Self-monitoring apps for substance use: what to look for, what to avoid

When recommending a digital tool to a client, you are implicitly endorsing several things: the clinical rationale for the tool, the quality of its implementation, and its appropriateness for that client's specific situation. In a market where the majority of substance use apps are built around sobriety tracking, AA-adjacent framing, and recovery community features, a systematic approach to evaluation is warranted.

This is a framework for assessing digital self-monitoring tools for substance use — particularly for clients who are not in active treatment or who don't identify with recovery framing.

The population problem in the current market

Almost all widely available substance use apps are built for one specific population: people who have decided to stop using and are seeking support for that commitment. "I Am Sober," "Nomo," "Sober Grid," and similar tools are sobriety counters. They track abstinence streaks. They provide relapse prevention support. They connect users with recovery communities.

These are legitimate and useful tools for the population they serve.

The problem for clinicians is that this population represents a small fraction of people with problematic substance use. The majority of people drinking hazardously, or using cannabis or stimulants in ways that are affecting their functioning, have not identified as having a problem, have not decided to stop, and would not describe themselves as "in recovery." They are your precontemplative clients — and they are precisely the people for whom self-monitoring evidence shows the strongest independent effect.

There is essentially no overlap between what those clients need from a digital tool and what sobriety apps provide. Recommending a sobriety app to a precontemplative client is not just ineffective — it can actively damage engagement by communicating that the clinician has already decided what the outcome should be.

What to look for

Privacy architecture, not just a privacy policy

For many AOD clients, privacy is not an abstract concern. Employment consequences, family dynamics, legal proceedings, and insurance implications can all be affected by a documented record of substance use. A conventional app that requires name, email, and account creation stores identifiable data that could, in theory, be subject to legal disclosure.

The distinction worth making is between a privacy policy (a statement of intent about data use) and privacy architecture (a technical structure that makes identification impossible regardless of intent). An app that stores no identifying information — passphrase-only login, no name or email — is not making a promise about what it will do with your client's data; it is structurally unable to identify them.

For clients with legitimate disclosure concerns, this distinction matters.

All-substance coverage

Many digital tools are built around alcohol specifically, or designed primarily for a single substance type. AOD clients' presentations are rarely that clean. A tool that logs alcohol but not cannabis, or stimulants but not alcohol, produces an incomplete picture and implicitly hierarchises substances in ways that may not match your client's reality.

A clinical-grade self-monitoring tool should accommodate any substance the client uses, without implicit framing that some substances are more legitimate subjects of attention than others.

Validated clinical instruments

Self-report diary data is useful; contextualised, validated, scored data is more useful. A tool that incorporates AUDIT, ASSIST, DASS-21, K10, PHQ-9, or comparable instruments provides population-level context for individual data and supports clinical risk identification without requiring a separate assessment process.

Importantly, these instruments need to be correctly implemented. The scoring algorithms for validated instruments are precisely specified. A tool that has built its own variant of the AUDIT, or modified question wording, is not administering the AUDIT — it is administering something else, and the normative data that gives AUDIT scores their clinical meaning does not apply.

Low barrier to entry

Completion rates for between-session tasks are inversely related to their complexity. An app that requires a 20-minute sign-up process, account verification, and a lengthy onboarding before the client can log their first entry will have low uptake. For clients who are ambivalent about monitoring in the first place, friction is decisive.

The target specification is: a client should be able to register and complete their first entry within five minutes of receiving a recommendation. Anything more complex than this is clinically impractical for most precontemplative or early-stage clients.

Tone and framing

The language an app uses communicates its assumptions about the user. An app that refers to users as "in recovery," uses the language of sobriety and relapse, or frames monitoring as part of a commitment to abstinence will be rejected by clients who don't identify with that frame — often before they've given the monitoring any real opportunity to work.

The appropriate tone for a precontemplative population is curious and non-judgmental: "understand your patterns" rather than "track your sobriety." The difference is not cosmetic; it determines whether the tool is usable by the clients who most need it.

What to avoid

Social and community features

Recovery apps often include community forums, peer support channels, and social accountability features. For clients who do identify with recovery framing and want community support, these can be valuable. For precontemplative clients, they are a significant deterrent — and for clients with privacy concerns, they introduce risks that a clinician should not be implicitly endorsing.

A self-monitoring tool for clinical adjunct use does not need community features. Their presence should prompt the question of whether the tool is designed for a different population.

Abstinence-only metrics

A tool that measures success exclusively as days since last use is clinically inappropriate for harm reduction goals, moderation goals, or monitoring-without-commitment goals. Progress for many AOD clients is not abstinence — it is reduced consumption, changed patterns, improved mood and sleep, or simply clearer self-knowledge. A tool that can only represent progress as a sobriety streak is not fit for purpose in most AOD clinical contexts.

Data monetisation

Some free apps generate revenue through data licensing or advertising. For substance use data, this is an ethical concern. Review the privacy policy of any tool you recommend for explicit statements about data sale or advertising use. The absence of a clear statement that user data is never sold should be treated as a concern, not as neutral.

The clinical reporting question

One feature worth specifically evaluating is whether a tool supports structured client-to-clinician reporting, and how it is implemented.

The clinically appropriate model is client-controlled sharing: the client generates a report and chooses to send it to their clinician. This preserves autonomy, is consistent with MI principles, and avoids the dynamic where monitoring feels like surveillance.

Avoid tools where the clinician has direct access to client data without explicit client action. Beyond the ethical concerns, this framing changes how clients relate to the monitoring — from self-knowledge to accountability — in ways that reduce its therapeutic utility.

ayodee is designed for the clinical adjunct use case described in this article. Clinicians can purchase bulk access codes and provide a printable QR card for clients to scan and register. No client data is accessible to clinicians without explicit client action. No identifying information is stored.