---
title: "Reader feedback and site search"
description: "Two first party signals that are unusually direct: what readers tell you, and what they went looking for."
updated: "2026-08-18"
verifiedAgainst:
  - apps/web/app/api/public/feedback
  - packages/shared/src/site-search-analytics.ts
---

Most analytics is inference. These two are readers telling you directly.

## Reader feedback

Readers can leave feedback on a published post. It arrives in the workspace
attached to the article it is about.

It is rate limited per reader, so the mechanism cannot be used to flood you.

<Callout variant="tip">
  Feedback volume is low by nature and worth reading every item of. One reader
  saying an article is wrong about your own pricing is worth more than a week of
  impressions data.
</Callout>

## On site search

Every search performed on your blog is recorded, with daily rollups. This is the
most underrated dataset in the product: it is your audience, in their own words,
telling you what they expected you to have written.

Two patterns worth acting on:

- **Searches with no results.** Each is an article that does not exist.
- **Searches with results but no click.** The article exists and its title does
  not look like the answer.

Both convert directly into Topics. See [Topics](/content/topics).

## Homepage engagement

The blog's front page is measured separately: what readers do there and which
links they follow. It tells you whether your featured selection is the right one.