Penny — the same budgeting app, redesigned as a feedback intervention

Before/after home screens for the Build With Kris video "Your Dashboard Is a Feedback Intervention." Tap the numbered pins for the research behind each change.

Before 11 numbers, feedback about the worker Shame state, real-time noise, empty praise
9:41
Penny August
OVER BUDGET: –$63 😬
This month
In
$2,140
Out
$2,203
Balance
$1,274.87
Budget left
$312
Daily avg
$73.43
LIVE Starbucks –$6.40 · 2m ago
Dining
$412
Transport
$181
Fun
$208
Bills
$890
🎉 AMAZING! You logged expenses 3 days in a row!!
Vacation fund
4%
After 6 widgets, feedback about the work One anchor, a direction, a next action
9:41
Penny August
Safe to spend today
$42
$118 less than last week
Dining is $63 over. Move $50 from Fun budget?
Done — Dining balanced.
3 days logged — your spending picture is now accurate.
This week
$486
smoothed trend
$120 saved so far
Vacation fund
Goal: $500 by October
1The anchor: 11 numbers → 1 hero
One calm number replaces the before screen's wall of eleven. The best apps already do this; most dashboards don't.
Our audit (n=76 top-app screens, Aug 2026): 34% of consumer home screens lead with a single hero metric; only 7% show red/negative framing; 42% of SaaS dashboards show 9+ numbers above the fold
2The delta chip: velocity beats a snapshot
A static total says where you are. The chip says which way you're moving — the moderator with the biggest lift.
Velocity feedback d = .55 vs .28 without; feedback with the correct next step d = .43 (Kluger & DeNisi moderator table)
3Shame → next action
Same fact as the red banner — Dining is $63 over — reframed as a one-tap task about the work, not the worker.
More than 38% of 607 feedback effects were negative (Kluger & DeNisi 1996 meta-analysis, 131 papers)
4Confetti → competence info
The streak survives, but as information about accuracy instead of empty celebration — because praise barely moves behavior.
Praise: d = .09 — the near-zero row of the moderator table (Kluger & DeNisi 1996)
5Live ticker → weekly rollup
Real-time noise makes people chase every wiggle. A smoothed weekly line is slower — and works better.
Less frequent feedback beat constant feedback in noisy environments (Lurie & Swaminathan 2009)
A visible page counter cut voluntary continuation from 48.5% to 27.3% (Etkin 2016, Experiment 6)
6The framing flip
Try the toggle: the same 24% ring reads "$120 saved so far" for a new user and "$380 to go" for a committed one.
To-date framing motivates uncertain users; to-go motivates committed ones (Koo & Fishbach 2008)