Getting a person involved

The clearest finding in the self-help literature is that guided beats unguided.

Cuijpers and colleagues found guided self-help produces effects roughly comparable to face-to-face therapy for depression and anxiety (a post-test difference of about d = −0.02), while unguided self-help is reliably less effective — self-guided treatment for depression showed only a small effect, around d = 0.28 against control. The practical translation is not "stop using this site". It is: get even light human support around it — a therapist doing guided self-help, an NHS Talking Therapies practitioner, or a coach.

In the UK you can self-refer to NHS Talking Therapies without a GP referral.

Step up promptly if any of these are true

• Your symptoms are severe or getting in the way of work, relationships or daily tasks.
• GAD-7 is staying at 15 or above.
• PHQ-9 is staying at 10 or above, or there is persistent low mood or loss of interest.
• You are using alcohol or other substances to cope.
• Physical symptoms are frightening you.
• You have plateaued or worsened despite about eight weeks of consistent effort.
• Any thoughts of suicide or self-harm — treat that as urgent, not as something to work through here.

On medication

MBSR was statistically non-inferior to escitalopram in the TAME trial, so medication is not obligatory for mild-to-moderate anxiety. But it is an effective, guideline-endorsed option — particularly if symptoms are moderate-to-severe, if there is depression alongside, or if psychological approaches stall. That is a conversation for a GP or psychiatrist, not for a website.

Digital options with real trial evidence

• Unwinding Anxiety — RCT showed 67% vs. 14% GAD-7 reduction against usual care at two months (n=65). Developer-run study; commercial app.
• SilverCloud / Space from… — used within NHS Talking Therapies.
• Daylight — digital CBT for GAD; feasibility RCT.
• Woebot — RCT showed reduced depression symptoms in young adults vs. an information control.

Treat single-group and developer-run studies cautiously — that includes most of this list.