Company Research and Business Understanding Questions

I treat those shifts as a change in the data-generating process, not just noise. First I re-check the business constraint: if regulation changes what we can collect or automate, the model may need to become simpler, more interpretable, or partially rule-based. If there is a macro slowdown, I expect demand elasticities, conversion rates, and churn to change, so I would retrain on recent windows, segment by cohort or region, and monitor calibration and drift more aggressively. For platform dependency or supply-chain issues, I would add external signals and scenario features, then recommend actions with ranges rather than point estimates.

My goal is to make the model decision-aware:

• Separate structural drivers from temporary shocks
• Build scenario analysis around optimistic, base, and stressed cases
• Add guardrails so recommendations remain valid under constraints
• Communicate when the right answer is to pause, hedge, or conserve cash

For example, if supply shortages hit fulfillment, I would stop optimizing only for revenue and instead optimize for contribution margin, service level, and customer retention risk. That keeps the model aligned with business reality.

I would start by decomposing growth into the core funnel: traffic, conversion, retention, and monetization. Then I would ask for the smallest set of data that lets me localize the break point.

First data I want:

• Time series of new users, qualified leads, revenue, and active customers
• Channel-level acquisition spend, CAC, and conversion rates
• Cohort retention and repeat purchase curves
• Pricing history, discounting, and ARPU / AOV trends
• Market context such as seasonality, competitor moves, and geography mix

Then I would test each hypothesis:

• Acquisition efficiency: is CAC rising, conversion by channel falling, or traffic quality declining?
• Conversion: are landing page, sales, or checkout rates down after a process change?
• Retention: are newer cohorts churning faster, or is expansion revenue weaker?
• Pricing: did price increases reduce win rates or volume more than expected?
• Market saturation: are we seeing flattening new-customer growth even when spend rises?

I would also compare marginal returns by channel and cohort, because rising spend with slowing growth often means we are buying lower-quality demand.

In a previous role, I was asked to support a pricing decision in a domain I did not know well. I had to learn the customer workflow, the commercial levers, and the constraints that sales and operations were dealing with before I could give a useful recommendation.

Situation and task: Leadership wanted a data-backed view on whether a price change would help revenue without hurting retention.

Action: I spent the first few days with sales, finance, and customer success to understand how customers actually bought and renewed. I reviewed call notes, renewal reasons, and usage patterns, then built a simple segment analysis instead of a one-size-fits-all view. I also shared my assumptions early so stakeholders could correct them.

Result: The recommendation shifted from a broad price increase to a segmented approach with guardrails for sensitive customers. That made the proposal more credible and easier to act on.

What I did differently afterward was treat domain learning as part of analysis, not a pre-step. I now ask faster questions about incentives, decision rights, and customer pain points before I model anything.

I would use a core set of health metrics that apply across businesses, then add metrics specific to the revenue model.

Common metrics:

• Revenue growth and gross margin
• Customer retention and concentration risk
• CAC payback and LTV / CAC
• Cash efficiency and operating leverage
• Product usage tied to the main value proposition

For a subscription business, I would emphasize:

• MRR or ARR growth
• Gross and net revenue retention
• Churn, expansion, and cohort retention
• Activation and feature adoption

For a marketplace, I would focus on:

• Liquidity and match rate
• Take rate and transaction volume
• Repeat usage on both sides of the market
• Supply-demand balance and unit economics by geography or segment

For an ad-supported business, I would track:

• DAU / MAU and session depth
• Fill rate, CPM, and revenue per user
• Ad load versus user experience
• Retention of both users and advertisers

The difference is that each model has a different engine: subscriptions depend on retention and expansion, marketplaces on liquidity and trust, and ads on engaged audiences plus monetization efficiency.

I usually start outside the company, then map what I learn back to analytics opportunities. I would review the website, pricing pages, annual reports if public, investor materials, app reviews, and customer case studies to understand the business model and who pays. Then I would look at product flows and job postings to infer what the company values operationally.

My checklist is:

• Who are the customer segments and what problem do they buy for?
• How does revenue come in: subscription, transaction, ads, usage, services?
• What are the main growth levers: acquisition, retention, expansion, pricing?
• Where are the likely bottlenecks in the funnel or operating model?

From there, I prioritize data science opportunities by impact and feasibility. For example, if revenue depends on repeat usage, retention analytics and churn prediction matter more than top-of-funnel models. If revenue is transaction-based, I would look at conversion, fraud, and pricing elasticity. The goal is to identify the few metrics and decisions that truly move the business, not just produce interesting analysis.