Promovolve is an attempt to get back what magazine advertising had!
// readme
Promovolve is an attempt to get back what magazine advertising had: relevant ads matched to what the reader is actually reading, with no cookies, no user profiles, no cross-site tracking, and no degradation of the reading experience. The page’s content is the only targeting signal — an article about hiking gets ads for hiking gear because of what it is, not because of who is reading it.
Being open source is not incidental: transparency is the product. Publishers and advertisers can read the auction, the pricing, and the pacing logic and verify there is no hidden manipulation — something no closed ad network can offer. For the guided version, read the Promovolve book — 13 short chapters on why the system is built the way it is.
Demo
How it works
- A publisher adds two lines of HTML — a script tag and a
<div>per ad slot. No SDK, no ad server account, no JavaScript to write. - Pages are classified on demand from real traffic: an LLM maps the page’s text into IAB content categories the first time it serves.
- Auctions run periodically per site, not on every…