UX Research · Case study

Smart Wake on Pixel Watch: Early-stage Diary Study

Ran an early-stage diary study to test whether Smart Wake on Pixel Watch helps people wake up more refreshed. The findings led the team to narrow Smart Wake's wake-up window and refocus it on a more targeted user group.

Role

UX Researcher

Partners

PM, TPM, software & test engineers

Timeline

[Add timeline]

Method

Diary study

Context & Problem

Smart Wake already worked on Fitbit. On Pixel Watch, would it perform the same, and would its users want it?

Smart Wake wakes people during light sleep, up to 30 minutes before their alarm, so they wake up feeling more refreshed. It was an existing Fitbit feature, now moving to Pixel Watch.

However, the form factor changed and Pixel Watch brought a new group of users. The team didn't know if Smart Wake would still make waking up easier on the new device, or whether these users would want it at all.

How did I choose the methodology?

From research questions, to the data needed, to the method

1. What are the research questions

RQ1 · Feeling

How do users feel when they wake up with Smart Wake?

RQ2 · Behavior

How does Smart Wake change the way users get out of bed, such as snoozing?

2. What kind of data needed to answer the questions

Subjective perception

How do users feel when they wake up with Smart Wake?

Qualitative context

Why do users feel this way with Smart Wake?

Objective behavior

How do users actually wake up, e.g. snoozes?

Comparative baseline

How does Smart Wake compare with a standard alarm?

3. Which methods to use to get the data

Candidate 1

Log analysis

Gains: large sample, objective

Trade-off: no feelings or why

Candidate 2

Survey

Gains: large sample, subjective

Trade-off: recall bias about grogginess days later

Candidate 3

✓ Chosen

Diary study

Gains: in-the-moment feelings, real-life context, less recall bias

Trade-offs: participant burden, self-reported

Methodology & metrics

A diary study, chosen to capture how people feel the moment they wake up, and why

Participants

30 users

Already wore a watch to bed, so new habits wouldn't skew results

Design

4 nights

Smart Wake compared with a normal alarm

Measures

Alertness & snoozes

Sleepiness scale (KSS), picked over vigilance tasks because it is quick to answer half-awake

study Process

Sleep with the watch

Wear the watch to bed, with Smart Wake on some nights and a normal alarm on others.

Log each morning

Rate alertness, count snoozes and describe how waking up felt.

Compare conditions

Smart Wake mornings against normal-alarm mornings.

Exit survey

Overall satisfaction and open feedback on Smart Wake.

Daily morning log

KSS alertness

Snooze count

Sentiment

End of study

CSAT

Qualitative feedback

Key Insights

Smart Wake felt no different from a normal alarm, and trust was the bigger barrier

Alertness & snoozing · RQ1, RQ2

No difference

in alertness or snooze frequency compared with a normal alarm.

Held across snooze habits, wake-up status and alarm device.

Satisfaction

3.6 / 5

average CSAT, and about 2 in 5 felt no difference from a normal alarm. Fans said it eased getting out of bed and helped them feel ready.

Qualitative feedback · RQ1

Felt sleep loss & insecurity

Both psychological: the 30 minutes felt taken away, and unpredictable timing didn't feel safe.

“I prefer to have an alarm at a specific time since my hours of sleep is more predictable.”

Impacts

Drove the decision to narrow Smart Wake's window, and helped PM shift product strategy toward a more targeted user group

I recommended

What changed

Narrow the early wake-up window

Early wake-ups read as lost sleep.

→

Algorithm change

Fewer very early wake-ups, which lowers the mental barrier to trying Smart Wake and helps users adopt it.

Re-test at a larger scale

Longer period, objective alertness measure.

→

Next study

Moved to dogfood with a larger audience after the algorithm change.

Define who Smart Wake is for

Clarify the target user profile.

→

Product direction

Proposed tying Smart Wake to wider sleep context, such as sleep debt, instead of a standalone alarm.