For twelve years I worked in industrial gas production. I started as a junior engineer and became plant manager and co-owner. Today I do SEO and AI search visibility for B2B companies.
People often ask how those two jobs connect. The answer is one habit: find the real cause before you fix anything.
On a plant, guessing is expensive
When a line stops, the easy move is to replace the part that looks broken. Sometimes that works. Often the line stops again a week later, because the real cause was somewhere else: a pressure setting, a supplier change, a step someone skipped.
So we learned a simple order:
- Describe the problem with data. Not "the line is slow", but "output dropped on these shifts, from this date".
- List the possible causes. All of them, not only the obvious one.
- Test the most likely cause first. Change one thing, then measure.
- Write down what you found. So the next person doesn't start from zero.
When I built an ISO/IEC 17025 testing lab from scratch, this order was not optional. Every result had to be traceable to a method, a calibration and a record.
SEO has the same trap
Websites have their own version of "replace the part that looks broken". Traffic drops, and the first idea is "we need more content" or "we need more links".
Sometimes that is right. But often the cause is technical and much simpler:
- pages that Google can't crawl, or crawls but doesn't index
- a canonical tag that points to the wrong page
- language versions that compete with each other because hreflang is wrong
- a site move that broke old URLs without redirects
- a page that answers a different question from the one buyers ask
New content won't fix any of these. It just adds more pages to the same problem.
The same order, applied to a website
- Describe the problem with data. Which pages, which queries, which countries, since when? Search Console and analytics usually answer this in an hour.
- List the possible causes. Crawling, indexing, duplicate or competing pages, intent, internal links, page speed, and recent changes on the site.
- Test the most likely cause first. Fix one thing, and note the date.
- Measure, then write it down. What changed in clicks and positions, and what did not.
It sounds slow. In practice it is faster, because you stop paying for fixes that don't touch the problem.
If your site is indexed but not ranking, this guide walks through the checks in order: Website indexed but not ranking? A diagnosis guide.
AI search raises the stakes
AI assistants like ChatGPT and Perplexity now answer many buyer questions directly. They repeat what they find most often and most clearly about a company. If your own pages disagree with each other, or old pages still say old things, the answer can be wrong.
The fix is the same plant habit: one source of truth, no contradictions, and a record of what changed. I built an AI assistant that answers questions from 666 pages of technical manuals. A system like that can only be as good as its source documents. Websites work the same way.
Work with me
I run SEO and AI search projects through my agency, Layerly. If you want a clear diagnosis before you spend money on fixes, that is where to start.