Operating Like a Start-up inside the World’s Biggest Company — Ramit Arora, Microsoft

Microsoft 365 consumer plans (personal, family, Copilot Pro) are sold via App Store/Play in-app purchase alongside ~10 other consumer channels and a large direct web business. Users already have a card attached and trust Apple/Google/Amazon to manage subscriptions, and many use the phone as their primary device. The team treated the store fee as the normal cost of any acquisition channel and invested in the mobile apps.

ASO covers branded keywords (brand value) and unbranded keywords. User research and Gen Z trends feed keyword optimization. A jobs-to-be-done list defines ~10 things each app should rank for (e.g. resume, budget, invoice), weighted toward jobs people are willing to pay premium for, so acquired users convert. Goal: keep at least the top three spots in most categories, since startups can outrank Word on any keyword.

One Microsoft 365 app and subscription covers Word, Excel, PowerPoint, notes, PDF, signing and a scanner ($9.99 family, $6.99 personal in the US). Many productivity jobs are one-off per month (one scan, a couple of documents), but combined across the suite users keep finding a use each month, which holds retention. Priced up to 50-60% below buying the same scenarios individually.

Product framework from a Microsoft VP: keep neutralizing competitor moves, keep differentiating, keep incubating, and maintain. After each WWDC the team asks how new features apply and ships day one (Dark Mode, Vision Pro). Competitor features are judged against the app's jobs to be done (e.g. Word: resume, note, book) rather than copied. Mix shifts with maturity: early on mostly neutralization and differentiation, now ~30% incubation (AI features).

AI features are picked by mapping generative AI capabilities (summarization, outline-to-document, retrieval-augmented generation, translation) onto the apps' jobs to be done: outline to document in Word, slide generation in PowerPoint, insights from Excel data, meeting recaps, and status updates pulled from the user's own files via Copilot. Aim: full document creation on mobile without 10 browser tabs.

The VP requires an estimated-impact figure for every idea, vendor or optimization. A large idea list from many sources is triaged; at double/triple-digit million MAU, work must affect tens of millions of users, not 1 million, because team size does not grow with MAU. The biggest wins are often simple: purchase flow, first-run experience, performance, reliability, file-open speed, app size, and 100% transaction completion (5-6% of transactions were failing at one point, fixed first).

In 2015-2018 the team added an MMP and an engagement platform while the mobile apps were still finding product-market fit. Implementations could not cope with fast-changing privacy law (GDPR) and a single app serving both enterprise and consumer users, and the vendors were young. Only after the mobile apps reached product-market fit did they integrate an MMP again (early 2024).

Uses Apple's promotional offers and promo codes, and is evaluating contingent pricing (new at the time). Aim is to stay current with every App Store monetization feature.