Is DashPass working — and should you change it? — DoorDash Consumer & Growth Data Scientist Interview

Is DashPass working — and should you change it?

Measure successHard

Problem. Leadership asks whether DashPass (the subscription that waives fees) is working, and whether to change its price or benefits. How do you measure success and decide?

Before you reveal: say your answer out loud, as if you were in the real interview — get your reasoning across clearly first. There is no single correct answer: reading what the interviewer is really after and defending your own thinking is what makes an answer strong.

Clarify the question first

"Success = incremental orders/frequency and retention from DashPass, or subscriber count?"
"Key risk: cannibalization — do subscribers just get free delivery on orders they'd have placed anyway?"
"Are we optimizing member LTV, or subscription revenue?"

Lay out your approach

Anchor on incremental frequency and retention, net of cannibalization. DashPass works if members order more than they would without it and retain better — not if they just save fees on existing orders.
Measure incrementality. Compare members vs a matched control (or a holdout / staggered offer), watching that subscribers self-select (heavy users buy it), which biases a naive comparison.
Full economics: incremental order margin from higher frequency minus the fees waived and the cannibalized-fee revenue; plus subscription fee.
For a price/benefit change: model conversion vs retention and cannibalization by segment; test where feasible.

Then check in: "Success is DashPass driving incremental frequency and retention net of cannibalized fees and self-selection, not subscriber count. Want me to work the incrementality and economics?"

Simulate the back-and-forth

Interviewer

Subscribers order way more than non-subscribers. Isn't DashPass clearly working?

You

Not from that comparison — it's self-selection. Heavy users buy DashPass because they already order a lot, so subscribers would out-order non-subscribers even if DashPass did nothing. That comparison measures "who buys subscriptions," not the subscription's effect. To isolate the causal effect I need members vs a matched control (same pre-subscription ordering), or ideally a randomized/staggered offer, and measure the change in frequency after subscribing versus that control.

Interviewer

How do you handle cannibalization in the economics?

You

DashPass waives fees, so for every order a member would have placed anyway, DoorDash loses the fee revenue it would have earned — that's cannibalization. So the economics are: incremental orders (from higher frequency) × their contribution margin, minus the fees waived on the orders that would have happened anyway, plus the subscription fee. DashPass is worth it if the incremental frequency and retention (and the subscription fee) outweigh the cannibalized fees. A member who orders the same amount but now pays no delivery fee is a cost, not a win — so incremental frequency is the crux.

Interviewer

Should you change the price?

You

It depends on the elasticity of conversion and retention, and on cannibalization, by segment. A lower price converts more members (more incremental frequency from light users) but waives more fees and may attract users who'd order anyway; a higher price reduces cannibalization but shrinks membership. I'd model conversion and retention vs price and the cannibalization each segment brings, and test a price/benefit change where feasible (a randomized offer). The decision is member LTV (incremental margin net of cannibalization) across the subscriber base, not subscriber count or subscription revenue alone.

Land a clear conclusion

You

So DashPass "works" if it drives incremental frequency and retention net of cannibalized fees, measured against a matched control or randomized offer (not a self-selected member-vs-non-member gap). I'd judge it on member LTV — incremental order margin plus subscription fee minus waived fees — and evaluate a price/benefit change by its effect on conversion, retention, and cannibalization by segment, tested where feasible. Subscriber count and subscription revenue alone are the wrong targets.

Key summary — what really matters.

Success = incremental frequency + retention, net of cannibalized fees — not subscriber count.
Beat self-selection — members vs matched control / randomized offer (heavy users self-select).
Economics: incremental order margin + sub fee − waived fees on would-happen-anyway orders.
Price/benefit change — model conversion, retention, cannibalization by segment; test.
Optimize member LTV, not subscription revenue alone.

Self-selection vs causal. Naive vs matched:

	Non-member	Member (naive)	Member (matched control)
Orders/month	3	9	7 (pre-subscribe baseline)

The naive gap says +6 orders, but members already ordered 7 before subscribing (self-selection), so the incremental effect is 9 − 7 = +2. The economics: +2 orders × margin + subscription fee − delivery fees waived on all 9 orders. DashPass is worth it only if that nets positive — the +2 incremental (and better retention), not the +6 illusion, is the crux.

Dimension	What interviewers look for	Below the bar ❌	Above the bar ✅
Problem framing	Whether success is incremental.	"Subscribers order more, works."	Incremental frequency + retention net of cannibalization.
Methodology and math	Whether you beat self-selection.	Member vs non-member.	Matched control / randomized offer.
Business judgment	Whether you model cannibalized fees.	Ignore waived fees.	Incremental margin + fee − waived fees.
Deployment thinking	Whether you test a price change.	Guess.	Model by segment; randomized offer.
Communication	Whether you headline the incremental number.	"26M members."	"Incremental +2 orders, net of cannibalization."

Interview tips

Strong-answer signal: isolating DashPass's causal effect from self-selection, and modeling incremental frequency net of cannibalized fees. Common trap: declaring DashPass a success from the member-vs-non-member gap, ignoring self-selection and cannibalization.