Back to all articles
Project Notes

Building MeowTested: Running My Own Site Like a Client Engagement

The full build story of MeowTested — an Astro affiliate content site with a real tester network, a Search Console-driven content loop, and one big business-model pivot.

Conor Gotzens
3 min read
Educational content only. Examples and performance figures are illustrative and not guarantees. Results depend on your business context, implementation quality, and market conditions.

MeowTested is a cat-product testing site I founded, designed, built, and run. It's also the project where I get to be my own most demanding client — every process I offer people professionally gets battle-tested here first, with my own money on the line.

This is the build story: the stack, the systems, and the one big pivot.

The premise

Most pet-product "review" sites are affiliate pages written by people who never touched the product. MeowTested's bet is the opposite: recruit a real panel of cat owners (60+ volunteers and growing), score products on a disclosed research framework, publish the failures alongside the winners, and only recommend gear that earns it.

One brand principle shapes everything, and it's stricter than you'd expect: the hunt must resolve. MeowTested will never sell or recommend laser toys, at any margin, because a cat can't ever catch a laser dot — and unresolvable hunting drive causes real frustration in cats. When a welfare principle can veto a revenue line, the rest of your recommendations get easier to trust. (Regular readers of this site have already met this principle: it's why the cats down there chase a catchable toy mouse.)

The stack

The architecture is deliberately boring, which is a compliment:

  • Astro static build on Netlify. Every page ships as HTML. Fast, nearly free to host, easy to iterate.
  • Netlify Functions + Resend for the application funnel. A cat owner applies through a form; a serverless function mints a tester ID, calculates points, adds the contact to the audience with structured properties, and sends a branded confirmation email. Referral codes and point balances are all contact properties — the "database" is the email platform.
  • A JSON content engine. Blog posts are structured data — typed sections (comparison tables, FAQ blocks, a snippet-optimized TL;DR box) rendered at build time. Adding a post is appending JSON, not designing a page.
  • A Search Console data pipeline. One command pulls query and page performance into the repo. What gets written next is decided by what real people already almost-find us for — not by a keyword tool's fantasy volume numbers.

That last one is the whole playbook in miniature: ship fast pages, publish content aimed at real queries, measure with first-party data, reinvest in what works.

The pivot

MeowTested didn't start as an affiliate site. The first version was dropship e-commerce — a Shopify checkout, webhooks awarding loyalty points on purchases, the works.

It was the wrong model. Dropship commodities diluted the positioning (why would a testing brand sell generic products?), margins were thin, and every operational hour went to fulfillment instead of the thing that makes the brand defensible: the testing network and the content.

So I killed it. Products moved to Amazon affiliate links, the dropship catalog was retired with 301s into the relevant category roundups, and the Shopify integration is being wound down. Revenue is now affiliate commissions on recommendations — which aligns the incentive exactly where it should be: MeowTested only earns when a recommendation is good enough to act on.

Killing your own launch model six months in isn't fun to write about, but it's the most useful thing in this post. The data said the model was wrong; the sunk cost said keep going. Side projects are where you learn to side with the data.

What's next

The tester panel keeps growing, and organized product-testing rounds through the network are the roadmap's next phase — that's the moat, because cat-to-cat variation is exactly what single-reviewer sites can't capture. Until those rounds run, recommendations stay clearly framed as what they are: structured research and scoring, not in-home lab results. The honesty is the strategy.

If you're building a content site of your own, the transferable lesson isn't the stack — it's the loop. Publish, measure against real search data, cut what the data kills, and double down on what it feeds. It works the same whether the site sells cat fountains or commercial construction.