Email and Slack alerts
Who gets told what, and when: verdicts, KPIs that stop firing and shipped-change problems, by email and in Slack.
You shouldn't have to watch a dashboard to know if the needle moved. Needlr tells you when something needs a decision: a verdict, or a KPI that stopped counting.
Needlr tells you when something needs a decision, so you don't have to keep checking. Alerts come by email, and to a Slack channel if you add one.
Think of a smoke alarm
You don't stand in the kitchen all day, watching for smoke. The alarm watches for you, and only beeps when it matters.
Needlr's alerts work the same way. They stay quiet until there's something to act on.
Who gets them
Every Owner and Editor of your account gets every alert, each in their own email. Viewers don't get alerts (roles). The demo project never sends any.
Don't want some of them? Open Team and use Your emails to turn off Results, Alerts or the others, just for you. Slack keeps getting everything.
What you're told, and when
- A verdict. When an impact check reaches its verdict (Improved, Got worse or No clear effect), you get an email within a few minutes, with both rates. An adoption check says Target reached, Below target or Near target. Each check is announced once.
- A split test stopped because it hurt. When a split test your coding agent asked for is clearly hurting the KPI or a guardrail KPI, it stops by itself, and you're told the same way as a verdict: Stopped · it hurt.
- A KPI stopped firing. A KPI that fires every day had almost no hits in the last 24 hours. Usually a release changed the button or the page. Running checks that use it go on hold until it fires again (KPI health).
- A KPI is firing again. One more email when it counts again. Checks that use it carry on.
- A shipped change isn't on your site yet. You pressed Ready for verification, but most visitors still saw the original (more).
- The original is back. After Needlr stopped applying a shipped change, most visitors saw the original again.
- Shipped changes are going back. Your plan ended, so changes shipped through Needlr go back to the original in 30 days. You're told on the day, 7 days before and 1 day before.
No daily noise: if a KPI stops again soon after a “firing again” email, Needlr waits two days before it tells you again.
Plan, usage and account emails are in Every email Needlr sends.
Add a Slack channel
On every account, the same alerts can also go to one Slack channel, for every project in your account.
- 1Open your project and go to the Settings tab. Find Alerts for the whole account.
- 2In Slack, go to Apps › Incoming Webhooks › Add to Slack. Pick the channel for the alerts, and copy the Webhook URL.
- 3Paste it into the box and press Connect. Needlr posts a first message in the channel, to check it works.
- 4Want to check again later? Press Send a test message.
If Slack stops
- Channel gone: Slack says the channel or the webhook doesn't exist any more, so alerts stopped. Paste a new webhook address.
- Disconnect makes Needlr forget the address. Emails keep coming.
Common mistakes
- Waiting for alerts as a Viewer. Only Owners and Editors get them. Ask an Owner to make you an Editor.
- Leaving a “stopped firing” email for later. Every check on that KPI is on hold until it's fixed.
- Pasting the wrong address. It must start with https://hooks.slack.com/services/.
What the video says
You shouldn't have to watch a dashboard to know if the needle moved. Needlr tells you when something needs a decision: a verdict, or a KPI that stopped counting.
When an impact check reaches its verdict, every owner and editor gets an email within minutes: improved, got worse or no clear effect, with both rates.
If a KPI that fires every day goes almost silent for 24 hours, you hear about it. Usually a release moved the button. Checks that use it go on hold.
When it fires again, one more email says so, and those checks carry on measuring. No daily noise: if it stops again, the next alert waits two days.
On every account, add a Slack channel. Paste an incoming webhook in your project's Settings, and the same alerts land where your team already talks.
Shipped changes are watched too: a failed verification, or the original coming back. Stop checking dashboards. Let the needle come to you, and decide.