For most of the last decade the app market rewarded volume. Ship fast, buy installs, raise the next round. That playbook has quietly expired. Consumer spending across the two major stores is still setting records every quarter, yet the number of products capturing it keeps shrinking. The market did not slow down — it became selective.

The shift shows up first in acquisition costs. Paid install prices in competitive consumer categories have roughly doubled since the start of the privacy-first advertising era, while the median first-week retention of a new app has barely moved. When the cost of a user rises faster than the value that user produces, the only sustainable answer is a better product. That is the whole story of the current cycle, compressed into one sentence.

Fewer apps, deeper sessions

Users are not installing less software; they are consolidating. The average smartphone now holds more apps than ever, but active weekly usage concentrates in a small group of tools that earn their place on the home screen. Everything else becomes a one-time install that is quietly forgotten after a single session.

For product teams this changes the target. The question is no longer "how many people can we reach" but "what job are we the default answer for". Apps that own a specific, repeated moment in someone's day — a meal, a commute, a shift handover, a training session — survive the consolidation. Apps that offer a general-purpose surface without a repeated trigger do not.

The most valuable metric of this cycle is not downloads or even revenue. It is the number of days per month a person chooses to come back without being reminded.

Monetization follows the same logic

Efficiency has reshaped pricing too. Flat monthly subscriptions still dominate revenue, but the fastest-growing products layer several models on top of each other: a generous free surface to establish habit, a subscription for the core loop, and consumable or usage-based pricing for the expensive parts — typically anything that calls a model on a server.

That last part matters more than it looks. AI features carry a real marginal cost per action, which breaks the classic assumption that software gross margin is close to fixed. Teams that priced AI as an unlimited subscription perk in the first wave are now rebuilding their plans around metering. Teams that started with metering are quietly compounding.

What we tell clients

  • Instrument the first ten minutes. Almost every retention problem is an onboarding problem wearing a disguise. If a person does not reach the core value in one session, no amount of push notifications will recover them.
  • Price the expensive part separately. Model calls, exports, rendering and storage should be visible in your pricing model before they are visible in your margin.
  • Build one loop properly. A single, well-instrumented habit loop beats five half-finished features in every cohort analysis we have run.
  • Treat store presence as product surface. Screenshots, first frame of the preview video and the first line of the description convert or lose more users than most in-app experiments ever will.

The efficiency era is not a downturn. It is a market that finally pays for craft. For teams that were already building carefully, the next few years are the best window in a decade.