Watch the full debrief

Two voices. One question. The insider reaction you don't usually see.

Also on YouTube 5–7 min 2026

Question decoded

"Design Instagram's News Feed ranking system. How would you decide what content to show each user?"

Competency tested

Focus on Long-Term Impact

Who asks it

Bar Raiser · HM · Peer

What they're really asking

Can you separate long-term user value from short-term engagement?

Answers compared

The answer that fails — and why

Candidate answer Does not raise the bar — Focus on Long-Term Impact

I'd design a two-stage system: candidate generation pulls the top N posts from a user's social graph and followed accounts, then a ranking model scores each post. For signals I'd use engagement history — likes, comments, shares, watch time — plus social proximity, content type affinity, and recency. I'd train a gradient-boosted model or a lightweight neural ranker. For scale, I'd precompute candidates and cache ranked feeds, refreshing every few minutes. I'd validate changes with A/B tests measuring click-through rate and session length.

Bar Raiser evaluation

⚑ Success defined entirely through short-term engagement proxies; no retention lens

⚑ No acknowledgement of the engagement-versus-wellbeing tension Meta explicitly debates

⚑ A/B metric is session length — known to be gameable and misaligned with long-term value

⚑ No mechanism proposed to detect or prevent ranking optimising for addictive content

Prefer to hear it? Watch the video for the two-voice delivery with live reaction commentary.

Meta debrief · SWE loop · Bar Raiser evaluation Below Bar

Meta Value: Focus on Long-Term Impact

Does not demonstrate Focus on Long-Term Impact.

✗ Defines ranking success solely via short-term engagement; no long-term user value framing

✗ Proposes CTR and session length as A/B metrics — both are known proxies, not outcomes

✗ No acknowledgement that optimising engagement signals can degrade long-term retention

✗ Architecture is competent but candidate shows no awareness of the measurement problem

interview101.com · Focus on Long-Term Impact · Meta SWE · Bar Raiser debrief reference

→ Now here's what a strong answer actually sounds like

The answer that works — in full

Strong answer Raises the bar — Focus on Long-Term Impact

Before I touch architecture, I want to define what good means here — because that's where most ranking debates at Meta actually live. Short-term, good is clicks and watch time. Long-term, good is whether a user comes back tomorrow and leaves feeling the time was worthwhile. Those two objectives can conflict. I'd design a two-stage system — candidate generation from the social graph, then a ranker — but I'd instrument two separate metric tracks: a short-term A/B signal like meaningful interactions, and a long-term holdout measuring 90-day retention and self-reported satisfaction. I'd explicitly cap the weight on pure engagement features to prevent the ranker from drifting toward content that's compulsive but not valuable. Any ranking change that improves short-term engagement but degrades the 90-day retention holdout doesn't ship.

Bar Raiser evaluation

✓ Opens by defining what 'good' means across two time horizons — rare signal

✓ Explicitly names the engagement-versus-retention tension without prompting

✓ Proposes a long-term holdout as a first-class measurement mechanism

✓ Hard gate on shipping changes that trade retention for engagement — concrete and defensible

Meta debrief · SWE loop · Bar Raiser evaluation Raises Bar

Meta Value: Focus on Long-Term Impact

Strong signal. Raises the bar.

✓ Defined success across short- and long-term horizons before proposing any architecture

✓ Named engagement-versus-retention tension explicitly and without prompting from interviewer

✓ Proposed a 90-day retention holdout as a concrete long-term measurement mechanism

✓ Articulated a hard shipping gate protecting long-term retention from engagement optimisation

interview101.com · Focus on Long-Term Impact · Meta SWE · Bar Raiser debrief reference

Fix your answer before your loop

Run your story through these three questions

1

Have you defined what 'good' means beyond the next A/B test?

If not, the Bar Raiser reads your answer as tactically competent but strategically blind.

2

Can you name one way your ranking system could hurt users long-term?

If you can't articulate the risk, you haven't shown you've thought past the sprint.

3

What is your concrete mechanism for measuring long-term user value?

Vague answers like 'monitor retention' are not mechanisms — they're aspirations.

Get your personalized report

How do your real stories score?

Get a personalized report scored against the interview rubric Meta uses for your role.

Get your Meta Software Engineer report →

More Meta Software Engineer debriefs