How do you run a UX competitive analysis?

In short

In a competitive analysis, you look at the products users look at before they choose you. I complete the same task in each product from start to finish and note where it’s easy and where it’s hard.

NN/g calls this work a “competitive usability evaluation”. The definition is short: comparing your product against several competing designs. The comparison can cover the whole site, a single feature or a piece of content.

The competitive analysis on this page doesn’t look at market share or price. Marketing’s competitive analysis is a different job. Here the question is: in which product does the user complete the same task with less effort?

When to do it

A competitive analysis is most useful before you design. NN/g sees competitive evaluations as a good initial research activity for a project. They can inform the design direction and uncover whether a feature is needed at all.

I recommend a competitive analysis in these cases:

• Before you draw a new product or a new feature. A screen you draw without knowing the familiar solutions can feel strange to users.
• When sales says, “The customer chose the competitor.” In my experience, the reason is often friction in a flow.
• After a UX audit. The audit finds the problem; the competitive analysis shows how others solve the same problem.
• When a redesign discussion turns into a battle of opinions. Competitors’ solutions bring the discussion back to concrete examples.

It works even without a product of your own. According to NN/g, designs don’t need to be fully developed or interactive to be evaluated. You can compare your paper prototype with competitors on the same task.

Which competitors to choose

NN/g writes that a typical competitive evaluation focuses on 2 to 4 competitors. Any more than that becomes too expensive and too overwhelming to analyze. I usually start with two competitors and one example from another industry too.

Direct competitors

Products that sell the same thing to the same customers. For an online store, another store in the same category; for a banking app, another bank’s app. These are the products users are most likely to compare you with.

Indirect competitors

Products that meet the same need in a different way. An indirect competitor of accounting software can be a spreadsheet. If users see it as an alternative, they bring their habits from there to you.

Examples from other industries

Products that solve a similar task well but don’t compete with you at all. If you’re building a quote form for insurance, it pays to look at a good flight search form. NN/g also recommends looking not only at direct competitors but at businesses that are only tangentially related. There you see practices that have proven themselves in other industries.

If there are many candidates, take a quick look at several first, then choose. Products with similar content and functionality are the first candidates. NN/g also recommends looking at the ones with the best overall user experience and the ones your customers are most likely to compare you against.

Don’t choose only bad products. A weak competitor shows what doesn’t work, but you learn more from a good one.

What to look at

If you don’t rate every product on the same criteria, you can’t compare them. I use six points. The same points are the rows of the matrix below.

Core tasks. Choose 3 to 5 tasks that users come to the product for. In a shipping app, for example, “Create a shipment” and “Track a shipment”. Note whether each task succeeds in each product. The task success rate is the cleanest measure here.

Flow and number of steps. How many screens, fields and taps does the same task need? Is the same information asked for twice? Can you check out as a guest, or is an account required?

Content and microcopy. Read button labels, field descriptions, price and delivery information. Is the question on the user’s mind answered on screen? In my experience, you see the difference between competitors fastest in the UX writing.

Accessibility. Can the task be completed with the keyboard alone? Is the focus ring visible? For this, run the same scan on each product with one of the accessibility testing tools. Then repeat the keyboard round by hand.

Performance. How quickly is the page usable, and does content shift while loading? To compare competitors’ field data, you can use the method from the guide to Core Web Vitals.

Errors and feedback. Deliberately do something wrong, such as entering an invalid coupon. Does the product say what went wrong and what to do now?

A competitive analysis step by step

1. Write down the goal. NN/g recommends making the goals clear before any work starts. Which design challenge are you trying to solve? “Why do we lose customers at the payment step” is a good goal; “Let’s take a look at the competitors” isn’t.

2. Choose the competitors. Two or three direct competitors and one example from another industry are enough.

3. Set the tasks and criteria. In each product, you do the same tasks with the same test data. Write the list of criteria before scoring and don’t change it afterwards.

4. Use each product and record. I do each task separately on mobile and desktop, with a screen recording. I note every place where I get stuck, with a screenshot of that moment.

5. Score. Each product gets 0 to 3 points for each criterion. I give the scores in a separate round after the notes are finished, not while using the product.

6. Draw out patterns. NN/g’s questions help here. What are the biggest strengths of competing designs? What trends arise across products? Is there a gap that nobody solves well?

7. Turn it into a decision list. Each finding goes one of three ways: fix, adopt or stand out.

Competitive testing with users

Going one step beyond the expert review means having real users do the same tasks in the competitors’ products. NN/g says each participant typically completes tasks on 2 or 3 sites here. More than 3 can overwhelm participants.

Alternate the order of the products from person to person. What someone learns in the first product makes the second look easier than it is. At the end of the session, ask participants to compare the products. But for the decision, what people did counts, not what they say. You’ll find the details in the guide to usability testing.

The scoring matrix

The matrix puts the analysis on one page. Rows are criteria, columns are products. Each cell gets a value from 0 to 3.

Points	Meaning
0	Missing or not working. The task fails or the feature doesn’t exist.
1	Weak. The task succeeds, but the user clearly struggles.
2	Adequate. Small snags, but the work gets done.
3	Strong. Solved so well that you can take it as a model.

Why 0 to 3? Four levels without a middle make it hard to move on with “so-so”. Always write a one-sentence reason next to each value. A value without a reason means nothing to anyone a week later.

Build your own matrix with the tool below. Change the product names, choose the scores and then copy the table as Markdown into your document.

Competitive analysis matrix

Column 1

Product name

Core tasks, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Flow and number of steps, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Content and microcopy, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Accessibility, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Performance, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Errors and feedback, Your product0 · Missing1 · Weak2 · Adequate3 · Strong

Total 11 / 18

Weakest criterion Flow and number of steps, Accessibility (1)

Column 2

Product name

Core tasks, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Flow and number of steps, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Content and microcopy, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Accessibility, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Performance, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Errors and feedback, Competitor A0 · Missing1 · Weak2 · Adequate3 · Strong

Total 13 / 18

Weakest criterion Accessibility (1)

Column 3

Product name

Core tasks, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Flow and number of steps, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Content and microcopy, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Accessibility, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Performance, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Errors and feedback, Competitor B0 · Missing1 · Weak2 · Adequate3 · Strong

Total 11 / 18

Weakest criterion Content and microcopy, Performance (1)

Your matrix

Criterion	Your product	Competitor A	Competitor B
Core tasks	2	3	2
Flow and number of steps	1	2	3
Content and microcopy	2	2	1
Accessibility	1	1	2
Performance	3	2	1
Errors and feedback	2	3	2
Total (max. 18)	11	13	11
Weakest criterion	Flow and number of steps, Accessibility (1)	Accessibility (1)	Content and microcopy, Performance (1)

Without JavaScript, you can change names and scores. The totals and table then stay at the sample scores.

The starting scores are examples I made up, not a rating of any real product. What you enter isn’t sent anywhere or stored.

Give each cell 0 to 3 points: 0 missing or not working, 1 weak, 2 adequate, 3 strong. The total and the weakest criterion update as you score. In the example, two products have the same total but different weak spots.

Don’t get stuck on the total. As NN/g stresses, the main goal of a competitive evaluation is not to declare a winner but to improve your design. Even in a product with a high total, the weakest criterion can be the place where you lose users.

Don’t read too much into the numbers either. One person’s scores are an opinion, not a measurement. The real value lies in the notes and screenshots behind the cells.

Combining it with a heuristic evaluation

If you combine the competitive analysis with a heuristic evaluation, your scores get a foundation. Instead of “The flow is weak”, you write which heuristic is violated. A competitor might not show which step you’re on during checkout. That violates the first of Nielsen’s 10 usability heuristics: visibility of system status.

NN/g’s guide to heuristic evaluation also recommends these 10 heuristics as the list.

In practice, I look at each product with the same list in two passes. As in the guide, in the first pass I get familiar with the product, and in the second I compare it with the heuristics. The guide also recommends narrowing the scope: one task, one area or one device.

Two warnings from the guide apply to competitive analysis too. Ideally, three to five people evaluate the same interface independently. And a heuristic evaluation can’t replace user research.

The risk of copying, and Jakob’s law

One trap in competitive analysis is simply copying a screen you like. NN/g puts it in one clear sentence: you want to beat the competition, not copy them.

The risk with copying: you see the competitor’s screen, but not the decision behind it. The solution may have been chosen for their users, their price and their technical limits. Maybe it failed in their own tests, and you copy it without knowing.

On the other hand, doing everything differently is also a mistake. That’s what the principle called Jakob’s law describes. Jakob Nielsen wrote in 2000: “Users spend most of their time on other sites.” That’s why they prefer your site to work the same way as the sites they already know.

So in a competitive analysis, I keep two separate lists:

• Conventions to follow. Things that work the same way at every competitor, such as the position of the cart or the order of the checkout steps. If you change them, users have to relearn.
• Chances to stand out. Places no competitor solves well. The row where everyone gets a 1 is the row where you can pull ahead.

My rule is simple: be similar in conventions, differ in weaknesses. If users have a clear expectation of how a screen works, stick to it. This expectation is called a mental model. You make the difference where your product does its actual job better.

My note

The most common mistake in competitive analyses: the analysis turns into a feature list. One column says “available”, the other “missing”. But the fact that a feature exists doesn’t mean it can be used.

That’s why I start the matrix with the task. First I try to complete the task in each product, then I look at the features. A competitor that looks incomplete on paper sometimes completes the core task in fewer steps.

And don’t think of competitive analysis as a one-time job. Competitors change their interfaces. Repeat the same tasks once a year with the same criteria. It’s a cheap way to see change.