Is the portfolio approach for indie apps a myth?

By Matheus Mello, solo founder at Quantivo AI — 10 subscription apps and a couple of web games shipped. Every figure below was pulled from my own RevenueCat account on July 19, 2026 (trailing 28 days: June 21 – July 19). Published July 19, 2026.

Mostly yes, as a diversification story: revenue follows a power law. Two of my ten apps produce 87% of my (tiny) $64 MRR, and in the well-known $602K portfolio, one app is effectively the whole business. A portfolio only earns its keep as a search strategy — finding your winner faster and killing the rest sooner.

Here’s the full data, including the embarrassing parts.

What does a 10-app portfolio actually earn?

Here is my entire subscription-app portfolio, unfiltered. These are real RevenueCat numbers, not screenshots of a good week:

Totals: $64 MRR, 15 paying subscribers, $87 collected in the last 28 days, 548 new users across the portfolio. Scope note: this is app-store subscription revenue only — my web games earn ad revenue that isn’t in these numbers, and most of these apps are under six months old.

The distribution is the story. The top app holds 67% of MRR. The top two hold 87.5%. Six of ten earn exactly zero. The power law showed up before the revenue did.

Isn’t one hit always the whole business?

Usually, yes — and it’s worth conceding this plainly, because it’s the strongest argument against the portfolio approach.

The clearest public case: Sebastian Röhl’s roughly $602K/year portfolio is four apps, but HabitKit — about $28K MRR from ~25,100 subscribers paying $1–2/month — is effectively the entire business. The other three apps combined reportedly account for under 1% of revenue. And HabitKit took about 2.5 years to reach $10K MRR. Even the portfolio-maximalist counter-story — the 30-app portfolio doing $22K/month — is a strategy of shipping many bets precisely because the author couldn’t predict which would hit.

My data agrees at 1/400th the scale. If concentration this extreme shows up at $64 MRR, there’s no reason to expect diversification magic at $60K MRR. A portfolio does not smooth your revenue. It will not protect you the way index funds protect investors. One app will be most of the business, and you don’t get to pick which one in advance.

If one app wins anyway, why run a portfolio at all?

Because the power law is an argument for portfolio discipline, not against portfolios.

Nobody — including Sebastian, by his own account — knew in advance which app would be the one. A portfolio is not a diversification strategy; it’s a search strategy. The job is to identify your winner faster and kill the rest sooner, and to do it with data instead of sentiment. The failure mode isn’t having ten apps. It’s having ten apps and no idea which two matter.

Two things my own numbers forced on me this quarter:

• Zombies survive by invisibility. Calorie Tracker App has been live for 15 months. It has one subscriber, $3 MRR, and zero active users in the last 28 days. I hadn’t consciously decided to keep it — I just never had its number in front of me next to the others. Scattered dashboards are how a project stays alive for a year past its expiry date.
• Sentiment misallocates time. The app I most enjoyed building (RepBro) earns $0. The app that pays (Nutri AI) is the one I was treating as a side experiment. I would not have believed the ratio without seeing both numbers in the same view.

So the honest claim is narrow: portfolios don’t beat focus, and focus doesn’t beat portfolios. Visibility beats blindness in either strategy. If you run one app, you need its numbers. If you run ten, you need them side by side, or the power law makes your decisions for you — slowly and expensively.

When do you kill a project? A framework from the actual data

Reading my own table, the projects sort into four buckets, and each bucket has a different correct action. Conflating them is how founders kill the wrong things:

1. No demand → kill. 60+ days after launch and at least one real distribution attempt, with near-zero active users and zero new installs. LookProof AI and UsTonight are here: zero revenue, zero-to-one users. No dashboard, redesign, or price change fixes absent demand.
2. Revenue without usage → zombie, sunset it. Calorie Tracker App: $3 MRR, one subscriber, zero actives, 15 months old. One loyal subscriber is a kindness, not a business. Freeze it or shut it down; either way, stop the ambient maintenance cost.
3. Usage without revenue → fix monetization, don’t kill. RepBro has 85 active users, 82 new installs, three people in trials — and $0 MRR. That’s not a dead product; it’s a broken paywall or wrong pricing. Killing it would destroy the only asset that’s actually hard to get: people who use the thing.
4. Too young to judge → set a review date, don’t decide. SnapSweep, ColdCart, and Antique Identifier shipped in June. Twenty-eight days of data on a base this small is noise. The discipline here is the opposite: resist reading anything into it, and put a decision date on the calendar.

Sleep Arc is my borderline case — two new users in 28 days, $5 MRR. It gets one deliberate distribution push; if the trend line doesn’t move, it moves to bucket 1. Writing the rule down before the push is the point: the data decides, not my attachment.

How do you actually track revenue across 10 apps and sites?

No single free tool covers a mixed portfolio, so be clear about what each category does and doesn’t do:

I build SoleOS, and the portfolio above is its dogfood data — that’s both my credential and my bias, so weigh accordingly. For plenty of founders the honest recommendation is a row above it: AppWatch is free for three apps, PostHog’s free tier is generous, and a spreadsheet you actually update weekly beats any dashboard you don’t open.

What a portfolio dashboard cannot do

• It won’t fix a product nobody wants. LookProof AI is dead on every dashboard ever built. Aggregation surfaces the corpse faster; it doesn’t revive it.
• Projections from small MRR bases have wide error bars. At $64 MRR, any “you’ll hit $1K by November” line is directional at best. SoleOS’s own projection engine refuses to fit under 21 days of data, labels low-confidence fits (r² below 0.6) as rough estimates, caps ETAs at 18 months, and prints a plain “stalled” when the data says so — because pretending otherwise would be lying with charts.
• It adds a subscription on top of the tools it aggregates. You still pay for (or run) Stripe, RevenueCat, PostHog, and your backends. Aggregation is a convenience layer, not a replacement layer.
• Who should not buy it: founders with one or two projects. Your Stripe or RevenueCat dashboard plus a weekly spreadsheet habit covers you for $0. The math only changes around 3+ projects on mixed stacks, when checking everything means eight tabs and the zombies start hiding.

FAQ

Portfolio data: author’s RevenueCat account, pulled July 19, 2026. Case-study figures: buildmvpfast’s breakdown of Sebastian Röhl’s portfolio and Indie Hackers; not independently audited.