Digital CBT-I: Can an App Fix Your Insomnia?

11:32 pm. I am lying in bed with my phone, which is not what CBT-I recommends, but the phone is showing me my sleep diary data and calculating my sleep efficiency and telling me to get out of bed and come back when I am sleepy. The therapist is an algorithm. The algorithm is not wrong.

This is digital CBT-I, and it exists because there are not enough human therapists to treat the number of people who need this treatment.

The access problem

CBT-I is the recommended first-line treatment for chronic insomnia. Both the American Academy of Sleep Medicine and the American College of Physicians say so in their clinical practice guidelines. Approximately 10% of adults meet diagnostic criteria for chronic insomnia disorder. The number of clinicians trained and certified in CBT-I is a small fraction of what that prevalence requires. Most primary care physicians do not deliver CBT-I. Many do not know what it is. Many have never heard the abbreviation.

The result is a treatment gap: the evidence says CBT-I should be offered first, but most people with insomnia are offered medication first because it is available and CBT-I is not. A prescription takes five minutes. A CBT-I referral requires a specialist who may not exist within a hundred miles and who, if they do exist, may have a wait list measured in months. Digital programs are one attempt to close that gap, and the evidence suggests they close it meaningfully, if imperfectly.

I tried to see a CBT-I therapist before I tried a digital program. The nearest one was a two-hour drive. The next one had a three-month wait list. The program on my phone was available that night. Imperfect access beats no access.

What the programs deliver

Digital CBT-I programs deliver the same core components as in-person therapy: sleep restriction (with automated calculation of sleep windows from diary data), stimulus control instructions, cognitive restructuring exercises, sleep hygiene education, and relaxation training. The delivery happens through web platforms or mobile apps, typically over six to eight weeks, with new content unlocked as the program progresses.

The better programs include interactive sleep diaries that calculate sleep efficiency automatically, personalized sleep window recommendations that adjust weekly based on diary data, and educational content about why each instruction works. Some include an animated therapist or brief video sessions. None include the real-time judgment and flexibility of a human clinician.

What they do include is structure. Left to my own devices with a self-help book, I adjusted my sleep window when I felt like it, skipped the cognitive exercises because they felt abstract, and abandoned stimulus control after two nights because getting up felt pointless. The app gave me the same instructions but made them sequential and tracked my compliance. It calculated my sleep efficiency before I could fudge it. It told me when to expand my window based on the data rather than my impatience. Structure turns out to matter when you are sleep-deprived and your judgment is compromised, which is to say, always, when you have insomnia.

What the evidence says

Zachariae et al.’s meta-analysis of 11 randomized controlled trials found that digital CBT-I produced significant improvements in insomnia severity (measured by the Insomnia Severity Index), sleep efficiency, subjective sleep quality, and sleep onset latency [1]. The effects were clinically meaningful, not just statistically significant. The improvements crossed the threshold where someone would notice a difference in their daily functioning.

Espie et al.’s trial of the Sleepio program randomized 164 participants to digital CBT-I, placebo, or treatment as usual [2]. The digital CBT-I group showed large improvements in sleep efficiency and sleep onset latency compared to both control groups. Effects were maintained at follow-up, suggesting the behavioral changes persisted after the program ended.

Ritterband et al. tested SHUTi (Sleep Healthy Using the Internet) and found significant improvements in insomnia severity, wake after sleep onset, and sleep efficiency compared to a waitlist control [3]. The program had been developed at the University of Virginia and was one of the first rigorously tested digital platforms.

The Seyffert et al. meta-analysis confirmed these effects across multiple platforms and examined what moderated treatment response [5]. The overall direction was consistent: digital CBT-I programs produce meaningful improvements in insomnia, and the effects are durable.

The gap between digital and in-person

The evidence suggests that face-to-face CBT-I produces larger effect sizes than digital delivery. This is not surprising. A human therapist does things an algorithm cannot: reads hesitation, adjusts pacing, addresses the specific fear that surfaced this week rather than the generic fear the program anticipated, and provides the kind of accountability that comes from another person knowing whether you followed through.

But the comparison that matters for most people is not digital versus in-person. It is digital versus nothing. A program that produces moderate effects and is available tonight is more useful than a therapist who produces large effects and is available in three months. The meta-analytic data supports digital CBT-I as effective; the access data says it may be the only viable option for millions of people.

What works less well

The limitations are real. Dropout rates in digital CBT-I programs run around 25 to 30%, which is higher than in-person therapy [5]. The early weeks of sleep restriction are difficult, and without a therapist to encourage persistence, some people quit before the treatment takes effect. Sleep restriction feels terrible during the first week: you are more tired, not less, and the program’s reassurance that this is temporary and therapeutic feels less convincing from an app than from a person who has guided fifty patients through the same miserable week.

Programs cannot screen for other sleep disorders. Someone with undiagnosed sleep apnea needs a sleep study and possibly CPAP, not sleep restriction. A therapist would recognize the red flags (loud snoring, observed apneas, excessive daytime sleepiness despite adequate sleep opportunity). An app cannot. If you snore heavily, if a partner has observed you stop breathing during sleep, or if you are excessively sleepy during the day despite spending enough time in bed, see a doctor before starting any behavioral program.

Programs cannot adjust in real-time. A human therapist modifies the sleep window based on clinical judgment, adjusts the pace of treatment for someone with high anxiety, and responds to the specific concerns that emerge week by week. An algorithm follows a protocol. For uncomplicated chronic insomnia, the protocol is usually sufficient. For complex presentations, co-occurring depression, PTSD, chronic pain, or medication interactions, it may not be.

How to choose

The CBT-I Coach app, developed by the Department of Veterans Affairs, is free and available on iOS and Android [4]. It was designed as a companion to therapist-delivered CBT-I, providing sleep diary tracking and educational content between sessions. It can also be used independently, though the evidence base is strongest for the companion use case.

Programs with stronger standalone evidence include Sleepio and SHUTi, though availability and pricing vary. When choosing a program, look for: structured sleep diary tracking with automatic sleep efficiency calculation, personalized sleep restriction windows, a multi-week format (not a single session or a collection of tips), and published peer-reviewed evidence for that specific platform. A program that cannot cite its own trial data is making claims it has not tested.

What makes this hard

The hard part is the same as in-person CBT-I: following through. Sleep restriction is difficult. Getting out of bed when you are awake is counterintuitive. Sticking with a program for six to eight weeks requires motivation that insomnia itself undermines. You are asking someone who cannot sleep to commit to a program that will make them more tired before it makes them better rested. That is a hard sell at 2am.

An app cannot call you when you miss a session. It cannot hear the discouragement in your voice and adjust its approach. It cannot tell you that the first week is supposed to feel this bad and that it gets better, in a way that feels personal rather than generic.

What it can do is be available at 3am, when no therapist is. It can track your data and show you the trends that are hard to see from inside a bad night. And it can deliver the same evidence-based protocol that would otherwise require months on a waiting list or a specialist who does not exist in your town.

For me, the app was a starting point. I used it for three months, improved enough to function, and eventually saw a therapist who refined what the app had started. The therapist caught things the app could not: a subtle avoidance behavior I had not recognized, a cognitive distortion I had rationalized rather than challenged, a sleep window adjustment I had been making too aggressively. For many people, the app may be sufficient on its own. Either way, it is better than the alternative most people currently have, which is nothing.

If your sleep is affecting daily life, talking to a GP or sleep specialist is a good first step. The American Academy of Sleep Medicine has a provider finder at sleepeducation.org. If insomnia is accompanied by depression, anxiety or other concerns, those are worth raising with your doctor as part of the same conversation.