Skip to content
Ivan Petrovic

Process

A short plan that goes live beats a long audit that does not.

I do not produce hundred-page SEO audits full of issues nobody will implement. Every crawler will generate hundreds of warnings on any site worth auditing, and almost none of them are the reason it is underperforming.

The work is to find which ones are, say why, and put them in an order someone can act on. These five stages are how that happens, and each one ends with something you hold rather than something you were told.

  1. 01

    Understand

    Week 1

    The business, the website, the market and what success actually means.

    Before any analysis, I need to know what the business sells, who it sells to, what a qualified enquiry is worth, and which pages are supposed to produce them. SEO recommendations made without that context tend to optimise the wrong pages very thoroughly.

    What you hold at the end

    A written statement of what the site is for, which pages are meant to earn money, and what we will call success.

  2. 02

    Diagnose

    Weeks 1–3

    Technical health, search visibility, competitors and where the opportunity sits.

    This is the measurement phase: crawl and log data, Search Console at page level, how competitors actually cover the category, and what the search results for each target look like today. Findings are verified against the live page rather than accepted from a tool's summary field.

    What you hold at the end

    A measured picture of the site: what is indexed, what is crawled, what ranks, and where the structure contradicts the intent.

  3. 03

    Prioritise

    Week 3

    Separate the issues that move search visibility from the ones that only fill a report.

    Every crawler produces hundreds of warnings. Very few of them are the reason a site is not ranking. Prioritisation means naming the constraint, ordering the work by what it unblocks and what it is worth, and being explicit about what is deliberately not being done.

    What you hold at the end

    A ranked list, each item with the evidence behind it, the owner, the dependency, and what it unblocks. Plus what is deliberately not being done.

  4. 04

    Execute

    As agreed, from week 4

    Implementation by me, by your team, or both, as agreed.

    How the work gets done is agreed up front. I can make the changes myself in the CMS, the markup, the structured data and the content, give your developers and writers precise instructions, or split the work between us. Either way every change has an owner and is checked on the live site, because a recommendation that never goes live changes nothing.

    What you hold at the end

    Changes live on the site, whether implemented by me or by your team, each verified on the live page before and after release.

  5. 05

    Measure

    From week 1, reported monthly

    Rankings, organic visibility, AI citations, conversions and technical health.

    Measurement is set up before the work starts, so the effect of a change is attributable rather than argued about. That means tracking the searches that matter, the pages meant to earn them, the technical health of the site, and increasingly whether the brand is cited in AI answers at all.

    What you hold at the end

    A baseline taken before anything changed, and reporting against it on the pages and searches that were the point.

Principles

What sits behind the method.

These four positions decide what makes it into a recommendation and what gets left out on purpose.

SEO is not a checklist

Technical health, content, authority and user experience are not separate workstreams that can be scored independently. A site with excellent content and a broken indexation setup has one problem, not two. The job is to see how the parts interact and to fix the one holding the others back.

Find the constraint

The most important problem on a website is rarely the longest list of warnings in a crawler. It is usually one structural decision: which pages exist, how they are linked, what the site tells search engines to keep. Identify that, and most of the remaining list stops mattering.

Prioritise for impact

Resources are finite and most SEO recommendations are never implemented. I would rather deliver five changes that go live than fifty that sit in a spreadsheet. Every recommendation carries the evidence behind it, so it can be argued with rather than taken on trust.

Build for search beyond ten blue links

AI Overviews, ChatGPT and Perplexity assemble answers from sources, and they select those sources on clarity, structure and authority. That is not a new discipline. It is the same fundamentals, applied to pages that have to be quotable as well as rankable.

Tooling

What I use to get the evidence.

Listed because clients reasonably ask, not because the tools do the thinking. A crawler tells you what it managed to check; the page tells you what is true.

Analytics and search data

  • Google Search Console
  • Google Analytics 4
  • Google Tag Manager
  • Looker Studio

Data and research

  • DataForSEO APIs
  • Search Console and Analytics APIs
  • Python data pipelines
  • AI answer sampling across assistants

SEO platforms

  • Ahrefs
  • Semrush
  • Screaming Frog
  • PageSpeed Insights

Platforms and code

  • WordPress
  • Webflow
  • Laravel
  • HTML
  • CSS
  • JavaScript

Technical

  • JSON-LD structured data
  • Core Web Vitals
  • Cloudflare
  • Server log analysis

To see the method applied end to end, read the case studies or the detail of each service.

Want this applied to your site?

Start by telling me the symptom. The first stage is understanding the business and the site, and that conversation costs nothing.

Discuss Your SEO Project

Or email directly

hello@ivanseo.com