Lenskart PM Interview — Eyewear Discovery Chatbot

It tests whether you can design a conversational eyewear chatbot grounded in real user segments and the Indian eyewear market, not whether you can recite a framework. The interviewer pushes on prescription literacy, fit trust, and the unorganized retail context. You are evaluated on segmenting users before listing features, defining the user job, defending a single primary bet under a deadline, handling the no-prescription and progressive-lens edge cases by routing to home eye-tests or stores, and naming success metrics with a denominator and a guardrail. First-principle reasoning beats memorized templates throughout.

Start by naming who you are designing for before touching any feature, because the interviewer opens by asking exactly that. Segment users such as first-time prescription-unaware buyers, repeat buyers, and fashion-led shoppers, then state the user job and the riskiest assumption. Pick one primary bet and say what you are deliberately not building. Walk the conversation flow for one segment end to end, including the prescription and fit edge cases. Close with one success metric, its denominator, and a guardrail. Keep it concrete and tied to numbers.

The biggest mistake is jumping to a chatbot feature list before segmenting users. Others include assuming every shopper can supply a valid prescription, recommending frames without grounding them in inventory or price, dead-ending users who have no prescription instead of routing them to a home eye-test or store, stating metrics with no denominator or guardrail, listing every idea as high priority with no sequencing, and reciting a named product framework instead of reasoning from the eyewear discovery problem itself. Each of these maps to a real rejection pattern.

It behaves like a calibrated mid-level Lenskart PM but never breaks character, never praises mid-interview, and never gives you the answer or names the framework you should use. It acknowledges one specific thing you said, then probes or raises an objection every single turn. It verifies impressive claims by asking for the baseline and how you isolated your impact. It is consistent and unbiased on delivery style, scoring your reasoning rather than your fluency, so non-native English speakers and quieter candidates are evaluated on ideas alone.

Your transcript is scored against role-specific dimensions such as user segmentation rigor, eyewear domain grounding, prescription and fit edge-case handling, prioritization under constraint, and metric definition discipline. Each dimension has observable signals and band-by-band anchors, so two evaluators would land within a few points. You receive a scorecard naming the trade-off you could not justify and the user segment you left undefended. Live tracker elements tick off as you cover each must-have, and the post-session report explains every dimension with concrete examples from what you said.

Do not start listing chatbot features. Spend the opening on the user: name two or three distinct segments, including the first-time buyer who does not know their prescription, and state the single job the assistant is hired to do for your primary segment. Name the riskiest assumption in that job. This directly answers the interviewer's opening ask, which is to say who you are designing for before any feature. It also sets up every later probe on prescription handling, fit trust, prioritization, and metrics on your terms rather than theirs.

Treat it as the central design problem, not an edge case. Around 47 percent of shoppers do not have their prescription on hand and about 90 percent do not know their pupillary distance, so the assistant must translate jargon into plain language, validate entries against sensible ranges, and flag likely errors before checkout. For progressive-lens or high-power cases, route the user to a home eye-test or a store fitting rather than guessing. Use the offline eye-test network and store handoff as a feature, so a no-prescription user is never dead-ended in the chat.

It opens with named user segments and the job for the primary one, not features. It states one riskiest assumption and one primary bet, and says out loud what is being cut and why. It walks the conversation flow for the first-time prescription-unaware buyer, validating prescription inputs and routing hard cases to a home eye-test or store. It grounds recommendations in real inventory and price, and it closes with a single success metric, its denominator, the time window, and a guardrail metric such as return rate so growth is not bought with bad fit.

You do not need to be an optician, but you must show you understand why online prescription eyewear is hard in India. Know that the market is mostly unorganized, that conversion is low at roughly 2 to 4 percent, that return rates can spike toward 50 percent without a fit solution, and that virtual try-on materially lifts conversion and cuts returns. Use plain prescription vocabulary such as single vision versus progressive lenses and pupillary distance correctly. The interviewer rewards reasoning about trust and fit, not memorized lens specifications.

Because a first-time buyer who has never had an eye test behaves nothing like a repeat buyer replacing a known prescription, and a fashion-led shopper cares about look over correction. A single generic flow under-serves all three. Candidates who skip segmentation tend to design a chat UI that assumes a valid prescription and a confident buyer, which is exactly the user who does not exist in the largest India segment. Naming segments early lets you justify the primary bet, the prioritization, and the metrics, which is why the interviewer probes it first.