Ping Slack when Algolia zero-result searches spike

Every weekday at 8am, check my Algolia search health and only ping Slack when something is actually off.

Pull yesterday's total search count and top zero-result searches from Algolia analytics for my index. Also pull the same figures across the trailing 14 days so you have a baseline to compare against.

Compare yesterday's zero-result rate against the trailing 14-day baseline. If yesterday's rate is materially above the baseline (expose the spike threshold as a configurable knob, defaulting to somewhere around 1.5x the baseline rate or +2 percentage points, whichever is larger), or if a brand new zero-result query cracked yesterday's top 10 by search volume, that counts as an anomaly worth flagging. Algolia's own guidance is that a no-results rate above 3-5% is not optimal and teams should target under 2%, so use that as a sanity check on whether the baseline itself is already unhealthy.

When there is an anomaly, draft a concise Slack message to the configured channel (also a knob) that names each offending query, its search count for yesterday, and a suggested fix. Pick one of: add a synonym for a likely misspelling or alternative phrasing, create a query rule to steer the result, or flag it as a content gap the catalog does not cover. Keep the message short and skimmable, do not paste raw analytics tables.

If yesterday's numbers look normal, post nothing at all. The channel should stay quiet on healthy days so people actually pay attention on the days you do speak up. This runs alongside the existing weekly Monday recap, which fires unconditionally, so anything routine belongs there, not here.