UX case study

2-minute skim

Sweetgreen UX audit

Making ordering faster, safer, and more user friendly for pickup customers

Sweetgreen website home page

Role

Lead evaluator

Team

3 evaluators

Timeline

2 weeks, Fall 2026

Methods

Heuristic evaluation, severity rating, theme analysis, mockups

Tools: Figma

The opportunity

Overview

Sweetgreen's website is visually appealing and easily allows the user to find what they're looking for, with clear and nicely placed images of their menu items and an order-building process that is user friendly and customizable. However, the most severe problems appear at the points where the user has to make a decision: checking allergens while building a salad, choosing a pickup location, and completing checkout without an account. Our team of three evaluated the pickup ordering journey, and as lead evaluator I compiled our findings into three recommendations.

The audit at a glance

13

usability issues found

8

rated Major or Catastrophe

3

recommendations made

Issues by severity

1 bowl = 1 issue · 13 total issues

Catastrophe3

Major5

Minor4

Cosmetic1

The scenario

One lunch hour. Six tasks.

You work in an office in the Empire State Building and have one hour for lunch. It is 1pm, and you want to place an order for a custom salad for pickup at the closest Sweetgreen to you. You are allergic to gluten, so you need to confirm that the salad you are ordering doesn't include any ingredients with gluten.

  1. 01Find the nearest Sweetgreen location to the Empire State Building and select for pickup.
  2. 02Build a custom salad with the following ingredients: quinoa, kale, apple, pickled onions, almonds, 2x chicken, avocado, parmesan, and green goddess ranch.
  3. 03Confirm the order is gluten-free before proceeding.
  4. 04Add a Spindrift Lemon to your order.
  5. 05Proceed to checkout.
  6. 06Attempt to complete your order without creating an account or signing in.

Finding 01

Allergen information

CatastropheFound independently by all 3 evaluators
BeforeExisting experience

The problem

When building a salad or looking at the menu, users can't see which ingredients contain common allergens like gluten, nuts, and dairy, or filter by dietary restriction. This information is only in a large nutrition chart that is hard to find and has to be cross-checked by hand: a health risk for users with allergies.

Sweetgreen ingredient builder showing dressings with no allergen information
Figure 1: Ingredient builder without allergen information
Dense nutrition and allergen spreadsheet
Figure 2: Separate nutrition and allergen chart
AfterProposed redesign

The proposed fix

Adding a dietary filter bar to the ingredient builder so users can select their needs and hide items that don't fit, and labeling every ingredient with tags for the allergens it contains. This brings the allergen information Sweetgreen already has in its nutrition chart into the builder, so users can confirm their meal is safe while building it instead of cross-checking a separate chart.

Ingredient builder with a dietary filter bar, Gluten-free selected
Figure 3: Dietary filter in the ingredient builder
Dressings grid with allergen tags such as Contains milk and Contains wheat on each item
Figure 4: Allergen tags on ingredients. Allergen tags are illustrative.

'Allergen info is buried at the bottom in an overwhelming chart.'

From the independent evaluation

Finding 02

Guest checkout

CatastropheFound by one evaluator, included for its Catastrophe rating
BeforeExisting experience

The problem

Users only find out they need an account after building their salad and clicking Continue to checkout. The next screen offers only sign in or account creation, with no guest option, which can frustrate and deter first-time or occasional customers from completing their order.

Sweetgreen cart panel with a single Continue to checkout button
Figure 5: Only one option forward in the cart
Sign in or create an account screen with only an email field
Figure 6: Sign in or create an account, with no guest option
AfterProposed redesign

The proposed fix

Adding a 'Check out as a guest' option alongside 'Sign in or create account' in the cart, so users choose how to check out before being redirected. The guest path asks only for a name and one contact method, while sign in stays available as a link for returning customers. This removes the account creation barrier for first-time and occasional customers without taking anything away from rewards members.

Cart panel with two buttons: Sign in or create account, and Check out as a guest
Figure 7: Guest and sign-in options before redirecting
Guest checkout form asking for a name and a phone number or email
Figure 8: Guest checkout with a name and one contact method

18%

18% of surveyed US online shoppers cited 'The site wanted me to create an account' as a reason for cart abandonment, (Baymard Institute, 2025).

US online shoppers, not Sweetgreen customers. Excludes just-browsing shoppers.

Tradeoff: Sweetgreen likely requires accounts to grow its rewards program, so the design keeps sign in prominent.

Finding 03

Location search

MajorFound independently by all 3 evaluators
BeforeExisting experience

The problem

The location finder is easy to find on the website's main page, and the map helps users see how far they are from Sweetgreen locations. However, it only searches by zip code or store name. Typing a full, valid street address returns an error that doesn't suggest an alternative search to try. Once stores appear, they show no distance from the user, and the map pins can't be used to select a store.

Location finder showing an error after a street address search
Figure 9: A valid street address returns an error
Map of Manhattan with store pins and a store list with no distances
Figure 10: Location pins and stores without distance labels

When search fails

Give the customer a way forward.

When a location can't be matched, the message explains why and offers two paths forward: specific formats to try, or a link straight to the full location finder.

AfterProposed redesign

The proposed fix

Adding the ability to search by street address as well as zip code, with autocomplete so the user can confirm the full correct location, and a clear error message that suggests alternatives when a search fails. Additionally, letting users select a location directly from the map pins and organizing locations by distance. This lets users search the way they naturally think of where they are and compare stores by distance.

Location search with a dropdown of suggested street addresses
Figure 11: Address search with autocomplete
Error state suggesting a street address or zip code and linking to the locations page
Figure 12: A clearer error state with actionable guidance
Map with a Penn Plaza card and Select for pickup button, and distances on each store
Figure 13: Distance labels and a selectable map card

The process

How we got there

The three evaluators completed the tasks independently, each noting the problems they found, the heuristics violated from Nielsen's 10 usability heuristics (Nielsen, 2024), and a severity rating on Nielsen's 0 to 4 scale (Nielsen, 1994). As lead evaluator, I compiled the findings from all three evaluations into a single document and identified overlapping themes. Findings were prioritized by severity first. Where multiple evaluators independently flagged the same problem, I treated that consensus as extra evidence of a real, repeatable issue and used the highest rating reported rather than an average, to avoid under-prioritizing a potentially critical issue.

  1. 01Evaluate independently
  2. 02Compile and group themes
  3. 03Prioritize highest severity

What I learned

Reflection

Leading this evaluation taught me to separate how often a problem was reported from how seriously it could affect someone. Prioritizing by the highest severity helped me keep allergy safety in focus while balancing it with ordering speed.

References

Baymard Institute. (2025, September 22). 50 cart abandonment rate statistics 2026. https://baymard.com/lists/cart-abandonment-rate

Moran, K., & Gordon, K. (2023, June 25). How to conduct a heuristic evaluation. Nielsen Norman Group. https://www.nngroup.com/articles/how-to-conduct-a-heuristic-evaluation/

Nielsen, J. (1994, November 1). Severity ratings for usability problems. Nielsen Norman Group. https://www.nngroup.com/articles/how-to-rate-the-severity-of-usability-problems/

Nielsen, J. (2024, January 30). 10 usability heuristics for user interface design. Nielsen Norman Group. https://www.nngroup.com/articles/ten-usability-heuristics/

Sweetgreen. (n.d.). Sweetgreen. Retrieved September 24, 2026, from https://www.sweetgreen.com/