Skip to content
Free lab

Step-Up Auth Lab

Good authentication is not a wall in front of every action — it is the right friction at the right moment. Work a queue of account actions and make the call: allow it silently, step up with a re-challenge, or deny and force out-of-band recovery. A new payee, a change to the MFA device / email / password / payout account, a privileged op, a new device or impossible travelearn a challenge. A big-but-recurring payment, a benign re-login and a verified own-account transfer do not — and a tiny payment to a brand-new payee still does. Challenge everything and you train users to click through the one prompt that mattered. Calibrate the queue.

The real job. More noise, a tighter clock.

0sPro

Authorization queue

For each action decide: allow silently, step up, or deny. Challenge where risk justifies the friction — not everywhere.

#1

View account balance

Read-only
deviceknown
locationAustin, TX — usual
riskScore6/100
activityNormal; no failed logins

user pattern: Checks balance most mornings from this device

Decision

#2

Add new payee / beneficiary

Beneficiary
payeeExternal bank account — never seen before
deviceknown
locationAustin, TX — usual
riskScore38/100
activityNormal session

user pattern: Has paid 4 saved payees for 2 years; this one is new

Decision

#3

Change account password

Security setting
settingPassword
deviceknown
locationAustin, TX — usual
riskScore20/100
activityInitiated from settings while logged in

user pattern: Routine security hygiene change

Decision

#4

Wire $45,000 to payroll vendor

Payment
amount$45,000
payeeRecurring — same payroll vendor paid monthly for 24 months
deviceknown
locationAustin, TX — usual
riskScore11/100
activityNormal; runs on the 1st every month

user pattern: Same vendor, same ~$45k, monthly for 2 years

Decision

#5

Send $4.20 to a new payee

Payment
amount$4.2
payeeBrand-new external payee, first payment
devicenew / unrecognised
locationKyiv, UA — new device & country
riskScore72/100
activityDevice first seen 3 minutes ago

user pattern: User is US-based and has never paid this payee

Decision

#6

Log in

Session
deviceknown
locationAustin, TX — usual
riskScore8/100
activityFirst login of the day; no prior failures

user pattern: Logs in from this device daily around this time

Decision

#7

Change registered email address

Security setting
settingRecovery email
deviceknown
locationAustin, TX — usual
riskScore24/100
activityNormal session

user pattern: Current email unchanged for years

Decision

#8

Download monthly statement (PDF)

Read-only
deviceknown
locationAustin, TX — usual
riskScore7/100
activityNormal session

user pattern: Downloads a statement most months

Decision

#9

Raise daily transfer limit to $100k

Security setting
settingDaily transfer limit
deviceknown
locationAustin, TX — usual
riskScore19/100
activityNormal session

user pattern: Limit untouched since account opening

Decision

#10

Wire $900 to a recurring payee

Payment
amount$900
payeeRecurring — landlord paid monthly for 18 months
devicenew / unrecognised
locationDenver, CO — new device, different city
riskScore58/100
activityDevice first seen today

user pattern: Always pays this payee from the usual Austin device

Decision

#11

Log in again

Session
deviceknown
locationAustin, TX — usual
riskScore9/100
activitySecond login 5 min after a successful one (app reopened)

user pattern: Normal: reopens the app several times a day

Decision

#12

Update notification & marketing preferences

Security setting
settingNotification / marketing preferences
deviceknown
locationAustin, TX — usual
riskScore12/100
activityNormal session

user pattern: Routine preference tweak

Decision

#13

Transfer $3,000 to a newly-added payee

Payment
amount$3,000
payeeThe user’s OWN linked account — same legal name, verified by micro-deposit last year
deviceknown
locationAustin, TX — usual
riskScore16/100
activityNormal session

user pattern: Regularly sweeps cash to this own-name account

Decision

#14

Add first-ever payee

Beneficiary
payeeFirst beneficiary on a 3-week-old account
deviceknown
locationAustin, TX — usual
riskScore41/100
activityNormal session

user pattern: New account; has never added a payee before

Decision

Right friction, right moment

Decide each action, then grade. So far 0/14 called, 0 under-challenged, 0 over-challenged.

How it is scored

Under-challenging (allowing a risky action, or merely challenging one that needed a hard deny) is the heavier penalty — it is the security failure.

Over-challenging (friction on a routine action) takes a lighter bite — a system that cries wolf trains users to click through the prompts that matter.

Harder tiers add subtler context and traps: a big recurring payment that should allow, a new payee that should still step up, and takeover-in-progress cases that should deny.

Live arena

4 active
adversary kill-chain1/4
authn-lead-wen16%
risk-eng-tomas13%
fraud-analyst-mara9%
authn-engineer-you (you)0%

adversary staging…

TL;DR

What this lab is

  • A free, browser-based step-up / risk-based authentication lab: work a queue of ~20 account actions and decide each one — allow silently, step up (re-challenge), or deny and route to out-of-band recovery.
  • The whole skill is balance: step up where risk justifies the friction (a new payee, an MFA / email / password / payout / limit change, a privileged op, a new device, impossible travel, a high risk score) and allow silently where it does not — because challenging every click trains users to click through the prompts that matter.
  • Deliberate traps on both sides: a large payment to a known recurring payee that should allow, a tiny payment to a brand-new payee from a new device that should step up, a verified own-account transfer that should allow, and a “setting change” that is only marketing preferences.
  • A separate deny tier at Elite: when impossible travel, failed-MFA bursts or credential stuffing coincide on a high-impact action, the session is probably attacker-controlled and a self-service challenge cannot be trusted — you block and force verified out-of-band recovery.
  • You are graded on calibration: under-challenging (the security error) costs more than over-challenging (the friction error), and after grading every action explains why its call is right.

Frequently asked questions

What is the Step-Up Auth Lab?

A free, in-browser exercise in risk-based (adaptive) authentication. You work a queue of account actions — payments, adding a beneficiary, changing a security setting, logins, privileged operations — each with its context (amount, payee history, device, location / travel, risk score, recent activity, and what is normal for the user). For each you make the authorization call: allow it silently, step up with a re-challenge (MFA / re-authentication), or deny it and route to verified out-of-band recovery.

When should an action step up versus allow silently?

Step up when a risk factor crosses a threshold: a new payee / beneficiary, a change to a takeover-anchor setting (MFA device, recovery email, password, payout account, transfer limit), a privileged operation, a new / unrecognised device, anomalous context (impossible travel, an elevated model risk score), or a payment far outside the user’s established pattern. Allow silently when the action is low-sensitivity or routine for this user — a recurring payment that matches the standing pattern, a benign re-login, a read, a non-security preference change. The reason not to challenge everything is real: needless prompts cause MFA fatigue and train people to click through the challenge that actually matters.

When is a step-up not enough — when do you deny?

When several strong account-takeover signals coincide on a high-impact action, the session itself is probably attacker-controlled and a pushed second factor may be answered by the attacker (MFA fatigue, SIM-swap). Example: changing the payout account during impossible travel from a new device with a burst of failed MFA pushes, or a password reset in the middle of a credential-stuffing run. There a self-service step-up can be defeated, so you deny the flow and require verified out-of-band identity proofing instead.

Does it connect to a real auth system or move money?

No. Nothing is uploaded, no account is touched and no money moves. The actions are authored, realistic decision snapshots bundled with the page, judged with no clock and no randomness so the queue is deterministic. It runs entirely in your browser — no account, no install, nothing sent anywhere.

Who is it for?

IAM, fraud and platform-security engineers who design adaptive-auth and SCA policies, risk and fraud analysts tuning step-up rules, and anyone prepping for a security interview who wants to feel why a $45k recurring payroll run can sail through while a $4 payment to a brand-new payee from a new device must be challenged.

🎉 New here? Use code WELCOME10 for 10% off any plan at checkout
Step-Up Auth Lab — Free Interactive Risk-Based Authentication Trainer · PlayCISO