, a form that fails silently on mobile. Opportunities are additions: new pages, or sections that could rank with better structure.

Each finding gets a line in a backlog with four fields: the evidence, the expected business impact, the effort required, and who owns it. That backlog is the audit, written for someone who decides where money goes rather than for another SEO. On the first handover you get a short brief: the blockers, what we are fixing first, and what we are deliberately leaving alone. Not a document listing several hundred issues.

What happens in a normal month

Months are not identical, but they follow the same shape.

The four tracks

Fixes. Blockers and costs from the backlog. We do the work ourselves where the site allows it. Where it does not, you get an instruction specific enough for whoever controls the code: the change and why.

Content. Either improving a page that earns impressions but converts poorly, or adding a page the site is missing. Both start with the query data and the real questions your customers ask, and both are reviewed by a person before publishing.

Measurement. Every month we confirm the tracking still records enquiries correctly. A change to a form, a phone provider or a template breaks measurement quietly, and a broken month is a month you cannot judge.

Baseline check. We re-crawl and re-read the core numbers against where we started.

What we ask you to do

Very little, but it matters. Approve or decline the priority list if something commercial has changed. Give us access to the accounts rather than a monthly export. Tell us when the business changes - a new service, a closed branch, a different enquiry process - because that invalidates work faster than anything technical. And tell us when the phone rings off a fix we made, because that is often the first real signal and does not always show in the data that month.

How we decide what to work on next

Priority is not a feeling. Each backlog item is scored on three things: the size of the likely impact, how confident we are that the impact is real, and the effort between us and it.

Confidence carries the most weight. A large fix we are unsure about is usually worth less than a smaller fix we are certain about, because uncertain work consumes a month and returns a conversation instead of a result. Blockers override the scoring entirely - a broken month cannot be judged.

There is a size limit per month. We would rather finish four things than start eleven.

When something is not working

Sometimes a fix we expected to move enquiries does not. When that happens we say so in the update, in plain language, then do one of three things: check whether the measurement is broken, check whether we were wrong about the intent behind the page, or close the item and stop spending on it.

We do not keep reporting activity and hope nobody checks. Nothing stays on the list indefinitely because it is difficult to admit it failed.

What you get and when

Every client has the same two things throughout, and the frequency does not change when a month is quiet.

Access, not attachments

The prioritised backlog stays open. You can see what is being worked on, what is queued, and what is blocked. When a month is slow on visible output, the backlog explains why.

The monthly update

At least once a month you get a written update: what changed, what moved against the starting baseline, what we learned, and what is next in order. Enquiries and calls lead it.

Once a month we also hold a short call, around thirty minutes. We walk through the priority order and you tell us what has changed commercially, because that regularly reorders the list. It is where the plan gets corrected, not a sales meeting.

You should never be unsure what we are working on, and you should never find out in month six that something was not working.

What follows is the short version of everything above, in order.

Step 1: We find what's stuck.
We don't start with a keyword list. We start with the traffic picture. Is Google crawling your important pages? Is your organic traffic actually human? Are there technical blockers that make every other investment less effective?

Sometimes the answer is simple - a misconfigured robots.txt, a redirect chain, index bloat from thin pages. Sometimes it's deeper - polluted measurement data, disconnected systems, a site that ranks but doesn't convert.

We find the highest-confidence fix and do it first.

Step 2: We assign intent to every page.
Every URL on your site should serve a clear purpose. Some pages exist to inform. Some exist to convert. Some exist to support a cluster. Most sites have pages that serve no clear intent - they were created because "we needed content."

We map each page to its commercial role. Pages that don't have one either get repurposed or removed. This step alone usually reveals why previous SEO didn't work.

Step 3: We fix the architecture before we add more content.
Adding content to a broken site structure amplifies the problem. We fix crawl, indexation, internal links and silos before writing anything new. Most agencies skip this because it's less visible than "we published 10 articles this month." We don't.

Step 4: We write content that serves a search intent and a business outcome.
Not content for the sake of publishing. Every page answers a real question a potential customer is asking, then guides them toward a next step. We don't write for word count. We write for usefulness.

Step 5: We measure against clean data.
If the traffic baseline is polluted with bot visits, we clean it first. Then we measure what actually changed - enquiries, calls, qualified leads - not just rankings and sessions.

---

This process doesn't fit into a package. Every engagement is scoped after reviewing the site, market and problem. That takes longer up front. It produces work that actually fits.

See what this looks like for your site →