Talk to Franz

SEO · 01

External SEO Lead Consulting.

Underwater, two men climb a wooden ladder passing a model house upward while a woman at its foot holds the ladder and wraps green tape around the next rung; every rung above already has the same green band
AI-generated image (OpenAI, via Codex) for Full Stack Optimization

SEO rarely breaks in an audit. It breaks between releases, when nobody owns it. I am that owner from the outside, until your team can run it without me.

You hire me when

  • Nobody owns SEO, or the owner left.
  • Releases ship every week, nobody checks them for search.
  • Your first in-house SEO just started.
  • A new product needs search designed in, not patched on.
  • You are buying a company for its organic traffic.

What you get

  • An hour budget or a regular jour fixe, with a KPI deck from Search Console.
  • Quality control after every release, in writing.
  • Your questions answered with your own data, not with opinions.
  • Prioritised tickets your developers can close.
  • SEO Rookie: I train your new in-house SEO until they run the channel alone.
  • Concept workshops for new products: page types, URLs and markup before the first line of code.
  • SEO due diligence before an acquisition.

Why SEO breaks between releases

Every release changes templates, links, markup or URLs. Most of them are fine. One of them is not, and nobody notices for weeks, because nobody owns search between the releases.

Google is clear about who carries that risk: “Ultimately, you are responsible for the actions of any companies you hire” (Google Search Central, Do you need an SEO?). The same is true for your own releases.

So an external SEO lead is not a monthly report. It is one person who checks every release, answers every question with your data, and keeps the routine running. My goal, as it says on my workshop deck: “Our goal is to make all SEO companies - including our own - redundant.”

What breaks when nobody owns SEO

Missing titles happen, and someone usually notices. These do not get noticed. This is what I found in my ongoing jobs, and what I specify for each:

What breaksWhat I measuredWhat I specify
The sitemap drifts away from the siteA sitemap with under a tenth of the URLs Search Console knewThe sitemap built from the live pages, checked against Search Console every month
Markup breaks without an errorMore than half of one markup type invalidThe Rich Results Test on every template, after every release
Routines stopA routine that ran for a while, then stoppedOne owner, one date, a failed check that blocks the release
Goals nobody can measureA “Top 3 ranking” as the goalClicks from Search Console as the goal, brand and non-brand apart
Fixes marked done, not testedAudit fixes re-checked four times in half a year until almost all were really doneEvery task re-tested on the live page
Questions answered by opinionNo AI Overview on any of over a thousand queriesEvery question answered from your own data, in writing

The sitemap drifts away from the site

On one site, the sitemap listed under a tenth of the URLs that Search Console showed for the site.

Google on what a sitemap is for: “A sitemap tells search engines which pages and files you think are important in your site” (Google Search Central, Sitemaps overview). A sitemap with a tenth of the site tells Google the rest is not important.

What I specify: the sitemap is generated from the pages that are live, never kept by hand. Once a month, its URL count is compared with the pages Search Console reports.

Markup breaks without an error

Structured data does not throw an error when it breaks. The page still looks the same. On one site I check, more than half of the items of one markup type were invalid.

Google: “Items that are missing required properties are not eligible for rich results.” (Google Search Central, Structured data policies)

On another site, job ads carried markup without a location Google could read. The details are on the technical SEO page.

What I specify: each template with structured data goes through the Rich Results Test after every release. The fields come from data the site already has.

Routines stop

A team agrees on a routine. In one 2026 call I heard how it usually ends: they did it for a while, then they stopped again.

That is the normal case, not a failure of the people. As I wrote in a 2026 workshop answer: “The pattern is always the same: not missing knowledge, but missing feedback.”

What I specify: one owner and one date per routine. A check that fails turns the build red, so the routine cannot be skipped quietly.

Goals nobody can measure

In one goal-setting talk, the goal on the table was a “Top 3 ranking”. A ranking is one query, one place, one day. I pushed back and asked for goals the team can measure.

What I specify: clicks from Search Console as the main number, brand and non-brand apart, compared year over year.

Fixes marked done, not tested

A task list shows what someone did. It does not show what Google sees. In one 2020 job, I re-ran the SEO tests four times in about half a year and re-checked the performance audit, until it said “95 percent done”.

What I specify: a task counts as done when it passes on the live page. Every open task gets re-tested after the next release.

Questions answered by opinion

A team sends me its questions by email. Each one gets a written answer with data.

One question was about AI Overviews. Of over a thousand queries I checked, none showed an AI Overview. That answered the question better than any article about AI search.

What I specify: every question gets a written answer, with the data it rests on and the source of every number. If a number cannot be traced to its source, it does not go into the answer.

How I work as your external SEO lead

  1. A budget, not a retainer. An hour budget per year, or a fixed number of jour-fixe calls. One hour budget has been renewed every year for several years. A retainer only if you ask.
  2. Every release checked. The three SEO tests on each page type, before and after the release: PageSpeed Insights, the page with JavaScript off, and Google’s live test. In one 2020 job, four times in about half a year.
  3. Questions in writing, answers in writing. Between the calls, your team writes its questions down. In one 2026 jour-fixe package, the written answers covered crawl rate versus server response time, and internal links.
  4. Hypotheses, not opinions. In one multi-year job, about two dozen client questions got about two dozen separate decks, and about two dozen hypotheses were tested. The reports come out of a reproducible build: the same data, the same report.
  5. The months after a relaunch. After one relaunch, a bi-weekly jour fixe with a KPI deck: a briefing per page type with its robots status, segmented sitemaps, and the redirects in Google’s crawl stats.
  6. Your own person, trained. From 2015 to 2017, I onboarded the first in-house SEO at several companies, two to three months each. My book says why: “SEO must always be a shared responsibility.” The in-house SEO training is the program for it.

This site: every release checked, on every push

I hold my own site to the same rule. Since 12 November 2026, every push to the main branch of fullstackoptimization.com runs one routine in GitHub Actions. A red step means nothing goes live.

  1. 96 unit tests in 15 test files, and Astro’s type check with 0 errors.
  2. The build, with 8 checks after it: every internal link resolves (about 4.7k of them), every lead form has its slider, no page is noindex, every canonical points to the www host, and more.
  3. The deploy to Cloudflare Pages.
  4. The launch check against the live site: every row of the relaunch URL map, fetched without following redirects, plus title, h1, canonical and JSON-LD on every page. On 12 November 2026: 656 checks, 0 failed.

The whole run takes under two minutes. That is the routine I want for your releases too: small, automatic, and it stops the release when it fails.

You can check one part of it yourself: the sitemap index lists every page, and every page in it answers 200.

Questions I get before an external SEO lead job

Do we need a retainer?

Usually not. An hour budget or a fixed number of jour-fixe calls is enough for most teams. Unused calls lapse, so nobody pays for a contract that nobody uses.

What should we expect from an external SEO?

Transparency about every change. Google’s own question to ask an SEO: “Will you share with me all the changes you make to my site, and provide detailed information about your recommendations and the reasoning behind them?” (source) My answer is yes, in writing, with the data.

How often do you look at our data?

After every release, and Search Console every week. Google’s own baseline for site owners is lower: “you might want to check your account around once every month, or when you make changes to the site’s content, to make sure the data is stable” (source). With weekly releases, once a month is too late.

Can you guarantee rankings?

No. Google: “No one can guarantee a #1 ranking on Google.” (source)

When should we bring you in?

As early as possible. Google: “a great time to hire is when you’re considering a site redesign (the earlier, the better)” (source). For a relaunch, see relaunch SEO.

Will we depend on you?

Not for long, if it works. The job ends when your own person runs the routine and the release gate without me.

Sources for the Google quotes on this page, read on 2026-11-12: Do you need an SEO?, Get started with Search Console, Sitemaps overview and Structured data policies, Google Search Central. The client numbers come from my own reports, jour fixes and handovers from 2011 to 2026.

From real jobs

  • A written report: every emailed question answered with data.
  • 2025 to 2026: a jour-fixe package with written answers between the calls.
  • For several years: an hour budget, renewed every year, with quality control after each release.
  • A multi-year retainer: about two dozen questions, each answered in its own deck, and about two dozen hypotheses tested.
  • A bi-weekly jour fixe with a KPI deck in the months after a relaunch.
  • 2020: the SEO tests re-run four times in about half a year, audit fixes tracked until almost all were done.
  • 2015 to 2017: the first in-house SEO onboarded at several companies.
  • One client for six years.