Most therapists keep their real clinical notes somewhere other than the EHR — a Word document, a private file, a notebook — because the official note has stopped being a place they can actually think. The data backs the instinct. In a study of more than 23,000 clinician progress notes, only 18% of the text was typed by the clinician; 46% was copied and 36% was auto-imported.[1] Across a separate analysis of more than 100 million notes, about half the text in the record was duplicated from prior notes.[2] When a note is mostly copy-paste, it stops being the clinical truth and becomes a compliance artifact — so the clinician keeps a shadow system to hold the part that matters. The fix is not more discipline. It is a record built for clinical thinking, where the note is drafted from the session and the clinician edits and signs it.
I am a licensed clinical social worker, and I have kept the shadow document too. This is a piece about why it exists, what it quietly costs, and what a record that earns the clinician's trust would actually look like.
Why do therapists keep notes outside the EHR?
Ask a therapist where the real account of a case lives and a surprising number will tell you it is not in the chart. It is in a Word file, a personal note, a spreadsheet. The official note exists, it is signed, it is billable — and it is not where they go to remember what is actually happening with the client.
This is a rational response to a broken tool. When a system forces everything into structured fields, rewards copy-forward, and auto-populates half the page, the note becomes something to produce rather than something to read. So clinicians split the work: the EHR note for the payer and the auditor, and the shadow document for themselves. When a meaningful share of your providers are maintaining a parallel record because the official one is too rigid to think in, the official one has failed at its only job — being the place the clinical truth lives.
What is note bloat, and why does it happen?
Note bloat is what you get when copied, templated, and auto-imported text accumulates until the note is long, repetitive, and hard to use. The numbers are stark. The UCSF analysis of progress notes found just 18% of text was manually entered.[1] A larger study of over 100 million notes found 50.1% of the total text was duplicated from earlier notes about the same patient, and that the duplicated share climbed from roughly one-third in 2015 to 54.2% by 2020.[2] The American Medical Association, summarizing this work, put it plainly: about half the words in clinical notes are copied and pasted.[3]
It happens because the incentives point that way. The click trail is the audit trail, so the system rewards volume of documentation over clarity of it. Copy-forward is faster than thinking on the page, and the tool was built to make copy-forward easy. The predictable result is a record that grows longer every visit while saying less.
What the shadow system actually costs
The first cost is clinical. A note that is mostly duplicated text is a note you cannot trust at a glance. Copying or importing raises the risk of carrying forward stale or wrong information, and one analysis of diagnostic errors traced a share of them to copy-and-paste content.[4] When the record is unreliable, the clinician has to reconstruct the case from memory or from the shadow doc — which means the EHR is adding work, not removing it.
The second cost is the split itself. Maintaining two records is slower than maintaining one, and the good one — the shadow doc — is the one that is not backed up, not secured to the same standard, and not part of the legal record. The clinician is doing the documentation twice and getting the worse version into the chart.
The third cost is trust. A clinician who has stopped believing the official note is real has, in a quiet way, stopped believing the system. That is the state most therapists I talk to are in, and it is why "my EHR" is usually said with a sigh.
What a record built for thinking looks like
The alternative is not a better template. It is a different relationship between the clinician and the note. A record built for clinical thinking has three properties.
It is drafted, not assembled. The note comes from the actual session — a consented transcript, the clinician's own dictation — rather than copied forward from last week. The clinician edits and signs a draft that reflects what happened, not a page stitched together from prior notes. Used this way, AI is a scribe that collapses the writing time without inventing content, which is the use the evidence actually supports.[5]
It is readable. The point of a note is the next clinician moment — the pre-session glance that reminds you where the work is. A record that surfaces the signal instead of burying it under boilerplate is one the clinician will actually use, which means the shadow doc has no reason to exist.
It is the clinician's. The reasoning, the patterns, the throughline of a case live in the chart, under the clinician's control, not in a Word file on a desktop. One record, trusted, owned, and good enough to think in.
The call
So here is the call. The EHR that wins the independent practice is the one that kills the shadow document — not by forbidding it, but by being a better place to think than a blank page. The note will come from the session. The clinician will read it, fix it, and sign it. And the real account of the case will finally live in the chart, because there will be no reason to keep it anywhere else. That is the one I am building, and the shadow Word doc is exactly what I am building it to retire.