product sense · standard
"How would you improve Instagram Reels?"
How would you improve Instagram Reels?
The failure mode on this question is proposing a feature backlog without a segment, a pain, or a business rationale. Interviewers at Meta are testing whether you understand that Reels is a two-sided market where content supply is the constraint, that it sits inside a portfolio of competing Instagram surfaces, and that in 2026 the discoverability problem is essentially solved. The unsolved problems are creator viability and viewer loyalty.
Before you walk in, know the scale: 2 billion users interact with Reels monthly, there are 200+ billion daily plays, and 55% of Reels views come from non-followers. Discovery works. The gap is downstream of discovery: creators who build an audience cannot earn from it below the 600,000 view threshold, and viewers who find a creator they like have no reliable way to build a relationship with that creator’s work rather than just consuming an algorithm feed.
One more business tension to name: Reels generates less ad revenue per view than Feed. That means Meta has a direct financial incentive to convert Reels watch time into higher-value inventory, which makes improvements to creator output and viewer retention commercially necessary, not just product-nice.
Scope before you propose anything
The strongest scope is mid-tier creators: accounts with 10,000 to 200,000 followers. They represent the long tail of original content supply. They build niche audiences and get real reach through the non-follower discovery system. But they churn because the economics do not work. If they leave for YouTube Shorts (which hit 200 billion daily views in 2026, parity with Reels) or a rehabilitated TikTok, Reels loses the content variety that makes the surface worth opening. The content ecosystem thins at the exact stratum that makes Reels feel like a real creator community rather than a loop of the same fifty accounts.
The December 2025 algorithm update made this worse: Instagram now weights the first 24 to 48 hours of a post’s signal heavily. Niche content builds through shares over days, not viral spikes in two days. Mid-tier creators posting for a specific interest community are structurally disadvantaged by a recency-and-velocity signal. Accounts that repost 10 or more times in 30 days are fully excluded from Reels recommendations, which means one of the common mid-tier tactics for reach is now penalized.
Structure a strong answer
strong
"Let me scope before I propose anything. I want to focus on mid-tier creators: accounts with 10K to 200K followers. They drive content supply, but they can't clear the 600K view monetization threshold consistently. If this segment churns, Reels loses the content variety and original voices that make the surface worth discovering. My goal is to make the platform economically viable for them, which in turn sustains the content supply that drives viewer retention and ad revenue.
Here's the pain loop: mid-tier creators post consistently, build niche audiences, and get non-follower reach through the algorithm. But the early-signal weighting (first 24 to 48 hours) punishes niche content that compounds through shares over days rather than spiking immediately. They miss the 600K view gate, earn nothing, get no feedback on why a specific video underperformed, and eventually slow their posting cadence. That cadence drop is the leading indicator of churn.
Three improvements tied to that thesis:
First, milestone monetization. Replace the hard 600K view gate with a tiered earnings structure starting at 50K views, funded by a small share of the ad revenue the creator's content is already generating. That video is monetized; Meta is keeping a disproportionate share below the 600K line. Redistributing a fraction costs no new ad inventory and directly fixes the retention problem. It changes the creator's relationship with the platform from 'I'm building toward a gate I may never reach' to 'every post earns something.'
Second, an AI hooks editor as a pre-publish tool. It analyzes the first three seconds of a Reel against the creator's own audience retention data and surfaces specific drop-off risk: whether to cut to action faster, add a text overlay in the first second, or reorder the opening frame. TikTok and CapCut have no equivalent first-party tool tied to the creator's own data. This addresses the feedback-loop gap, differentiates Instagram's native editor from third-party tools, and gives creators a reason to stay in the ecosystem. The AI model improves with tenure: a creator who has 200 past Reels generates better personalized signal than a new account. That creates platform-specific switching cost.
Third, a Following-only Reels feed. A chronological, algorithm-free tab showing only accounts you explicitly follow, in reverse time order. This addresses the viewer lovability gap: right now, a viewer who discovers a mid-tier creator through the algorithm has no reliable way to prioritize that creator's content. The main Reels feed serves whatever the algorithm surfaces. A Following tab gives viewers agency and gives creators a direct channel to their core audience outside the early-signal window.
North star metric: monthly Reels creators who earn at least $100, as a proxy for whether the platform is economically viable at the mid-tier. Primary guardrail: Reels watch time per non-follower session, to confirm discovery stays healthy while we improve the follower experience. Secondary guardrail: Reels ad CPM, to confirm tiered monetization does not compress ad rates.
What I'd explicitly not build: new editing effects and filters. TikTok and CapCut already win the creative tooling battle. Competing there is a maintenance task, not a strategic move, and it does not address why mid-tier creators are churning."
weak
"I'd add better music licensing, AI filters, longer video support, and a collab feature." No user, no pain, no goal, no tradeoff. This is a feature backlog. A second weak pattern: framework recitation that never lands on a real improvement ('Step 1: clarify the goal, Step 2: identify segments...') with no concrete ideas by the end. A third pattern: proposing AI improvements vaguely ('use AI to improve recommendations') without specifying the mechanism, the user pain it solves, or the metric it moves. Interviewers at Meta flag all three. A fourth: answering only from the viewer side and ignoring creators, which misses that content supply is the constraint on a two-sided platform.
The cannibalization subtext
Meta interviewers use this question partly to probe whether you recognize the intra-product tension. Reels competes with Stories for short-form creator attention, with Feed for posting behavior, and with Threads video for time spent on video inside Meta’s portfolio. The Following-only Reels feed proposed above directly touches the Stories dynamic: it creates a Reels surface that functions the way Stories used to for close followers. Name this tradeoff explicitly and argue it: Stories is declining as a creator-to-follower communication format; a Following tab recaptures that relationship inside a higher-quality video surface without introducing net-new cannibalization. Weak answers propose the feature without acknowledging the tension at all.
Metrics to know cold
Engagement rate for Reels averages 2.46% overall, but drops to 0.45% for accounts with 100K to 1M followers. Small and mid-tier creator economics are meaningfully different from large creator economics, and interviewers expect you to know that segmentation. 60 to 90 second Reels get the highest engagement; this is a design constraint, not a recommendation.
The 2026 framing
In 2026, feasibility is essentially free for Meta. The question is whether an improvement makes the creator economy viable at the mid-tier (which sustains content supply) and whether it makes viewers build creator relationships rather than just consuming an algorithm feed (which sustains retention). Candidates who anchor their answers to that thesis, rather than to feature novelty, sound like they understand Meta’s actual strategic problem.
See what happens when Instagram Stories cannibalizes posts for the intra-product follow-up interviewers use. The north star metric framework covers the guardrail structure in more depth. And feasibility is free is the lens that applies to every Meta product-sense question.