By Nick Babich
This page breaks down ten annotated mobile app screens, one per app type, from fitness and food delivery to smart home and an AI camera feature, and shows what each screen does well and which pattern to borrow.
Key Takeaways
- Start with the moment, not the look: each screen below is built for one person in one situation, mid-run, at the gate, at the curb, and it answers the one question that person has right then.
- Galleries show screens without the reasoning: in my November 12, 2026 check of the eight page-one results for this query, none annotated a screen with who opens it, what it answers and what to check before shipping.
- Design for the interrupted user: a gate change, a late courier, a lost signal or a call mid-run is the normal case on a phone, so the strongest examples show the change on the screen the user already has open.
- Put the main action under the thumb: Pause, Continue, Delivered and Retake photo all sit in the lower half, and risky actions like ending a run take a hold instead of a tap.
- Borrow the reasoning, not the skin: the ten products share no palette, typeface or theme, yet they follow the same few rules, and those rules are what transfer to your app.
A good mobile app screen answers one question for someone holding the phone in one hand, often while walking, waiting or being interrupted. Galleries rarely show that part. On November 12, 2026, I read the eight organic results Google shows on page one for “mobile app design examples” in the US. Seven were galleries, template libraries or example roundups, and one was a Reddit thread of designers trading inspiration sources. Muzli showed 60 concept shots, each with only a title. A Pinterest board loaded 25 pins. Mobbin loaded more than 80 real app screenshots before it asked me to log in. Two pages added a few sentences per app or shot, mostly about how it looks. None of the eight annotated a screen with who opens it, what question it answers and what to check before it ships.
How I counted: first page loads only, before scrolling or logging in, with logos and icons left out. Dribbble, the Medium post and the Reddit thread did not load for my check, so I describe those three from the results page itself.
This page takes the opposite approach. It shows ten annotated mobile screens, one per app type, from a fitness app mid-run to an AI plant identifier. Every product is fictional, with invented data, and each was drawn for this article in its own visual style: dark and light themes, iOS and Android, condensed, rounded and serif type. None is a screenshot of a real app. For each screen you get the person and their question, what works, and what to check before you ship something similar. It is written for founders, product managers and designers who review mobile work. If you would rather hire help than collect references, start with the mobile app UX design teams we have compared; the rest of this page stays on the screens.
Mobile App Design Examples at a Glance
Use the table to jump to the app type closest to yours. The last column is the test each screen has to pass.
| # | App type | Screen | Product (fictional) | User's main question |
|---|---|---|---|---|
| 1 | Fitness | Active run | Pacebird (iOS, dark) | “How far, how fast, how long to go?” |
| 2 | Food delivery | Order tracking | Saltpan (iOS, light) | “When will it get here, and is it on track?” |
| 3 | Travel | Boarding pass with a gate change | Skerry Air (iOS, light) | “Where do I go, by when, and will this scan?” |
| 4 | Podcast | Now playing | Hushwave (iOS, dark) | “Where am I in this episode, and how do I skip back what I missed?” |
| 5 | Messaging | Group chat with a poll | Kinloop (iOS, light) | “What did we decide, and what do I need to do?” |
| 6 | Language learning | Answer feedback | Wordwren (Android, light) | “Was I right, and what should I say instead?” |
| 7 | Smart home | Room control | Brightden (iOS, dark) | “Is it warming up, and when will it be comfortable?” |
| 8 | Delivery driver | Next stop | Dropwise (Android, light) | “Am I at the right place, how do I get in, and how do I close this stop?” |
| 9 | Event tickets | Ticket at the gate | Gatefold (iOS, dark) | “Will my ticket scan, and which gate do I use?” |
| 10 | AI camera feature | Plant identification result | Leafsight (iOS, light) | “What is this, how sure is the app, and how do I care for it?” |
How to Read a Mobile App Screen Example
A phone screen is used in conditions a desktop screen never meets: one hand, bright sun, a moving bus, a call that cuts in. So before you look at the examples, or save one from a gallery, run five checks on it. I apply the same lens to the shipped apps behind the UX design teams we review for consumer and mobile products, and it tells you far more than the color palette does.
- 01 Interruptions: what happens after a phone call, a lost signal or a return to the app two hours later? A good screen picks up where the user left off and says what changed.
- 02 Platform navigation: do back, tabs and gestures work the way iOS or Android users expect? Wordwren and Dropwise are drawn as Android screens for that reason.
- 03 Glanceability: can the user get the answer in one glance? Pacebird’s three numbers and Saltpan’s arrival window are the test cases.
- 04 Real conditions: does the screen hold up in sunlight, in motion, with large system text and in dark mode? Four of the ten examples are dark, and Dropwise is built for glare.
- 05 Thumb reach: can the main action be done with one thumb? Look at where Pause, Continue, Delivered and Save to My plants sit.
These checks are a reading tool, not a set of platform rules. The guidelines behind them, from touch targets to offline states and permission prompts, are covered in how platform guidelines, offline states and permission requests shape mobile design. Here they only point you at what to look for on each screen.
10 Annotated Mobile App Screens by App Type
Each example follows the same order: the image, the person and their question, what works, and what to check before you build something similar. The numbers in parentheses match the numbered callouts on each image, and every figure in the text is the one shown on the screen.
1. Fitness: Pacebird Active Run Screen
Who and what question: A runner glancing at the phone mid-stride, on an easy Zone 2 run: “How far, how fast, how long to go?”
What works: Three numbers fill the screen, readable at arm’s length (1): 6.42 km, 36:29 and an average pace of 5:41 /km. Current pace (5:38 /km) and heart rate (142 bpm, with the second of five zone segments lit) sit in a smaller row, and a goal strip says “Goal 8.00 km · 1.58 km to go.” There is one primary control, a large lime Pause button (2), and ending the run takes a hold (“Hold to end”) so a bump in a pocket does not stop it (3). The tab bar is gone during the run, so nothing else competes for a sweaty tap (4).
Check before shipping: What the distance does when GPS drops in a tunnel, whether an incoming call keeps the run recording, contrast in bright sunlight, and the same three numbers on the lock screen or a watch.
2. Food Delivery: Saltpan Order Tracking
Who and what question: A hungry customer who already paid: “When will it get here, and is it on track?”
What works: The headline is an arrival window, “Arriving 7:42-7:50 PM”, not a fake-exact minute (1). Below the map (the courier is 1.4 km away), one progress line runs Placed 7:08, Cooking 7:11, Picked up 7:29, On the way, Delivered, and it uses the same steps as the push notifications (2). Message and Call for the courier, Tomasz, are big buttons within thumb reach (3). The order itself is collapsed to “3 items · $34.60” because it no longer matters at this point (4), and the delivery note “Leave at door, buzz 4B” is one Edit away.
Check before shipping: Whether the window widens honestly when the courier runs late, what the screen says when the courier is reassigned, hiding the courier’s live location after delivery, and a lock-screen version of the same progress line.
3. Travel: Skerry Air Boarding Pass With a Gate Change
Who and what question: A traveler at the airport: “Where do I go, by when, and will this scan?”
What works: The gate change sits on the pass itself, in the alert color: “Gate changed: B12 → B31 · about a 9-minute walk”, with Show route (1). On the pass, the new gate B31 appears with the old B12 struck through rather than silently replaced (2), so a traveler who remembers B12 knows the change is real. Flight SK 418, BOS to LIS, boarding 8:55 PM, departing 9:35 PM, seat 23A, group 4. The QR code is the biggest object on the screen, with “Brightness raised for scanning” under it (3), and “Available offline” is stated, not assumed (4).
Check before shipping: What the pass shows with no signal at the gate, whether the gate-change push and the pass can ever disagree, how wallet passes and screenshots get updated, and the second passenger’s pass (the dots say 1 of 2).
4. Podcast: Hushwave Now Playing
Who and what question: A listener walking or driving: “Where am I in this episode, and how do I skip back what I missed?”
What works: Chapters make a 52-minute episode navigable: the pill reads “Chapter 3 · Curb cuts and strollers · 16:02” and the scrubber carries chapter ticks (1). The scrubber shows 18:44 played and 33:26 remaining, which adds up to the 52:10 episode. Skip back is 15 seconds, shorter than the 30-second skip forward, because catching a missed line is the common case (2). Play/Pause is the largest target and sits in the bottom half (3). Speed (1.2×), a sleep timer set to “End of episode” and a Downloaded label sit in one row, so download status is visible before the commute (4).
Check before shipping: Pause when headphones disconnect instead of playing through the speaker, lock-screen and car controls, resume position across phone and web, and very long episode titles.
5. Messaging: Kinloop Group Chat With a Poll
Who and what question: One of six friends planning a Saturday hike: “What did we decide, and what do I need to do?”
What works: Jonah’s quoted reply (“Then early, I can drive 3 people”) keeps the context of Mireille’s note that trailhead parking fills by 8 (1). The decision lives in the thread as a structured poll card, not in a separate tool (2). The result is visible without opening it: 7:30 AM with 4 votes and a check, 9:00 AM with 2 (3). The footer says “You voted for 7:30 AM” and “6 of 6 voted · closes Fri 9:00 PM”, so everyone knows who voted and when it closes (4).
Check before shipping: Long names and long poll options, how a muted group notifies about a poll, what the keyboard covers when someone replies to the poll, and how the card looks to a member added after it closed.
6. Language Learning: Wordwren Answer Feedback
Who and what question: A learner on question 7 of 12 who just answered: “Was I right, and what should I say instead?”
What works: The prompt was “Translate: I’d like two kilos of tomatoes,” and the answer “Quiero dos kilos de tomates” gets an orange sheet. The state is shown with an icon and the word “Almost!”, not color alone (1). The sheet names the exact word to change, “Better: Quisiera dos kilos de tomates” (2), and explains why in one sentence: Quiero is correct but sounds abrupt to a shopkeeper (3). “Hear it” plays the correct form (4), and the full-width Continue button sits where the thumb already is (5). This is the Android version, with a Material-style top bar and the system gesture bar.
Check before shipping: Color-blind contrast between the green correct state and the orange almost state, audio when the phone is on silent, accented input (é, ñ) on the keyboard, and what the Android back gesture does mid-lesson.
7. Smart Home: Brightden Room Control
Who and what question: Someone at home or on the way home: “Is it warming up, and when will it be comfortable?”
What works: Target and current temperature are both visible, 70°F set and 67°F now (1). A time estimate, “Heating · reaches 70° around 6:40 PM,” means slow heating does not feel broken (2). Heat is selected in the mode control, the schedule line reads “Next: Eco at 11:00 PM (62°F),” and tiles show lights 3 of 5 on. One device problem becomes one fixable row: “Hallway sensor offline since 2:14 PM · Fix” (3). Big minus and plus buttons beside the dial change the target without precise dragging (4).
Check before shipping: What the dial shows while a command is still sending, a whole-hub-offline state, what a guest or child account is allowed to change, and the °C/°F switch.
8. Delivery Driver: Dropwise Next Stop
Who and what question: A driver at the curb holding parcels: “Am I at the right place, how do I get in, and how do I close this stop?”
What works: The top bar reads Stop 14 of 38, with 13 done and 25 left. The address, 2218 Alder St, Apt 3, is set in large type for glancing in sunlight (1). The details a driver needs at the door are on the card, not behind a tap (2): recipient R. Lindqvist, by 2:30 PM, 2 packages, “Gate code 4471” and “Dog in yard. Leave at side door.” A mini map carries a Navigate button. At the bottom, Delivered and Can’t deliver are equal in size, so the failure path is not hidden (3), and Delivered says up front that a photo is required (4).
Check before shipping: A proof photo taken with no signal (queue it and retry), gloves and wet screens, what Can’t deliver asks next, and stops reordering mid-route.
9. Event Tickets: Gatefold Ticket at the Gate
Who and what question: A fan in the entry line: “Will my ticket scan, and which gate do I use?”
What works: Everything needed at the gate fits on one screen (1): The Lindens, Northbound Tour, Riverside Amphitheater, Sat, Nov 28, 2026, doors 6:30 PM, show 8:00 PM, Section 104, Row K, Seat 12. The rotating code is explained in plain words, “Code refreshes every few seconds · screenshots won’t scan,” so people do not screenshot it and get stuck at the turnstile (2). “Enter via Gate C” is printed on the ticket (3), and “Available offline” is stated before anyone reaches a venue with no signal (4). Add to Wallet and Send to a friend handle the second ticket (1 of 2).
Check before shipping: Low battery and low-power mode, two tickets on one phone for group entry, the brightness boost, and what happens when a sent ticket has not been accepted yet.
10. AI Camera Feature: Leafsight Plant Identification
Who and what question: A plant owner who just took a photo of a leaf: “What is this, how sure is the app, and how do I care for it?”
What works: The answer is phrased as likely, “Likely a fiddle-leaf fig” (Ficus lyrata), with a “High match” label (1). Look-alikes are one tap away: rubber plant and banana-leaf fig under “Could also be,” with Compare (2). Care basics fit in three short lines (3): Light: Bright, indirect. Water: When top 2 in of soil is dry. Tip: Rotate the pot once a month. Retake photo is as visible as Save to My plants (4), and “Not right? Tell us” gives a way to correct it.
Check before shipping: What the screen says for a blurry photo or a low match, behavior with no signal, and where a correction goes after “Not right? Tell us”.
What the Ten Examples Have in Common
- One question per screen: every example can be summed up as one person and one question, and the answer is the largest thing on the screen: the run distance, the arrival window, the QR code, the target temperature.
- The change shows up where the user already is: the gate change sits on the boarding pass, the poll result sits in the chat, the offline sensor sits on the room screen. Nobody has to find a separate inbox.
- Honest status: an arrival window instead of an exact minute, “Likely” with a match label, “reaches 70° around 6:40 PM,” “Available offline.” Each tells the user what the app knows and what it does not.
- Main actions in the lower half: Pause, Message and Call, Continue, Delivered and Can’t deliver, Save and Retake all sit where a thumb reaches, and failure paths are the same size as success paths.
- Built for the bad moment: hold to end, struck-through gates, screenshots that won’t scan, a proof photo that is required: these are designed for the bump, the rush and the dead signal, not the demo.
Where to Find More Mobile App Examples
Ten annotated screens teach you how to read a screen; they are not a library. When you need volume, these are the places the results page points to. If you design web software rather than phone apps, the same format is used in ten annotated SaaS product screens.
- Mobbin: screens from shipped iOS and Android apps, grouped by category and collection; it asks you to log in after the first rows.
- Dribbble: designer concept shots, many never shipped. Good for visual direction, weak as proof that a pattern works with real data.
- Muzli: a curated roundup of concept shots, often shown as several phones in one image, each with only a title.
- Figma Community: free mobile app templates and UI kits you can duplicate to see how a file is built. This article offers no templates.
- Pinterest: boards of saved shots collected by other designers, useful for a mood board.
Real shipped apps and concept shots answer different questions. A concept shot shows a direction with perfect data: short names, full signal, no error. A shipped app shows the compromises a team made for long translations, empty states, large text and lost connections. For product decisions, prefer shipped screens, and save each one with a one-line note: the person, the question and the one thing it does that your screen does not. A folder of unannotated screenshots turns into a mood board within a month.
Frequently Asked Questions
Where can I find free mobile app design examples and templates?
Mobbin shows some real app screens before asking you to log in, and Dribbble, Muzli and Pinterest are free to browse. For editable files, Figma Community has free mobile templates and UI kits you can duplicate into your own workspace. Treat a template as a starting structure, then redraw it for your own users and states.
How do you use mobile app design examples without copying them?
Write down the person and the question the screen serves, then the decision that answers it, such as “the arrival window is the headline” or “ending a run takes a hold.” Apply that decision to your own content, brand and platform. If you only take the palette and layout, you copy the look and miss the reason it works.
What are good mobile app design inspiration websites?
Use Mobbin when you need real shipped flows, Dribbble and Muzli when you want visual direction, and Pinterest to collect a mood board. Check every concept shot against the five checks above before you treat it as a pattern.
What mobile app design styles are common in 2026?
Dark and light themes are both common, and many apps support both. The ten examples here range from a near-black running app with lime numbers to a cream food app with rounded type and a high-contrast driver app in white and safety orange. Pick a style for the conditions of use, sunlight, night, glancing in motion, before you pick it for trend.
Conclusion
The ten screens above look nothing alike, and that is the point. What they share is a habit: name the person and the moment, put the answer first, show the change where the user already is, and keep the main action under the thumb. Read any example, from a gallery or from your own team, with the five checks: interruptions, platform navigation, glanceability, real conditions and thumb reach.
Use the at-a-glance table as a checklist for your own app. Write down the person and the question for each key screen, then look at the screen for one second and see whether it answers that question. Where it does not, the examples above show a way to fix it.