Design Guide: Test Your Add-on Customer Experience
While you can use the Certification Guidelines and Functional Requirements for guidance on functional testing, you'll want to test your add-on on an Alexa device to check it also meets a high CX quality bar.
How to use this guide
Run through your add-on's complete flow on your primary device multiple times, then repeat on additional devices. Most themes are observational: you can evaluate them while using your add-on normally. Test with a range of utterances, and don't just test the happy path. Try:
- Highly specific requests with all details included ("Book a table for 4 at Nobu on Saturday at 7pm")
- Vague or partial requests ("Find me a restaurant")
- Mid-flow corrections ("Actually, make that 6 people")
- Requests at the edge of your capabilities ("Do you have anything cheaper?")
- Requests outside your scope entirely (to verify graceful handling)
Test across devices. The same utterance can produce different results on different screen sizes and input modes. Run end-to-end testing on:
- Echo Show (primary)
- Mobile (Alexa app) — voice and keyboard
- Web (alexa.com) — voice and keyboard
- A voice-only device (for example, Echo Dot)
The combination of varied utterances × varied devices × varied input modes is where most real-world issues surface.
Pattern usage
Check whether UI patterns are used for their intended purpose. See Components and Patterns.
| Pattern | Intended for |
|---|---|
| List | Vertical stack of homogeneous items |
| Carousel | Browsing related items side by side |
| Card | One item deserving individual focus |
| Map | Location-based results with pins |
| Inline mode | Confirmations, summaries, single decisions |
| Fullscreen | Dense/complex interactions, games, panning |
Watch for:
- A Card being used to display multiple items (should be a Carousel or List)
- Fullscreen mode for simple confirmations (should be Inline)
- Frequent switches between inline and fullscreen without a clear pattern
- Results that aren't scrollable
Legibility and readability
Check that all text is readable at typical viewing distances.
Verify:
- All text on screen is readable at arm's length
- Essential information (price, name/title, primary action) is legible from across the room
Watch for:
- Text that stops mid-word with no ellipsis (…) or "see more" affordance
- Content that extends off-screen with no scroll indicator
- Any truncation on commit screens (checkout, booking confirmation, payment) — all details must display completely on these screens
Typography
Check that fonts are appropriately sized and styled.
Verify:
- No text is too small to read comfortably at arm's length
- No text is disproportionately large relative to the rest of the interface
Voice and screen agreement
Check that what Alexa says matches what the screen shows.
Watch for:
- Price discrepancies between voice and screen
- Dates, times, or quantities that don't match
- Alexa referencing an item that's scrolled off-screen or not visible
- Status or policy contradictions ("free cancellation" spoken, "non-refundable" shown)
Fit and finish
Check the overall visual polish of your experience.
Watch for:
- Misaligned elements
- Inconsistent corner radii between cards or containers
- Cropped images or content
- Typography inconsistencies (mixed sizes without clear hierarchy)
- Unintentional excessive white or black space
- Elements that look tappable but don't respond to touch
Image quality
Check that all images render cleanly.
Watch for:
- Pixelated or blurry images
- Stretched or distorted aspect ratios
- Broken image placeholders visible to customers
- Images that look correct in one mode but break in the other (light/dark)
Task completion
Check that customers can finish what they started.
Verify:
- Tasks your add-on declares in tool descriptions and your Alexa+ Store listing can be completed end-to-end within Alexa
- The results your search surfaces connect to a completable transaction or action
Watch for:
- Dead ends where the customer has no path forward
- Handoffs to a phone, app, or website for tasks that could resolve within the conversation
- Results that appear in search but can't be acted on (for example, a restaurant shown but not bookable)
Error recovery
Check what happens when things go wrong.
Verify:
- The experience remains recoverable after an error: Alexa can call your tools again to try a different path
Watch for:
- Generic "something went wrong, try again" messages with no guidance
- Loops where the same error repeats with no escape
- Unrecoverable states that require the customer to start the entire flow over
Friction and gating
Check that customers can explore before committing.
Verify:
- Customers can browse, search, and view results without signing in or linking accounts, if possible
- Account linking and permissions are requested only at the moment action requires them
Input flexibility
Check that your add-on handles the way real people actually talk.
Verify:
- The experience works with varied phrasings for the same intent
- Partial inputs still produce useful results (not just errors)
- Corrections mid-flow are handled gracefully
Watch for:
- Only accepting one specific phrasing to trigger a tool
- Failing on partial inputs instead of asking a clarifying question
- Ignoring corrections ("Actually, make that Thursday" doesn't change anything)
Cross-device rendering
After primary testing, visit 2–3 key screens on each device in your test set.
Per device, verify:
- Does it render at all?
- Is text equally readable?
- Is it equally navigable?
- Are there any device-specific visual issues?
- Can you complete the same tasks as on your primary device?
Test both light and dark mode on at least one screen device.
Follow-through: emails and notifications
If your add-on sends confirmation emails or notifications, check that they:
- Are professionally designed (not plain-text dumps)
- Contain actionable content (clear what to do next)
- Are consistent with the in-experience branding
- Show correct information matching what was confirmed on screen
Reporting Alexa+ issues
During testing you may encounter issues that originate from Alexa rather than your add-on. For example, Alexa pronouncing something wrong or talking too long.
Say "report a bug" to Alexa to log the issue directly from your device at any time. You'll receive an email confirmation when your report is received. If confirmation doesn't arrive, follow up with your assigned Solutions Architect.
Accessibility testing
Run through your add-on with each of the following settings and scenarios. All steps apply to multimodal devices unless noted.
Input
- Complete your full CX end-to-end using voice only; don't touch the screen.
- Complete your full CX end-to-end using touch and keyboard only; don't use voice. If your CX includes any form fields, verify all fields are reachable and fillable via the on-screen keyboard.
Captions and spoken output
- Go to Settings → Accessibility → Captioning. Enable captions and run your CX. Verify Alexa's responses appear as on-screen captions.
- Go to Settings → Accessibility → VoiceView. Enable VoiceView and navigate your CX by touch. Verify: (a) items on screen are read aloud, and (b) the information read by VoiceView matches what's shown on screen.
- If your CX includes sounds or audio cues (for example, a chime or alert): verify there is also a visual indicator on screen.
Information parity
- Complete your CX end-to-end with sound off and captions on. Verify the same quality of information is available via captions as via voice.
- Complete your CX end-to-end without looking at the screen (voice and audio only). Verify key information is conveyed audibly.
Assistive technology
- Go to Settings → Accessibility → Screen Magnifier. Pinch to zoom in and out. Verify your layout doesn't break and no content becomes hidden or unreachable.
- Go to Settings → Accessibility → Color Correction. Run your CX. Verify it remains usable.
- Go to Settings → Accessibility → Color Inversion. Run your CX. Verify it remains usable.
Related topics
| Go back to: Evaluate Your UX | Next: Store Listing |
Last updated: Jul 21, 2026
