[ ]productinterview
Run a free mock

product sense · standard

"How would you improve Gmail?"

How would you improve Gmail?

Updated Jun 2026 Calibrated to the strong-hire bar

This question is not asking you to redesign Gmail. It is asking whether you understand what problem Gmail still has, who feels it most, and whether you can defend a specific bet without proposing January’s product launch as your big idea.

The immediate trap: Google shipped a significant AI upgrade in January 2026. AI Inbox (a briefing-style overview that surfaces priorities) and AI Overviews (natural-language inbox queries) launched alongside Suggested Replies, Help Me Write, and Proofread, all powered by Gemini 3. AI Overviews and Proofread are gated behind the Google AI Pro and Ultra tiers, signaling exactly where Google sees willingness-to-pay. Help Me Write is free, which is Google’s customer acquisition play at the base tier. AI Inbox was still in trusted-tester status at launch, with broader rollout planned for US English users. A candidate who proposes “better AI search” or “smart compose” in 2026 is proposing shipped features, and a Google interviewer will notice immediately.

Know the product before you pick the problem

Gmail has approximately 1.8 billion active users, making it the world’s largest email client. But raw user count is not the relevant context. What matters for an improvement question is the business problem: engagement depth has stagnated. Users open Gmail, triage, and leave. The product generates value for Google in two ways: it anchors consumer Google account engagement (which feeds ad signals and cross-product retention) and it anchors the Workspace business (10 million-plus paying organizations, direct SaaS subscription revenue). These two surfaces require different improvement theses. A candidate who treats all 1.8 billion users as one pool is skipping the strategy question.

One more constraint worth naming: Google tried to migrate Gmail users to Google Chat for team communication. It did not stick. Any improvement that amounts to “add more Slack-like features to Gmail” will raise an interviewer eyebrow, because Google has already attempted that path and pulled back.

Structure a strong answer

Clarify the goal, choose a specific segment with a real unmet need, propose one solution, name success metrics that actually reflect improvement, and state the tradeoff explicitly.

strong

"Before I propose anything: Google shipped AI Inbox and AI Overviews in January, so I want to find the white space those don't cover rather than re-propose them. My goal is improving Workspace retention and expansion revenue, since that's where Google earns direct subscription revenue and where incremental value is easiest to measure. Consumer Gmail matters, but it's harder to tie a specific Gmail improvement to an ad revenue signal directly.

The segment I want to focus on is knowledge workers inside Google Workspace orgs, specifically people who use their Gmail inbox as a task-management surface. This is the most widespread email workaround in existence: starring emails as reminders, leaving threads unread to signal "needs action," using folders to manually track in-progress work. It has been the default workaround for 20 years. AI Inbox v1 does not solve it. It reprioritizes the list, but it does not convert a thread into a structured task with an owner, a deadline, or a next step.

My proposal: a Thread-to-Project handoff. A single action, accessible from any email thread, that converts the thread into a Google Tasks project: extracted action items, suggested owners pulled from the thread participants, and a due date inferred from the conversation. The result surfaces inside Gmail without leaving the app. This is adjacent to what Google Workspace already does in Docs and Sheets with Smart Canvas, but it is not yet native to the Gmail inbox.

Viability argument: Google already sells Workspace on the promise of productivity. This feature deepens platform lock-in, reduces churn to Microsoft 365 (which has a native Tasks integration in Outlook), and is a natural AI Pro upsell since extracting structured action items from unstructured threads requires the model. It also gives Google a behavioral signal they currently lack: whether a thread actually resolved.

Success metrics: Thread-to-Project conversions per active Workspace user per week (leading indicator), Workspace contract retention at 6 months for cohorts who use the feature at least once per week (lagging), and percentage of extracted action items marked complete within the due date (quality signal). I would explicitly not use emails sent per week as a metric. If the feature works, triage gets faster, email volume may drop, and that is a good outcome, not a bad one.

The tradeoff I'm accepting: I'm not improving the consumer Gmail experience, which has more daily sessions. But a consumer improvement is harder to tie to a revenue metric, and the Workspace segment is where the business problem is clearest. If I were pitching this to a product council, I'd frame it as churn defense against Microsoft 365, not as an engagement feature."

weak

"I would improve Gmail's search with better natural language queries, add a collaboration tab so teams can work together like Slack, and let users customize their inbox themes." This answer fails on three counts. Natural-language inbox queries are AI Overviews, which shipped in January. Slack-like collaboration is Google Chat, which already exists and hasn't displaced email. Customizable themes are in settings today. Beyond the product awareness problem, the answer never establishes why Gmail needs to improve or what the business problem is. The candidate spent four minutes on a three-bucket user segmentation ("power users, casual users, occasional users"), picked "power users because they send the most emails," and landed on "emails sent per week" as the success metric. Optimizing email volume for Gmail is like optimizing page views for Google Search: it's a proxy that doesn't reflect whether the product is actually solving the problem.

The 2026 reframe

In 2026, “how would you improve Gmail?” is really asking: given that AI can now draft, summarize, triage, and reply to email, what human problems remain unsolved? Feasibility is near-free for Google. The scarce resource is clarity about which problems are worth solving and whether users will actually change behavior to adopt the solution.

The inbox-as-task-manager behavior is the clearest example of an unsolved problem that no AI feature has addressed yet, because solving it requires understanding workflow context across threads, not just inbox content in isolation. A candidate who names that gap, proposes a specific intervention, and defends its viability against a Google interviewer’s standard pushback (Why Workspace over consumer? Why not just improve AI Inbox v1?) is demonstrating the judgment that matters in 2026: not execution IQ, but knowing what is worth building.

See the CIRCLES framework for the full product-sense structure and feasibility is free for the 2026 lens this question requires. For a question with the same structure but different product dynamics, compare how would you improve Google Maps?

Framework

Asked at

Related