Reverse-Engineer Instacart's Growth Loop from Public Signals
Objective: Given only public app-store listings, help-center pages, and press coverage, map Instacart's growth loop into the lesson's four parts: input, action, output, re-investment.
You're a growth analyst pitching a loop-mapping exercise to your VP before your own team builds a referral feature. You want a real precedent, not a hypothetical, so you reverse-engineer Instacart's loop from what's publicly visible.
Using only Instacart's own app, referral landing pages, and public reporting (no internal data), reconstruct the loop diagram and identify which of the four steps is weakest based on visible evidence.
Before you start
What you'll need
Free path (everything below is enough to finish)
Free, no account friction, easy to share with a manager
The process
1 step
Step 01 of 01
The lesson's playbook says a real loop must answer four questions without ever saying 'and then we run an ad': what triggers a cycle (input), what happens inside the product (action), what artifact gets produced (output), and how that artifact becomes a fresh input.
Open Instacart's referral page and a recent order confirmation flow. What is the input, what is the action, what is the output, and what turns that output into the next input?
Procedure
- Open Instacart's public referral landing page (instacart.com/referral or the in-app 'Invite Friends' screen) and note the exact offer language
- Place an order (or read a recent order-confirmation email screenshot from a review site) and note every prompt shown after checkout
- Fill in the four-column sheet: Input = an existing user needing groceries, Action = they place an order and are shown the referral prompt, Output = a shareable referral code/link with a credit offer, Re-investment = the invitee redeems the code, becomes a new input
- Flag any step where the evidence is thin (e.g. you can't observe the actual redemption rate) and mark it 'unverifiable from public data'
INPUT: Existing customer places a grocery order ACTION: Post-checkout screen surfaces 'Give $10, Get $10' referral prompt OUTPUT: Unique referral link + dual-sided credit RE-INVESTMENT: Invitee redeems link on first order, becomes a new INPUT next cycle WEAKEST LINK (visible evidence): Output-to-re-investment conversion is unverifiable publicly — no redemption-rate data disclosed
Healthy
All four boxes filled with real observed behavior, not internal assumptions, and the analyst names exactly which step lacks public evidence.
Unhealthy
Skipping straight from 'they have a referral program' to 'it's a loop' without ever writing what the re-investment step actually is.
What this means
A referral program is only a loop if you can point to the specific mechanic that turns the output back into a new input — otherwise it's a one-time incentive with a loop-shaped label.
So what do I do about it?
| Symptom | Action | Effort |
|---|---|---|
| You can describe the referral offer but not the re-investment mechanic | Go back to the product and trace exactly what the invitee sees and does after redeeming, not just the offer copy | 5 min |
Final deliverable
A one-page loop diagram (four labeled boxes) for Instacart, with the weakest, least-verifiable step explicitly flagged.
See a reference example
Rent the Runway loop map (excerpt) INPUT: Existing subscriber's box is due for return ACTION: App prompts 'Add an item to your next shipment' OUTPUT: A styled outfit photo the subscriber posts or shares RE-INVESTMENT: Unverifiable publicly — no data on how many viewers convert from shared posts WEAKEST LINK: Output-to-re-investment, same gap pattern as most consumer subscription loops
Success criteria
You're done when you can:
- All four loop steps are filled with specific, observed behavior (not guesses)
- Exactly one step is flagged as unverifiable from public data, with a reason