Tiny PR. +8 / -0. But it closes an instrumentation gap I've been meaning to close.
What changed
Four landing pages, two lines each: the Vercel Web Analytics script tag. No npm dependency, no new library, no build step — it's a <script src="/_vercel/insights/script.js" defer></script> added to the head of index.html, managers.html, owners.html, and investors.html.
Why this instead of (or in addition to) PostHog
These two tools measure different things and I want both:
- PostHog answers behavioral questions: which CTAs get clicked, what the funnel looks like from landing-page arrival to a demo request, where visitors drop off, what their session path was. It's the product analytics layer.
- Vercel Web Analytics answers performance questions. It's Web Vitals out of the box: Largest Contentful Paint, Cumulative Layout Shift, Interaction to Next Paint, Time to First Byte. I can see immediately if a change to the hero image tanks LCP on mobile, or if a font swap introduced CLS I didn't notice.
These two dashboards cover different failure modes. Behavior-level "nobody's clicking the CTA anymore" is one kind of problem; performance-level "the page loads like it's 2009" is another. I want both alarms wired.
Side effect: closing PR #10
This also supersedes PR #10, which had been sitting open as a "Vercel Agent" experiment that would have wired the same instrumentation a much more elaborate way. PR #10 is closed with a note pointing at this one. Smaller is better when the result is the same.
Test plan
- Merge, let Vercel redeploy.
- Visit each landing page in an incognito window.
- Check the Vercel dashboard → Analytics tab for incoming data. (The free tier has a short delay before first events show up, which is fine.)