Benchmarks

What is a Benchmark?

A benchmark lets you ask: Did this part of the product get better after we changed it?

You run an evaluation, make design changes, then run another evaluation. The benchmark keeps those runs together so Analytics can show the trend.

You do not need a multi-step user journey. Evaluations use a guideline set, so you can review a full checkout screen or something small like a dropdown.

Simple example

Imagine a pizza ordering app. Users need to add toppings. That need stays the same, but the UI can change:

v1: Pick a topping, then click Save (score: 58%)
v2: Pick a topping, and it auto-saves (score: 71%)
v3: Use chips to pick several toppings (score: 74%)

All three reviews belong under one benchmark, such as "Add pizza toppings".

Benchmark: "Add pizza toppings"
  ├── Evaluation: "Toppings v1: Save button"   (58%)
  ├── Evaluation: "Toppings v2: Auto-save"     (71%)
  └── Evaluation: "Toppings v3: Multi-select"  (74%)

Benchmark = the thing you keep measuring (the same need)
Evaluation = one review at one point in time (that version of the UI)

If you already write Use Cases in Requirements, you will often name a benchmark after one. They are not linked in the app. The name is just a helpful match.

You don't always need a benchmark

That's okay. A quick redesign, like updating a dropdown, can still get a full evaluation against your guidelines. Reach for a benchmark when you care about tracking scores over time. If you just need feedback once, a single evaluation is plenty.

Where to Find Benchmarks

Open Evaluations in the sidebar, then choose the Benchmarks tab.

Evaluations tab: individual review runs
Benchmarks tab: the groups those runs belong to

The Benchmarks Table

Each row shows:

Benchmark name: What you are tracking
Count: How many evaluations are in it
Last run: When the newest evaluation was updated
Edit: Menu for Analytics, edit, or delete

Create a Benchmark

Open Evaluations, then select the Benchmarks tab
Click New benchmark
Give it a clear name (e.g. "Add pizza toppings")
Optionally add a description
Click Save

You can also create one while starting a New Evaluation. You'll pick a benchmark when you create an evaluation: either a series you'll revisit for trends, or a simple group for a one-time review.

Edit, Delete, and Analytics

From the Edit menu on a row:

Analytics: View trends (only if the benchmark has at least one evaluation)
Edit: Change the name or description
Delete: Remove the benchmark

You can also open Analytics from an evaluation card with View Analytics.

If you delete a benchmark, the evaluations stay. They may just show as unassigned afterward.

When to Reuse vs Create New

Reuse the same benchmark when you are still measuring the same need, even if the UI changed a lot, such as from a Save button, to auto-save, to chips.

Create a new benchmark when it is a different part of the product (e.g. toppings vs payment).

Tip: Name evaluations after the version you reviewed ("Toppings v2: Auto-save"), so the list is easy to scan later.

Tips for Better Results

Reuse the same guideline set when you want fair score comparisons
Re-evaluate after meaningful design changes
Check Analytics after each round to see what improved or got worse

Benchmarks

On this page