The method
A plan, then practice that comes back before you forget.
Stica turns what you are trying to learn into a plan, then brings short practice back to you over the following weeks, checks what stuck, and gets in touch when you go quiet.
Eight steps. This page walks through all eight, in order, then shows the research each one rests on, graded honestly.
Free to start. No card, no clock.
How it works
The loop, start to finish
Eight steps. It writes the plan, brings the practice back, and nudges you when you go quiet. It never speaks to anyone else on your behalf and never acts in the world for you. Where a real commitment is needed, that step says so.
-
1
Where the plan comes from
A few questions
No card. Give an email, we send a sign-in link, and you are straight into the questions: what you want to change, why now, and where you will use it first.
-
2
Your plan
A frank read of your goal, the gaps behind it, and steps you could start today. Built from your answers, not picked off a shelf.
-
3
Your own material
Optional: paste what you already have and the plan is built around it. Upload your documents as well, and point your practice at them as you go.
Anything you paste goes to Anthropic's Claude to help write the plan, then comes straight back to you. Files you upload stay in our own storage and are not sent to any model. Full detail on how we handle both is on Stica.
-
4
What keeps coming back
Practice, spaced out
A commitment prompt lands the moment your plan does. After that, a small card arrives every few days, drawn from the researched map of your goal: something to react to, think about, try or recall, one at a time, never a pile. Skipping one is a real answer, and it steers what comes next. When the stretch you picked is up we check whether it has become habit, choice or effort, and if that says keep going, one more check follows a month later; miss a check and it stays open rather than disappearing. Go quiet for about ten days and we send one nudge; go quiet again and we ask whether this is still the right thing, rather than nudging a third time.
-
5
What checks whether it stuck
What you make of it
Open prompts, written from your own goal. Your answers are kept and you can read them back.
-
6
Your own read on it
On some checks, rate your confidence one to five. Stica draws the line so you can see whether it is moving.
-
7
What happens when it doesn't
A nudge when you go quiet
Ten days of silence on a topic, a thin answer, or a check falling due brings one note in the app, plus an email if you asked for one, and that same miss only ever gets the one nudge.
-
8
Where a person comes in
A coach, if you have one
This is where a commitment becomes real. On your own, you write your goal down and Stica practises you against it. With a coach, the session sets the commitment and the dates, and Stica surfaces each check-in as it falls due.
Where this loop needs a person
On your own, Stica plans, practises and nudges. Commitments and follow-up need a coach, because a commitment is something you make to a person.
Proof, not a promise
See the loop in a real plan
Not a screenshot from a free account, but a worked example of the same writing: the diagnosis, the gaps, the do-today steps and the practice arc. It's a demonstration, not a real client.
Illustrative example
See a plan worked through, end to end
An early worked example: one brief, taken through diagnosis, strategy and the first piece of content, built.
See the example plan →
Free to start. No card, no clock.
Why we built it this way
The evidence, graded honestly
None of the eight steps above are a guess. Here is the research behind them, and exactly how strong each piece is, so you can judge it yourself.
A grade is the strength of the evidence, not how often we repeat a claim. Where the research pools many independent studies, we treat it as load-bearing. Where it's a single strong study, practitioner convention, or an effect that shrinks under scrutiny, we still use it, held more loosely, and we say so. Nothing here is new science. What's different is living it end to end, and telling you exactly how strong each piece is.
A pooled, quantitative synthesis across many independent studies. The strongest evidence we use.
A well-used framework, a single strong study, or an effect that shrinks under scrutiny. We use it, held loosely.
The effect shrinks to near nothing once it's checked hardest.
Behind steps 4–6
Why practice beats delivery
These findings shape why every Stica touchpoint asks you to recall or use something, never to re-read it, and why the schedule is fixed by the research rather than by mood.
Recalling from memory beats re-reading
Testing yourself on material works better than reading it again: a pooled advantage of about g = 0.50, rising to g = 0.73 when the recall comes with feedback.
g measures how large an effect is: roughly 0.2 is small, 0.5 is worth building around, 0.8 is large. Where a bar appears beside a g or d figure on this page, the tick marks 0.5.
Most of the trials behind this are educational. Few run inside workplace learning, so we treat the direction as solid and the size as a reasonable estimate.
In Stica, this means
every prompt asks you to produce an answer, never to re-read one. That's steps 4 to 6 of the loop above.
Rowland, Psychological Bulletin, 2014.
Spaced practice beats cramming
Spreading practice out over time beats massing it together: about g = 0.28 in applied settings, higher again in the lab.
It only works if you actually come back. That's why the schedule in Stica runs itself, rather than waiting on willpower.
In Stica, this means
step 4, the spaced checks, is system-run on a fixed schedule, not left to whenever you remember, and you never set it yourself.
Cepeda, Pashler, Vul, Wixted & Rohrer, Psychological Bulletin, 2006.
Skills fade fast without use
Unused skills measurably fade within a month of training, and keep fading after that.
The decay curve for interpersonal skills is mapped less well than for technical ones. We say so, rather than borrow the stronger number.
In Stica, this means
every plan puts the first real application in its opening weeks, not left for “someday”.
Arthur, Bennett, Stanush & McNelly, Human Performance, 1998.
Planning the first use makes it happen
Deciding in advance (“when X happens, I'll do Y”) reliably helps people start something new.
The classic large effect has shrunk in later pooled analyses. It still earns its place, but for starting a behaviour more than for sustaining it, and we use it that way.
In Stica, this means
every plan anchors the first application to a situation you actually name. That's step 2 of the loop.
Gollwitzer & Sheeran, Advances in Experimental Social Psychology, 2006.
Behind steps 1–2
Why diagnosis comes before content
These findings shape why every plan starts by asking what kind of gap this actually is, and whether training is even the right lever.
Most behaviour problems aren't knowledge problems
The established diagnostic frameworks separate whether someone can do a thing from whether their environment lets them, and whether they want to.
The practitioner triage of “can't do” versus “won't do” has decades of field use and thin trial evidence. We apply it, and tell you when your problem looks like environment or motivation rather than a course.
In Stica, this means
step 2, your plan, is willing to say this isn't a training need.
Michie, van Stralen & West, Implementation Science, 2011.
The workplace often outweighs the training
Whether the workplace supports the new behaviour explains a real share of whether training sticks, roughly a third of it in the strongest pooled analysis (R² = .32), and supervisor support on its own is a moderate, reliable predictor (r = .31).
The main meta-analysis here is dated and leans on self-report, so we keep the claim modest. The direction is one of the most consistent findings in the field.
In Stica, this means
step 8 exists: plans engage a coach where one is available, not willpower on its own.
Blume, Ford, Baldwin & Huang, Journal of Management, 2010.
Wanting to use it predicts that you will
How much you actually want to use what you learned, what researchers call “motivation to transfer”, is among the strongest predictors that you will: a moderate, reliable correlation (r = .37). Framing material as personally useful adds a further, smaller lift (d = 0.49), though that evidence leans more on education than work.
Correlations and effect sizes here come from different kinds of studies, so we quote each on its own terms rather than force them onto one scale.
In Stica, this means
step 2's plan is built from your own goal and material, not a generic syllabus.
Colquitt, LePine & Noe, Journal of Applied Psychology, 2000.
Goals work best when you set them yourself
Specific, self-authored goals are one of the strongest individual levers in the literature.
Tight goals can hurt novel, complex tasks. That's why a Stica plan is drafted from your own answers, not assigned to you.
In Stica, this means
step 1 is guided capture and reflection, not a form you tick.
Locke & Latham, American Psychologist, 2002.
Where the evidence is thinner
What we hold loosely, and what we don't buy
Weaker evidence doesn't get quietly promoted. These are things we either use with a caveat attached, or don't use at all, labelled as what they are.
Habit timelines vary enormously
Habits typically take two to five months to form, with a documented range from 4 to 335 days.
This is a range, not a percentage scale: the spread is real variation between people and habits, not measurement error. There is no single "days to form a habit" figure we would put on a page. Our arcs end when the behaviour holds, not on a fixed date.
In Stica, this means
step 4's practice runs until the habit holds, however long that takes for you.
Singh et al., systematic review and meta-analysis, Healthcare, 2024.
Nudges alone can't carry a programme
Once publication bias is corrected, the average nudge effect is close to nothing: about d = 0.04 in the toughest pooled analysis.
We use prompts and defaults as amplifiers inside a designed programme, not as the intervention on their own.
In Stica, this means
step 7's nudge supports the plan. It never stands in for one, and it never fires twice for the same miss.
Maier, Bartoš, Stanley et al., PNAS, 2022 (correcting Mertens et al., PNAS, 2022).
AI-delivered learning is promising and unproven
Adaptive and tutoring systems show real pooled gains on knowledge outcomes. There's no pooled trial base yet for the newest AI tutors.
Engagement with digital behaviour-change tools decays without human structure, and AI assistance can lift output while leaving retention flat unless recall is designed in. Our answer is that design: the AI drafts and schedules, the practice makes you produce, and we treat claims beyond knowledge gain as hypotheses to measure.
In Stica, this means
steps 1 to 7 are the AI's job: plan, practise, nudge. Step 8, the commitment, still needs a person.
Kulik & Fletcher, Review of Educational Research, 2016 (older adaptive tutoring systems; not the newest AI tools specifically).
The same discipline, tested harder
The same fixes, tested hardest in medicine
Implementation science is the study of why known-good practice fails to spread. It has been tested hardest in medicine, where a lapse is a missed diagnosis rather than a missed deadline. A respected peer, one-to-one expert input, feedback on how you compare, even plain reminders: the same handful of fixes turn up again and again in that literature. They are the backbone of the loop above too.
| Approach | What it does | Effect | Confidence |
|---|---|---|---|
| Local opinion leaders | A respected peer champions the change | +10.8 pts | Solid, wide spread. Flodgren, Cochrane 2019. |
| Educational outreach | An expert visits one to one with the evidence | +4.8 pts | Strong for prescribing. O'Brien, Cochrane 2007. |
| Audit and feedback | Shows people how they compare to a standard | +4.3 pts, up to 16 pts when well designed | Strong. Ivers, Cochrane 2012/2025. |
| Printed materials alone | Handing out guidelines or leaflets | +4.0 pts, weakest alone | Low certainty alone. Giguère, Cochrane 2020. |
Figures are percentage-point improvements in the behaviour each study measured, not a shared 0–1 effect size, so no bar is drawn for them.
Two of these need a person. A respected peer and one-to-one expert input are what a coach adds at step 8, live once you have one. Audit and feedback needs no one else: it's step 6, your own confidence tracked over time. And Stica is built to avoid the weakest fix here, printed materials alone; that's what the loop above replaces.
How to read these numbers
No single approach reliably clears about ten percentage points of absolute improvement on its own. The value comes from stacking the right one at the right point, not from any one fix. The wide ranges are genuine context-dependence, not noise. Stacking more pieces on does not reliably do more either: pick the right one for the point you are stuck at.