SOAP notes: a practical guide for clinical psychologists

The SOAP format is one of the most widely used structures for documenting clinical sessions, and it was not born in psychotherapy: it comes from the problem-oriented medical record that Dr. Lawrence Weed proposed at the end of the 1960s. Its name comes from its four sections: Subjective, Objective, Assessment, and Plan. Used well, it produces notes that are clear, defensible, and genuinely useful for treatment continuity. This guide covers what belongs (and what does not) in each section, when to write the note, and how it fits into the clinical record your jurisdiction expects you to keep.

Why the note matters, beyond habit

Documentation is not bureaucracy that gets in the way of clinical work: it is clinical work. The APA Ethics Code (Standard 6.01) lists, among other purposes, what records exist for: so that you, or another professional, can continue the service later; to meet institutional requirements; to ensure accuracy of billing; and to comply with the law. The APA’s Record Keeping Guidelines — adopted in 2007 and still the most recent published edition — put it plainly: appropriate records help protect both the client and the psychologist if anything ever reaches a legal or ethical proceeding, and adequate records are generally a requirement for third-party reimbursement, insurers included. If you are ever asked what you did and why, your note is both your memory and your defense.

The four sections, in depth

S — Subjective

What the client reports in their own words: symptoms, concerns, relevant events since the last session, and verbatim quotes when they capture something clinically significant.

“Reports sleeping better this week, though nighttime rumination persists on Sundays. Mentions an argument with her partner she describes as ‘the same cycle as always.’”

What does not belong here: your conclusions. “Presents as resistant” is already your interpretation and belongs in the Assessment. Two boundary cases clinicians hit constantly:

Data that sounds objective but is reported. “I slept four hours” sounds measurable, but you did not measure it: it is the client’s report, so it goes in S.
What third parties say. If a mother, a partner, or a school provides information, it goes in S with the source made explicit (“mother reports that…”). Always attributing the source keeps someone else’s account from later reading as your own observation.

O — Objective

What you observe directly and measurably: appearance, affect, in-session behavior, and results of administered instruments.

“Arrives on time, well groomed. Congruent affect, broader emotional range than in previous sessions. PHQ-9: 8 (previous: 12).”

Standardized instruments are the best possible content for this section, because they turn “she seems better” into a number you can compare across sessions. The PHQ-9, for example, is free: the official site phqscreeners.com states that the instruments and their translations can be downloaded and reproduced with no permission required — including official Spanish versions, such as the Spanish for Mexico PHQ-9. Scores range from 0 to 27, and in the original validation study cutpoints of 5, 10, 15, and 20 mark mild, moderate, moderately severe, and severe symptoms. So the 12→8 in the vignette did not just “go down”: it crossed from moderate to mild, and that is a genuine data point about progress. Two cautions: if you work with Spanish-speaking clients, a 2023 meta-analysis found optimal cutoffs varying from ≥ 5 to ≥ 12 across Spanish-language samples; and no score is a diagnosis — these are screening and monitoring tools your clinical judgment integrates.

The classic boundary case in this section: “congruent affect” or “cried for ten minutes while discussing her father” are legitimate mental-status observations and belong in O; “she is sad” or “she is doing better” are inferences and belong in A.

A — Assessment

Your clinical interpretation: how the subjective and objective information integrates with the diagnosis, the prognosis, and the current treatment plan.

“Reduction in depressive symptomatology consistent with subjective report. The couple conflict remains the primary stressor and trigger of the rumination pattern.”

This is where a note written to check a box parts ways with a real clinical note. The typical failure is summarizing S and O instead of connecting them: the Assessment answers “what does this mean for the case?”, not “what happened today?”. It is also where sensitive reasoning lives: if a risk signal appeared in session — a positive PHQ-9 item 9, a mention of ideation — that signal calls for an immediate clinical risk assessment under your own protocol and training, and A is where you document that you did it, what you found, and what your decision rests on. A note that records the score but not the follow-up evaluation is exactly the kind of gap you cannot explain later.

And the reverse boundary: no new data appears in A. If something matters to your analysis, it has to exist first in S or O.

P — Plan

Concrete next steps: homework, techniques to work on, adjustments to the frame, next session date.

“Continue behavioral activation. Introduce thought records around arguments. Next session: August 12.”

Do not confuse the note’s Plan with the treatment plan: the treatment plan is the full map of the process, with goals spanning weeks or months; each note’s P is the next move within that map. If your notes say “continue treatment” session after session, the problem usually is not the note — it is that there is no treatment plan with verifiable goals behind it.

The litmus test for P: could you verify at the next session whether it happened? “Work on self-esteem” fails the test; “thought records around arguments, review in session” passes.

And if the full format feels like more structure than your style needs, there are leaner alternatives: compare DAP vs SOAP notes before you commit. And if your setting documents by intervention or by goal, there are BIRP notes and GIRP notes.

Common mistakes

Mixing subjective with objective. “The client is sad” is interpretation; “cried for 10 minutes while discussing her father” is observation.
An assessment that repeats instead of integrating. Section A doesn’t summarize the previous ones — it connects them to the course of treatment.
Vague plans. “Continue treatment” cannot be followed up on. Specify technique, homework, and timeframe.
Writing for no one. A good note can be picked up by another clinician (or by you in six months) and instantly show where the case stands.
Detail that helps no one. A note is not a transcript. Record what is clinically relevant; unnecessary intimate detail does nothing for continuity and enlarges what you would have to account for if the record is ever requested by an authority.
Cloned notes. Copying the previous note and changing the date produces a record that documents nothing: if March and June read identically, neither one supports that a process took place.

A broader tour — with the exact rule each mistake breaks — is in clinical documentation mistakes that put your practice at risk.

When to write it

The APA Record Keeping Guidelines’ rule is terse: entries should be made as soon as is practicable after the service. The reason is pure memory: a note written the same day records what happened; one written on Friday reconstructs it — and reconstructions tend to look suspiciously alike. In some places a per-session note is not just good practice but a normative requirement: in Mexico, for example, the clinical-record norm (NOM-004-SSA3-2012) requires a progress note every time outpatient care is provided. The habit that works best in practice is blocking the final minutes of each clinical hour for the note, rather than letting notes pile up at the end of the day.

How long to keep your records

Retention depends on where you practice, so treat what follows as orientation, not as your rule. As US guidance, the most recent published APA Record Keeping Guidelines suggest — explicitly “in the absence of a superseding requirement” — considering keeping full records until 7 years after the last date of service for adults, or until 3 years after a minor reaches the age of majority, whichever is later. Local law prevails where it exists: in Mexico, for example, the clinical-record norm sets a minimum of 5 years counted from the last act of care. Wherever you work, verify the rules of your jurisdiction and treat the legal period as a floor, not a target. And since everything in a note is sensitive health information, your informed consent paperwork should tell clients how records are stored, who can access them, and for how long they are kept.

How long it should take

There is no official standard for minutes per note, and you should distrust anyone who cites one. In our experience, a well-written SOAP note takes 10 to 15 minutes per session; at 25 clients a week, that pace adds up to four to six hours of documentation. It is exactly the kind of burden where AI is already proving useful: a 2025 systematic review across 23 studies found that AI tools significantly reduced clinical documentation time. gesell.ai applies that to your practice: it drafts the note from your session record — in SOAP, DAP, BIRP, or GIRP format — and you review and adjust every draft: the final note is yours. The clinical judgment is always yours; the typing time no longer is.