Skip to content
Academy
Marketing Academy · Field Work●Mental Models
All projects·From: Second-Order Thinking
MiniAudit· 20 minutes

The 'And Then What?' Matrix: Auditing a Referral Growth Campaign

Grab Holdings

Objective: Build a 3-tier consequence matrix in Google Sheets to audit an aggressive customer referral incentive, mapping immediate lifts, behavioral shifts, and systemic costs before budget approval.

You are a marketing strategist at Grab assessing a proposed 'Double Your Referral' campaign across Southeast Asia (giving $10 ride credits to both referrer and referee). The growth squad forecasts a 40% surge in new user signups. Your task is to audit the decision using a Second-Order Impact Matrix to determine if the campaign creates sustainable acquisition or an exploitable incentive trap.

Follow the step-by-step procedure in Google Sheets to audit the referral campaign: map the 1st-order lift, the 2nd-order behavioral responses (fraudulent multi-accounting, non-incremental cannibalization), and the 3rd-order equilibrium (blended CAC inflation, driver payout friction), and define guardrails.

Before you start

What you'll need

Free path (everything below is enough to finish)

Google Sheets
FreeSpreadsheet for building the 3-level Second-Order Impact Matrix

Free and accessible spreadsheet tool for tabular risk and consequence modeling.

No access? Notion or any spreadsheet application

The process

1 step

Step 01 of 01

The Playbook: The "And Then What?" Drill

Second-order thinking requires asking 'and then what?' at least twice beyond the immediate metric lift to uncover delayed behavioral reactions and unintended incentives.

What are the specific 1st, 2nd, and 3rd order consequences of doubling the referral incentive, and what are their estimated probabilities and mitigations?

Google Sheets— Open Google Sheets and create a tab named 'Referral Impact Matrix'

Procedure

  1. Create columns: Level (1st/2nd/3rd Order), Actor (New User / Existing User / Fraudster / Driver), Consequence Description, Impact Direction (+/-), Probability (%), and Operational Guardrail.
  2. Document 1st-order effect: immediate +40% surge in app referral downloads this month (+).
  3. Document 2nd-order effects: coupon-forum sharing causing non-incremental signups (-), existing users creating burner SIM accounts to claim $10 credits (-), referee 30-day retention dropping below 12% (-).
  4. Document 3rd-order effects: blended CAC rising 35% above lifetime gross margin, promo budget exhausted before genuine organic word-of-mouth activates.
  5. Formulate 2 operational guardrails to prevent exploitation while preserving referral virality.
Sample output
| Level | Actor | Consequence | Impact | Prob | Operational Guardrail |
|---|---|---|---|---|---|
| 1st Order | General Public | +40% referral signup surge in 30 days | +Gross Volume | 90% | Monitor acquisition run-rate |
| 2nd Order | Deal Seekers | Users post codes on public discount aggregators | -CAC Quality | 75% | Invalidate referrals from public coupon domains |
| 2nd Order | Multi-Accounters | Burner SIM account creation for $10 rides | -Promo Waste | 60% | Enforce device-ID fingerprinting & verified payment requirement |
| 3rd Order | Unit Economics | Blended CAC doubles; 60-day cohort retention drops to 11% | -Net Margin | 65% | Gate referee payout behind completion of 2nd paid trip >$8 |

Healthy

Matrix evaluates at least three tiers of consequences with specific probability estimates and identified actor behaviors across the entire ecosystem.

Unhealthy

Matrix lists only the intended 1st-order upside and vague generic risks without detailing specific behavioral causal chains.

What this means

If negative 2nd- and 3rd-order impacts carry >50% probability and erode net cohort margin, the incentive structure must be gated with behavioral guardrails before launch.

So what do I do about it?

SymptomActionEffort
High probability of referral credit abuse on public coupon forumsAdd qualification gate: referral credit unlocks only after referee completes second paid ride >$830 min
Referee 30-day retention drops below 15% across promo cohortsReplace single lump-sum $10 credit with staged $2.50 discount vouchers across rides 1 through 430 min
YouYou can do this yourself, no engineering access required.

Final deliverable

A completed Second-Order Impact Matrix in Google Sheets with three consequence levels, assigned probabilities, and two operational guardrails.

See a reference example
Sample output
Worked example for a different delivery company (Instacart grocery delivery referral bonus):
| Level | Actor | Consequence | Impact | Prob | Operational Guardrail |
|---|---|---|---|---|---|
| 1st Order | Casual Shoppers | +35% referral signups in 14 days | +$140k GMV | 95% | Track baseline volume |
| 2nd Order | Deal Seekers | Users create burner accounts with disposable emails | -$35k promo waste | 70% | Require phone SMS + card verification |
| 2nd Order | Existing Loyalists | Heavy organic referrers exhaust friends list | -$12k future referrals | 45% | Cap at 5 referrals/month per account |
| 3rd Order | Unit Economics | Blended CAC rises 22% due to zero-repeat promo hunters | Margin squeeze | 60% | Tie payout to $35+ basket size on 2nd order |

Success criteria

You're done when you can:

  • Maps at least 3 tiers of consequences across multiple ecosystem actors
  • Assigns realistic probability weights (not 0% or 100% assumptions)
  • Identifies at least one negative second-order behavioral shift
  • Proposes actionable friction or qualification guardrails to preserve first-order upside