TL;DR

  • I'm building Runometry, a price comparison site for sports gear, solo, with Angular/AnalogJS on Cloudflare.
  • After 53 days: 276,000+ live offers, 800,000+ prices recorded, 2,800+ models.
  • The KPIs that pay: 900+ clicks to retailers (estimated human clicks), 1 sale, €2.75 in pending commission.
  • Main reason: Google hasn't indexed the site yet. Main signal: 9 people set 13 price and stock alerts, in a market where people buy less than one pair a year.

Context

I'm a freelance full-stack TypeScript developer based in Montpellier, mostly Angular. Runometry started with running shoes and now also covers road, gravel and mountain bikes, GPS watches, headlamps, football boots, headphones and heart rate sensors.

The business model is affiliate. You click through to a retailer, you buy, I get a commission. So the only numbers that really matter are clicks and sales.

The vanity numbers

  • 53 days since the first commit
  • 2,400+ commits
  • 276,000+ live offers
  • 800,000+ prices recorded
  • 2,800+ models from 100+ brands
  • 30+ retailers, 15 markets, 10 languages
  • About 43,000 pages in the sitemap

They look great in a screenshot. None of them pays anything.

The reality check

  • 900+ clicks sent to retailers (estimated human clicks)
  • 1 sale: a €110 order
  • €2.75 commission, still pending
  • 0 validated sales

The stack

  • Angular + AnalogJS for the site
  • Cloudflare, with D1 as the database (my stats script queries it through wrangler)

How the data is shaped

The same shoe in the same colorway doesn't cost the same depending on the retailer and the country it ships to. So the catalogue is modelled like this:

model 2,800+ published └─ variant = colorway × retailer × market 316,000+ └─ offer 276,000+ active └─ price history 800,000+ rows 

An offer counts as active when it's in stock, not withdrawn, and attached to a model.

Scraping and aggregation

[CODE: how a retailer feed is fetched, normalised and matched to a model]

Prices come from 30+ retailers, and outgoing clicks go through 7 affiliate networks. Three things the data taught me:

1. Offer count is a vanity metric. One marketplace alone carries 149,000+ of the 276,000+ active offers. Without it, the headline number is cut in half.

2. A price drop is a bug until proven otherwise. My stats report lists the biggest drops between two consecutive readings of the same offer, some around -69%. I treat them as candidates, never as deals. A broken feed looks exactly like a great discount.

3. Check your tracking before you celebrate. While writing this post, my stats script reported 32,000+ clicks to retailers. My admin says 900+, filtered as human clicks. A few bots may still get through, so I treat it as an estimate. The most flattering number in a report is the one to check first.

SEO and indexing with lots of generated pages

  • 60,000+ pages generated
  • About 43,000 marked indexable and listed in the sitemap
  • 900+ commits in the last 30 days
  • Indexed by Google: not yet

What I've learned so far:

  • Generating pages is the cheap part. Getting Google to crawl, trust and index them is the real work.
  • The sitemap is what you submit, not what Google keeps. My database knows how many pages I list. Only Search Console knows how many Google indexed.
  • Changing a lot, early, has a cost. My read is that constant structural changes gave Google a moving target.
  • Competition is up. A site that isn't indexed yet starts further behind.

The signal

9 people set 13 price and stock alerts. And the buying cycle is long: in France, 9 million pairs of running shoes were sold in 2024 for 12.4 million runners (Observatoire du running, Union sport & cycle). Less than one pair per runner per year.

These aren't lost visitors. They're sales on hold.

What's next

Stop moving the structure, get product pages indexed, then see whether organic clicks turn into sales.

Question for you

If you've shipped a site with tens of thousands of generated pages: how long did Google take to index it, and what actually made the difference?