Senior PM Assignment · Product Concept

The Quiet
Window

Track the monthly financial rhythm, find the calm after every bill clears, and act at the right moment.

⚡ AI Engagement Platform 💰 Stability Ladder · Step 2 · Build a Cushion 🏦 B2B2C · Fintech · AI June 2026
product concept · ai-driven banking
The Core Insight

Why generic saving nudges fail

The problem
They fire at the wrong moment.
Every bank app says "save more" — right before rent, right after a loan clears, when the account looks full but isn't. Users feel nagged, ignore it, lose trust.
The insight
Everyone has a quiet window.
A 5–10 day stretch that opens after all fixed obligations clear and before new commitments begin. The only moment a saving suggestion is accurate and safe.
🗄️ Data source
Bank's existing transaction history. No cold start. The system reads the rhythm on day one, not via user input.
🎯 Target segment
Ages 25–45, regular salary, surplus ₪300–₪1,500. The "almost saving" cohort: intent every month, execution breaks down.
⚡ Trigger
Calendar-based: all obligations cleared + balance above comfort floor
📅 Cadence
Once per pay cycle, at the user's window. Monthly in IL/EU, bi-weekly in US.
The Core Insight

A typical monthly financial rhythm

💸 Payday
Salary in
Account looks full. But fixed obligations haven't cleared yet — the number isn't real.
🏠 Days 1–3
Rent & loans
Mortgage & rent standing orders fire on day 1. ~85% of loan installments clear by day 3.
💳 Day 10
Card bill
Cal, Max & Isracard all debit on the 10th. Last major obligation of the month.
🪟 Day 10+
The Quiet
Window
Last obligation settled. Window open. Notification fires within 24–48 hours.
"I always plan to save at the end of the month, but nothing's left by then. I know I should. I just never do."
Why this moment matters
Last recurring debit clears. Next salary date is known. The ₪300 buffer floor handles unknown one-offs. No forward scan needed.
The Intelligence

How the system learns the rhythm

The one capability: detect the precise day all recurring obligations clear, calculate the verified surplus after reserving a comfort buffer, and fire one notification per cycle at that moment.
01
Map obligations
Detect every fixed debit: rent, loans, standing orders, card dates.
02
Find the clearing date
Across 3 months, find the day the final obligation clears. That marks the window's start.
03
Calculate true surplus
Balance minus remaining debits. Card spend from this month won't debit until next month's bill.
04
Classify the state
Surplus above floor → save. Near zero → tight month. Recurring negative → Know Your Money.
📐 Rules
Obligation threshold · quiet-window date · comfort floor · notification cooldown (14-day default).
🤖 AI
Obligation classification · plain-language explanations · structural vs. one-time deficit detection. Why not rules? Binary thresholds misclassify variable-income users. ML handles trend analysis against shifting baselines.
Feature Design

Three states, three responses

🧭 Design principle
Every user who reaches the quiet window falls into one state, and none of them get silence. "Guidance before advice" applies to all three.
State 3 is not an exit. Deficit users route to Know Your Money, fix the gap, and return to Build a Cushion when the window reopens. The loop is what makes this a system.
State 1 · Surplus
Surplus detected
True surplus > comfort floor
→ ✅ Save-now notification
State 2 · Tight month
One-time deficit
Unusual expense this month
→ 🤝 Reassure + project forward
State 3 · Structural
Recurring negative
Income doesn't cover obligations
→ 🔍 Know Your Money
Who We're Building For

Three users, three realities

State 1 · Surplus
Shira, 32
Tel Aviv · Accountant · ₪14,000/mo net
"If someone just told me now is the moment, I'd probably do it."
  • Intends to save every month, never executes
  • Doesn't trust the running balance ("is this really free?")
  • Calculates too long, then spends it on something else
What it does
Catches her in the window, shows the verified free amount, one-tap transfer
State 2 · Tight Month
Tal, 29
Haifa · Developer · ₪22,000/mo net
"If an app tells me to save when I obviously can't, I delete it."
  • Needs context, not a number. Raw balance feels alarming.
  • High notification fatigue; generic notifications break trust
  • Knows why he's short. Needs the app to know it too.
What it does
Skips the notification, names the anomaly, books the next window. No pressure.
State 3 · Structural Deficit
Dana, 26
Rishon LeZion · Social media · ₪9,500/mo net
"I feel like I'm always almost okay, but never actually okay."
  • Doesn't know her real obligation total. Installments are invisible.
  • "Save more" advice feels impossible and shaming
  • On Step 2 before the Step 1 foundation is solid
What it does
Does not send a save notification. Routes to Step 1 instead.
State 1 · Surplus

All bills cleared. Save now.

What happened

All obligations cleared. Balance confirmed after the ₪300 buffer is reserved. Real, uncommitted money for the first time this month.

Happy path
Window detected
→
Surplus confirmed
→
Notification appears
→
User saves
→
Ladder updated
Alternate path: "Not now"
Notification appears
→
Not now
→
14-day silence
→
Passive re-surface
→
2nd dismiss → suppress & recalibrate
🛡️ Key safety check
The window opens only when every entry in the obligations calendar has cleared. The ₪300 buffer floor is always reserved. No forward scan: the calendar is the safety net.
Save Money10:14
Your quiet window is here 👋
Your moment to save
All bills cleared. You have ₪740 free.
Rent ✓ · Loan ✓ · Card ✓. No large expenses expected.
Cushion ₪930 / ₪3,000 goal
Save ₪350 now →
Change amount · Not now
State 2 · Tight Month

Near zero. Unusual this month.

What happened

An unusual expense ate the usual surplus. Flagged as a one-off, not a structural problem.

Flow
Surplus near zero
→
Check 3-month history
→
One-time anomaly
→
Tight-month card
🤝 Key principle: no silence
A user whose tight month goes unacknowledged feels invisible. The card builds trust by showing the app noticed, even when it can't help. Honest, calm, no pressure.
Save Money10:14
A check-in from your bank 👋
Tight month
Nothing to save this month, and that's okay.
Your car repair (₪1,200) made this month unusually tight. Your last 3 months averaged ₪580 free at this point.
📅 Next quiet window ~Jan 24. We'll check back then.
Got it ✓
See full month breakdown
State 3 · Structural Deficit

Income doesn't cover obligations

What happened

Negative at the window for 2+ consecutive months. Obligations exceed income. Saving is off the table until the imbalance is addressed.

Flow
Window negative
→
2nd month running
→
Reality card
→
Know Your Money
💡 Why Know Your Money, not an advisor
Routes to Step 1 for self-discovery: review spending, spot the leak, understand the pattern. Fix the gap and they return to Step 2.
Save Money10:14
An important update 👋
Financial reality
Your obligations exceed your income.
Monthly income  ₪7,200
Fixed obligations  ₪5,900
Variable spend  ₪1,800
Gap  −₪500 / month
Loan = 28% of income. Your biggest pressure point.
Review in Know Your Money →
Dismiss · See full breakdown
Design Principle

The Step 1 → Step 2
feedback loop

Ladder · Step 1
Know Your Money
Understand the pattern. Find where the money goes.
→surplus opens
Ladder · Step 2
Build a Cushion
Quiet Window fires. Save at the right moment.
→deficit detected
Back to Step 1
Diagnose & fix
Find the leak. Reopen the window.
💡 Why this matters
The deficit state is the bridge between steps. Users who can't save yet are coached to discover why. Guidance before advice, end to end.
Platform differentiator
No other solution handles the deficit case this way. A compelling angle for enterprise bank sales.
Business Case

Value for all three stakeholders

👤
Bank Customer
End user
Surplus month
Notification fires at the right moment. A growing cushion with visible ladder progress.
Deficit states
Reassurance in a tight month. Honest diagnosis when needed. Feels the bank is on their side.
Key KPI
Cushion delta · Feature NPS · ladder progression
🏦
Bank
B2B client
Surplus month
Deposit growth attributable to the feature. Savings activation and AUM increase per cohort.
Deficit states
Deficit message keeps users engaged. Spending corrections reopen future windows.
Key KPI
AUM growth · account openings · re-engagement · churn
⚙️
The Platform
AI Vendor
Surplus month
Clearest ROI story. AUM delta is verifiable and attributable per cohort.
Deficit states
Step-1 loop proves this is a system, not a feature. Deeper engagement, stronger case study.
Key KPI
Engagement · renewal rate · attributed ROI
B2B2C Design

What banks configure

⚖️Thresholds
Comfort buffer and save trigger. Defaults from platform surveys in the bank's country. Adjustable per segment.
🔀States
Which states to deploy. A conservative bank may disable State 3 entirely and surface only surplus notifications.
🎯CTA destination
Which savings product the notification routes to. Bank-owned account or a generic savings drawer.
📅Data window
Minimum months of history to detect a pattern. Default 3. Reducible with open-banking data.
🔔Cadence
Max notification frequency. Banks can extend the post-notification silence beyond the 14-day default.
✍️Copy tone
Formal or conversational, within the platform's AI guardrail templates.
🔐
Data scope & consent Feature is enabled at the bank level. User consent is covered by the bank's existing transaction data agreement — no additional per-user opt-in required. Banks control which customer segments are enrolled and can limit data scope to the transaction fields needed for obligation detection.
🌍 Market cadence
🇮🇱
Israel
Monthly · Day-10 card · 1 window ~18d
🇪🇺
Europe
Monthly · Variable card · 1 window ~8d
🇺🇸
United States
Bi-weekly · Variable obligations · 2 windows ~3–5d
Global Markets

One feature, three payment rhythms

🇮🇱
Israel
Monthly · 30 days
Rent
Card
🪟 Quiet Window · ~18 days
Day 1 · SalaryDay 10 · CardDay 11+Day 30
~18d
1 window/mo
🇪🇺
Europe
Monthly · DE/NL/Nordics
Standing orders
Card
🪟 ~8–10 days
Day 1 · SalaryDay 15 · CardDay 20+Day 30
~8d
1 window/mo
🇺🇸
United States
Bi-weekly · 2 pay cycles
Rent
Card?
🪟
Card?
🪟
Pay 1Day 7Pay 2 · Day 14Day 21Day 28
3–5d
×2 cycles
Salary / payday
Fixed obligations
Card debit
Quiet Window
v1 scope: IL + EU — monthly salary cycle, rule-based obligation calendar, reliable window. US: Phase 2 — bi-weekly cycles and variable obligation dates require ML-first detection. Scoped out of v1 to validate the IL calendar model before adding complexity.
Validation & Success

Three success metrics

Metric 01 · Primary
>20%
📈 Savings conversion, State 1
State 1 users saving within 48h of the notification. Generic nudge baseline ~7%. The 2.5x target reflects timing precision, not copy.
Metric 02 · Safety
<5%
🛡️ Post-notification overdraft rate
State 1 users who overdraft within 10 days. Rises above 5%: recalibrate obligation calendar and buffer floor before wider rollout.
Metric 03 · Trust
>30%
💬 30-day return after State 2/3
State 2/3 users who reopen within 30 days. Rising rate confirms honesty builds trust, not churn.
⚠ Worst-case failure
State 1 fires, user saves, user overdrafts within 10 days. Mitigation: Survey-calibrated comfort buffer (never touched) + obligation calendar must fully clear before any notification fires. Metric 02 is the early-warning signal.
3-step pre-build validation
1. Algorithm on real data: Run window detection on anonymised bank transactions. Target >60% detectable. Fail fast if the pattern isn't there.
2. Concept test, 5 users: Show 3 state cards as static mocks. "Does this moment feel right? Would you trust this number?"
3. Baseline existing harm: Pull overdraft rate after any bank nudge within 10 days. Know the number before claiming to fix it.
Execution

From idea to scale

Wk 1–2🔍 Discovery
8–10 user interviews. Run window-detection algorithm on bank transactions to confirm the pattern exists before designing anything.
PM · UX Research · Data · Bank
Wk 3–4✏️ Define
Concept-test 3 state cards (5 users). Define threshold rules with Data Science. Bank legal review of all state templates, especially State 3 deficit language.
PM · Design · Data · Bank Legal
Wk 5–10🏗️ Build
Obligation calendar, window algorithm, 3 state classifiers, AI copy generation, A/B flag. Bank data integration confirmed before build starts.
Engineering · Data · PM · Bank Tech
Wk 11–15🚀 Pilot
Bank UAT + sign-off. Soft launch: 10,000 users, 50/50 A/B. Iterate thresholds and copy. Gate: savings product and enrollment confirmed before first notification fires.
PM · Data · CS · Bank
Wk 16+📡 Scale
Roll out with config playbook (buffer floor, states enabled, CTA destination). CSM training on all 3 states, the loop, and bank parameters.
PM · Sales · CS · Bank
LIVE PROTOTYPE

Try the
Quiet Window

The Quiet Window runs inside the bank's native mobile app. Users see their bank, powered by the platform behind the scenes.

🛡️
Saves only when you actually can
Fires only when your obligations calendar is fully clear and your buffer is intact. Nothing fires if it isn't.
🪟
One notification per month
Fires once, at the Quiet Window, when the obligations calendar says the moment is real.
🔄
Goal-aware & adjustable
Users set what they're saving for, then control the pace with a live slider.
09:41
M
MY BANK
⚙️🔔
Hi Shira 👋🔔
Checking · Available balance
₪1,540
Updated 09:38 · All obligations cleared ✓
Build a Cushion Now
🪟 Your Quiet Window is open
All obligations cleared. 8 days to save when the number is real.
Available
₪1,540
Safety
✓ Clear
Closes
8 days
🔁
Transfer
🧾
Pay bill
📊
Insights
🕘
History
STEP 1 OF 5
What we found this month
✅
Surplus
14% above your 3-month average
₪1,540
🪟
Window open
Last obligation cleared Jan 11
8 days
🛡️
Buffer reserved
Always kept, never counted
₪300
Available to save
Up to ₪1,240 this month (₪1,540 − ₪300 buffer). You decide how much.
STEP 2 OF 5
What are you saving for?
or type your own
STEP 3 OF 5 · EMERGENCY FUND
How much do you need?
₪
Quick picks
STEP 4 OF 5 · EMERGENCY FUND
Which plan works for you?
STEP 5 OF 5 · EMERGENCY FUND
How much each month?
Monthly amount ₪380
₪50₪1,240 max
31% of your ₪1,240 surplus
Balanced — comfortable, sustainable pace.
Time to goal
8 months
By February 2027
📅
🎉
You're saving!
₪380/mo toward Emergency Fund.
Goal · Emergency Fund
₪3,000
Expected completion · 8 months
Saves automatically each month after all bills clear and your balance is real. If the balance is too low, nothing happens. You can cancel anytime.
🏠
Home
💳
Cards
🧭
Save
⋯
More
Pre-Mortem

Four risks that could end this

T1🚨 Launch-blocking
Safety miss → user overdrafts
Calendar misses a non-recurring charge. State 1 fires. User saves. Overdraft 3 days later. One bank complaint pauses the rollout.
Auto-pause if >3% of notified users overdraft within 7 days. Buffer floor is first line of defense.
T2⚡ Fast-follow
Window undetectable for 30% of users
Irregular income, variable card dates. Algorithm can't find a stable window. Users see nothing for months and assume the feature is broken.
Validate detection rate on anonymised data before launch. Target: >70% of users.
T3⚡ Fast-follow
Push notification never reaches users
Push opt-in ~45%, dismiss rate 60%. Effective reach: ~27%. The feature looks like it failed when the channel failed.
In-app home card as primary surface, push as accelerant. Both confirmed before launch.
T4👁 Track
Conversion at 8%, not 20%
Timing was right. User opened the notification. Didn't save. "I wasn't sure if that number was really safe." Safety signal didn't land.
A/B test explicit confirmation: "Rent ✓ Loan ✓ Card ✓ — ₪1,240 is actually yours."
Pre-Mortem

The risks nobody named

🗄️ E1 · Data
Bank delivered daily batch data, not real-time feeds
Pilot spec said "transaction feed." What arrived was a nightly batch with 24–36h lag. The window fired on yesterday's data — hidden cause of both overdrafts and low conversion.
Each bank configures a post-clearing delay to match feed latency. Window timing stays accurate regardless of batch lag.
⚖️ E2 · Legal
Legal blocked State 3 mid-build — nobody updated the business case
Legal removed the deficit card during build. The Step 1↔2 loop shipped without its defining piece. The commercial team didn't know until after launch.
Legal review of all state copy at design phase. Two approved variants agreed before build starts.
📦 E3 · Product
The bank's savings product had a 1.2% rate and killed conversion
Feature fired at the right moment. Users opened it, saw 1.2%, didn't save. The feature worked. The product it pointed to didn't — nobody locked the destination in the spec.
Savings product and enrollment steps locked in pilot spec before signing. Confirmed as a launch condition.
Bottom line
All three elephants are contractually preventable. A PM who owns the pilot spec owns the failure conditions before they become failures.
← → to navigate