Skip to content
LPR Studio
FREN
Contact us

Case studyCustom softwareGDPR

Case study: CompareHub, a fast and honest product comparison site

7 min read

CompareHub is a French-language product comparison site with two sections: tech and AI on one side, beauty on the other. Each page puts products head to head, records their prices and gives a verdict for each type of reader. LPR Studio designed, builds and runs it; it has been live at compare-hub.net since 19 October 2026.

It is not a client project. It is our own site, and a proving ground for technical choices we later recommend to clients.

The constraint we chose: no server at all

Content sites usually run on a CMS: a database, plugins and a server to monitor. For a site that publishes a few pages a day and serves thousands, that is a lot of moving parts for little benefit.

CompareHub does the opposite. Every page is generated at publish time and then served as a plain file from Cloudflare's network. No code runs when a reader opens a comparison.

ConcernTypical CMS siteCompareHub
HostingApplication server and databaseStatic files on a CDN
Cost of a traffic spikeMore servers or a slower siteUnchanged
Security updatesCore, theme and every pluginBuild tools only, nothing runs in production
PublishingEditor in an admin panelA structured data file, checked before release

Content as data, with rules the machine enforces

Each comparison is a structured file: the products, the comparison table, the prices with the date they were checked, the verdict and the FAQ. A generator reads every file before release and rejects anything that would break a page, such as a missing field, an unknown category or a promise of testing.

That last rule matters. A CompareHub page never says "test in progress" or "we will check". It states what the official specifications say and what the recorded prices show, then makes a call. The rule lives in the generator, so nobody has to remember it.

From those files, everything else is built automatically: section and category pages, a sitemap with each page's real last-modified date, the RSS feed, structured data for search engines, the llms.txt and llms-full.txt files that AI answer engines read, and a share image for every page. On 13 November 2026 the site carried more than 160 comparisons, none of which needed a line of code to publish. They cover everyday purchases, all written in French, such as Oral-B vs Sonicare electric toothbrushes, AirPods 4 vs AirPods Pro 3 earbuds, Meta Quest 3 vs Quest 3S VR headsets and Nespresso Vertuo vs Original coffee machines. Each one follows the same template, built from the same kind of file.

Trust you can check

An affiliate-funded site lives on credibility, so we made it verifiable.

  • Disclosed affiliate links. Every commercial link carries the attribute that flags it to search engines, and the methodology page explains how the site earns money.
  • A published score. Each product's score is computed from the data in the comparison, with the formula shown on the site. It is a reading of the spec sheets, not a lab test, and the page says so.
  • No invented authors. Comparisons are signed by CompareHub, not by a made-up writer persona. When a real person signs, they will have a page and profiles anyone can check.

Built for an average phone on a slow connection

Most readers arrive from Google on a phone, so the site is tuned for them. Images are converted to WebP and resized to the width at which they are displayed, with smaller variants of above-the-fold images for mobile. Fonts and images are self-hosted, with no third-party request on load. The content security policy lets the browser run only the site's own scripts, each allowed by its hash, and an automated check confirms after every deployment that the live policy covers every page.

Measuring without tracking

Analytics are off until the reader agrees. The one exception is an anonymous count of page views with no cookie, no on-device storage and no way to follow a reader across sites. Clicks through to merchants are only measured after consent. If you build for European users, our GDPR checklist for business applications covers the same trade-offs.

What it means for your project

You may not need a comparison site, but three ideas carry over.

  1. Run code only where it earns its place. Marketing sites, catalogues and documentation rarely need a server. Static is cheaper, faster and safer.
  2. Put the rules in the tooling. A build that refuses an incomplete page beats a review that will one day miss it. See how we approach AI-assisted development with human review.
  3. Make trust verifiable. Prices, methods and commitments should be something a reader can check, not something they have to believe.

Our other products are documented the same way: the Izi-Devis case study, voice-driven quotes and invoices for tradespeople, and the Tubissime case study, personalised songs on the web, iPhone and Android.

Frequently asked questions

Is CompareHub a client of LPR Studio?

No. CompareHub is a site published and operated by LPR Studio itself, under the same company registration, like Izi-Devis and Tubissime. We designed it, built it, put it online and write its comparisons. That is why we can describe its architecture and trade-offs in detail without revealing anything about a client.

Why build a content site as fully static pages instead of using a CMS?

Because a comparison site is read far more often than it is edited. Rendering every page at publish time removes the server to patch, the database to protect and the cost per visit. Pages load fast everywhere, and a CMS vulnerability cannot expose a site that runs no code when someone visits.

How can an affiliate-funded comparison site stay trustworthy?

By disclosing and showing its work. Every commercial link is marked as such, the scoring formula is published, and no comparison claims a hands-on test that never happened: each one relies on official specifications and prices checked on a stated date. Readers can verify every claim for themselves.