What Google Photos’ end of unlimited backups teaches PMs
A story about letting unlimited features scale costs—and how to choose caps, cohorts and communication to protect most users and the product.

I watched one of the big photo services end its unlimited backup offer. The announcement landed the way these things do: a technical reason framed as product improvement, a version of a cap, and lots of angry users. The change was inevitable. The lesson for product managers was not about PR — it was about prioritisation by unit cost.
In brief
How it unfolded
Usage kept growing. Average revenue per user and average costs told a flattering story. But the storage bill came from the heaviest users. The service ended the unlimited promise and introduced a cap. Backlash followed. Much of it could have been reduced with different prioritisation and clearer tooling.
Measure cost-to-serve by cohort
Don’t rely on averages. Split users into cohorts by storage use and device type. Calculate marginal cost per GB for each cohort. Often a small percent of users account for a large share of storage. That’s where the unit cost is highest and the product economics are most sensitive.

Design the change
Rollout and reduce backlash
Pair the policy with tooling and communication. Give users dashboards showing what they store, where it came from, and how to reduce it. Offer temporary exceptions for device cohorts (old phones, company devices) and use grandfathering for long-time users.
Worked example: model the trade-offs
Steps you can follow right now without guessing numbers:
- Compute cost_per_gb = total_storage_costs / total_gb_stored.
- Rank users by storage and find the 90th and 10th percentile usage. Note how much data the top 10% hold.
- Choose a cap, e.g. 15 GB. Estimate how many users lie above that cap from your distribution.
- Model outcomes: assume X% of capped users upgrade, Y% accept limits, Z% churn. Translate into revenue change: delta_revenue = upgrades * ARPU - churn_loss.
- Estimate cost savings: saved_gb = sum(storage_above_cap_for_users_who_leave_or_downsize); saved_cost = saved_gb * cost_per_gb.
- Compare net effect: saved_cost - delta_revenue. Iterate cap and exceptions (device cohorts) to protect users who matter most to retention or revenue.
Mistakes to avoid
- Hiding the true reason. Users will infer you’re monetising; be honest about costs and trade-offs.
- Broad blunt cuts that hit your most engaged users. Those users deliver value beyond storage metrics.
- Changing terms without tools. Don’t expect users to trim backups if you give them no way to identify and delete large items.
- Failing to show the net benefit. Explain how the change preserves the product for the many.
What it taught me
The core decision is a prioritisation question: which marginal cost can the product sustain, and which it cannot. When usage runs away, decide by unit cost per cohort, not by headline satisfaction numbers. Design limits to protect the majority, target the tail, give people tools, and manage exceptions. If you need a framework to pick which cohorts to protect first, consider a structured prioritisation exercise with your data and stakeholders and a tool like Prioritisation to keep choices transparent.


