Glossary

Words to know

Every word Needlr uses, in one plain line each.

In Needlr: Everywhere in Needlr

0:00 / 1:16

Quick tour of the words you'll see. A KPI is the one action you want more of. An impact check asks one thing: did this change move that KPI?

Every word you'll see in Needlr, in one plain line each. Use your browser's search to jump to a word.

The basics

Project: One website you measure. It has its own KPIs, impact checks and Tag Manager connection (more).
Demo project: A project full of sample results. It never touches a real site and doesn't count toward your plan.
KPI: The one thing you want more visitors to do, like click Book a demo, visit a page or send a form (more).
Impact check: What you create in Needlr to find out if the needle moved. One question: did this change move this KPI?
Verdict: An impact check's final answer: Improved, Got worse or No clear effect. Once it's given, it never changes (more).
Timeline: The project tab that shows every release and what it did to its KPI (more).

Kinds of impact checks

Split test: Some visitors see the old version and some see the new one, at the same time. A coin flip decides who sees which (more).
Split test (developer): A split test your developer turns on with one line of code, instead of the visual editor.
Visual editor: Opens your site with Needlr's editor on top, so you can click something and change it without code (more).
Before vs after: For a change already live for everyone. Compares the 4 weeks before the release with the 4 weeks after. Its verdicts are likely, not proof (more).
Adoption: For a new feature. Counts how many active people use it, against a target you set, like 20% within 30 days (more).
Control (A): The version visitors see today.
Variant (B): Your new version.
Share of visitors who see the new version: How many visitors get B in a split test. You pick it in the last step of a new impact check.

Reading a result

Needle: The gauge on every impact check. Right of the middle means better, left means worse. It stays grey until there's a verdict.
Lift: How much the KPI changed compared with where it started. From 3% to 3.6% is a lift of +20%.
Points: The plain difference between two rates. From 3% to 3.6% is +0.6 points.
Range: Where the real difference most likely is. On the needle, it's the shaded band.
Confidence: How sure the answer must be before Needlr calls it. 95% is the default. You can pick anything from 70% to 95%.
Sample: How many visitors each version needs before Needlr may call it. It's worked out before the impact check starts (more).
Smallest improvement worth knowing: The smallest change you care about finding. A smaller one needs more visitors. You'll find it under Advanced settings.
Full week: A split test never gets a verdict before 7 full days, because Monday visitors aren't Saturday visitors (why).
Collecting data: The impact check has just started. There aren't enough numbers to say anything yet.
Too early to call: Enough numbers to lean one way, not enough to decide. A lean is not a result.
Improved: The change moved the KPI the right way.
Got worse: The change moved the KPI the wrong way.
No clear effect: No real difference was found. If there is one, it's smaller than this impact check could see.
Data problem: Something is wrong with the counting, so the impact check is on hold until it's fixed (more).
Peeking: Looking at early numbers and stopping as soon as they look good. It finds winners that aren't real.
Uneven split: A and B got a different share of visitors than planned. Something is broken, so the impact check waits.
Stopped firing: Your KPI hasn't been counted for a while, maybe because a button changed. Impact checks that use it go on hold (more).

Setup and safety

Google Tag Manager (GTM): A free Google tool that adds small scripts to your site. Needlr puts its own tags there (more).
Container: Your site's box in Tag Manager. Its ID starts with GTM-.
Tag: One small script in the container.
Needlr snippet: Needlr's main tag. It loads on every page of your site.
Workspace: A separate drafting area inside Tag Manager. Needlr makes its changes in its own one.
Publish: Making Needlr's latest tags live on your site, through Tag Manager (more).
Undo: Putting your container back to the version that was live before Needlr's last publish. Also called a roll back.
Google Analytics (GA4): Optional. Needlr only reads it, so you can measure past releases with the history you already have.
Authenticator code: Six digits from an app on your phone. Every change to your live site needs one. One code covers 20 minutes (more).
No flicker: One line in your site's code, so a change at the top of a page shows without the old version flashing first.

After the verdict

Ship the winner: Show version B to everyone through Needlr with Ship version B, without waiting for a developer (more).
Shipped changes: The project tab that lists winners shown to everyone through Needlr, and helps move them into your site's code (more).
Measured visitors: Visitors who entered a running split test this month. The first 25,000 a month are free (more).

People

Owner: The person the account belongs to. Can do everything.
Editor: Sets up and changes impact checks, connects Tag Manager and publishes.
Viewer: Sees all the results, changes nothing. Viewers are always free (more).
Coding agent: An AI that writes code. With its own key it can ask Needlr questions and start impact checks, but it can never publish to your site (more).

What the video says

Quick tour of the words you'll see. A KPI is the one action you want more of. An impact check asks one thing: did this change move that KPI?

In a split test, the control is today's version and the variant is your change. Visitors are split at random, and each one keeps the version they got.

Say the control converts 3% and the variant 3.6%. That's a lift of 20%, or 0.6 points: lift is relative, points are the plain difference between the two.

Confidence is how sure the answer must be, and 95% is the default. The sample is how many visitors each version needs before Needlr is allowed to call it.

The verdict is the final word: improved, got worse, or no clear effect. Before that you'll see collecting data, then too early to call. Data problem means the impact check is on hold.

Tag Manager is the box of scripts your site already loads. Needlr puts its own tags there, and publishing makes them live, protected by your authenticator code.

Two more impact checks: before versus after compares weeks around a release that's already live, and adoption counts who uses a new feature. Now you speak Needlr.