Before Sirv, before Studio, before I had a useful title for what I do, I built my first commercial website.
It became a profitable business and reached roughly 50,000 unique visitors a day. Then Google changed its algorithm and most of that traffic disappeared.
That sequence taught me more about internet products than any clean success story could have.
I learned development and distribution at the same time
I built it because I saw demand and wanted to see whether I could turn a website into a business.
There was no meaningful boundary between the product and its distribution. I wrote code, structured content, watched search behaviour, fixed pages, improved navigation, and worked out monetization as one continuous job. A development decision could change traffic. A marketing decision could create a technical problem. An operational mistake could become a revenue problem before the day was over.
That is still how I think about products. Engineering, distribution, and the business model are not separate departments in the product’s actual life. They are one system, even when a company draws boxes around them.
Fifty thousand people a day changes the job
At small scale, a broken page is annoying. At roughly 50,000 daily unique visitors, the same mistake is an incident.
The site had to stay available. People had to find what they came for. Search engines had to understand the structure. The publishing flow had to keep moving. Traffic had to pay for the operation rather than merely decorate an analytics chart.
The site did. It became profitable through a traffic-based model. It was not a theoretical startup or a portfolio mock-up. It paid its way and gave me the dangerous confidence that comes from watching a system work for long enough.
The audience was real. My ownership of it wasn’t
Search brought the visitors. I knew that, obviously. What I did not understand deeply enough was the difference between earning distribution and owning it.
The pages ranked because the site was useful, established, and aligned with what people searched for. It was easy to interpret that as a durable asset. But the route between the audience and the product belonged to Google. One algorithm sat between a profitable business and most of its customers.
Then Panda changed that route.
Traffic fell hard. The business economics followed. There was no equally strong direct audience, no alternative distribution engine waiting underneath, and no clever technical fix that could restore the old terms.
Google did not literally press a button labelled “kill Igor’s website.” It changed the system it owned. The result from my side was the same.
What I got wrong
I optimized the machine that was working and underinvested in the one I would need if it stopped.
I should have built stronger direct relationships with returning users. I should have treated search as one channel rather than the foundation. I should have asked what part of the business remained defensible if rankings disappeared.
Most importantly, I confused current leverage with permanent advantage.
That error has aged well. App stores, social networks, marketplaces, model providers, cloud platforms, and search engines all make the same offer: enormous reach on somebody else’s terms. The reach can be worth using. The mistake is building as though the terms are yours.
The lesson survived the website
That experience is why I care about owned distribution. It is why I think product resilience is a commercial concern, not an infrastructure detail. It is why “features are not a moat” does not feel like a new AI-era revelation to me.
AI has made software construction dramatically cheaper. That makes the old lesson more important, not less. If somebody can reproduce your feature set quickly, the durable value has to live somewhere else: relationships, trust, data, operating knowledge, distribution, and the ability to keep serving people after the novelty wears off.
The website disappeared. The combined skill set did not. Development, SEO, monetization, operations, and platform scepticism all travelled forward into the businesses and products that came next.
I would rather the business had kept growing. But as an origin story, a profitable business broken by an algorithm is more useful than a clean win. It taught me early that a product is not alive because it shipped, or even because it makes money. It is alive while the whole system around it can survive a change it did not choose.