What should you build and sell to normies?

Are normies actually spending more money on apps now?

Yes. Normal consumers are spending more on apps today, and non-game apps have become a much bigger business than they were even a few years ago.

Sensor Tower's State of Mobile 2026 found that consumers spent $167 billion through mobile in-app purchases in 2025, up 10% in a year. The bigger change happened outside gaming. Spending on non-game apps jumped 21% and passed games for the first time, after nearly tripling over five years.

Health and fitness shows how much willingness to pay has changed. Sensor Tower counted about 3.96 billion downloads in 2025, barely higher than the year before, while spending rose 13% to roughly $4.5 billion. More people downloading apps did not explain most of the growth. Existing users were simply spending more.

There are now large consumer software businesses hiding inside very ordinary activities. Rocket Money generated $351 million from subscriptions in 2025. Life360 generated $369 million from subscriptions around family location and safety. PictureThis has passed 50 million Android downloads by helping people identify and look after plants. ReciMe says more than 10 million people use its cooking app.

So there is plenty of consumer money available. The hard part is convincing someone that our particular little app deserves it.

Why is it still so hard to make money from consumer apps?

Because building consumer software has become incredibly easy while getting consumers to care has barely become easier.

RevenueCat's latest State of Subscription Apps report gives us a brutal picture. Around 2,000 new subscription apps were launching each month in early 2022. That figure has climbed to more than 14,700. Roughly seven times more subscription apps are now fighting for attention every month.

Most of them go nowhere. RevenueCat found that one year after launch, the median subscription app makes only about $72 per month. Around $429 per month gets an app into the top quartile, while reaching roughly $2,574 puts it in the top 10%.

The old apps still dominate too. Products launched before 2020 generate 69% of subscription-app revenue in RevenueCat's dataset. Apps launched in 2025 or later account for only 3%, despite the explosion in new launches.

That should change how we think about ideas. A habit tracker that takes two weekends to build is easy for us because it is easy for everybody else too. The same applies to AI journals, generic planners, simple photo generators and most wrappers around foundation models.

These days, the scarce part is getting somebody to notice an app, understand it immediately and still want it three months later.

Do normal consumers really pay for AI?

Only a small minority do. Consumer AI usage is enormous right now, but willingness to buy a separate AI subscription remains surprisingly weak.

Menlo Ventures estimated that roughly 1.8 billion consumers were using AI at least periodically, including around 600 million daily users. Yet it estimated total consumer AI spending at only $12.1 billion. On its numbers, about 3% of consumer AI users were paying.

Where the money goes is even more revealing. Menlo estimated that 81% of consumer AI spending went to general assistants and only 19% to specialized AI products. ChatGPT, Gemini and the other big assistants have trained hundreds of millions of people to expect surprisingly capable AI for free.

Recent subscription data gives us another warning. RevenueCat found that AI-powered apps generate 41% more realized revenue per payer after a year than non-AI apps, but their retention is worse across weekly, monthly and annual plans. Annual retention, for example, was about 21% for AI apps versus 31% for non-AI apps. AI apps also had higher refund rates.

AI clearly helps people say yes at the paywall. Keeping them subscribed is harder.

The better consumer businesses are increasingly using AI to make an existing job easier rather than asking users to pay simply because AI is involved.

What kinds of consumer apps are actually winning now?

The strongest consumer apps solve embarrassingly simple problems and make the result arrive much faster than it used to.

Cal AI is one of the clearest examples. Instead of manually searching a food database and entering every ingredient, users photograph their meal. MyFitnessPal said Cal AI reached more than 15 million downloads and more than $30 million in annual revenue in under two years before acquiring the company.

ReciMe attacks another small annoyance. People discover recipes on Instagram, TikTok, YouTube and websites, then lose them or have to copy everything manually. ReciMe pulls those recipes into one cookbook and turns them into meal plans and shopping lists. The company now says it has more than 10 million users.

PictureThis does something even simpler: photograph a plant and find out what it is and how to care for it. Google Play now shows more than 50 million downloads. Recent Sensor Tower estimates put PictureThis at roughly 700,000 monthly iOS downloads and $9 million in monthly iOS revenue worldwide, plus around two million Android downloads and $4 million of Android revenue. Those revenue numbers are third-party estimates rather than company-reported figures, but the scale is hard to ignore.

All three products start with something people already have: a meal, a social-media post or a plant. The app removes the annoying steps between that input and the useful answer.

That pattern is much more interesting than building another app whose main pitch is “you can chat with AI about this.”

Should consumers even notice that the product uses AI?

Usually no. For mainstream users, AI works best when it removes work without creating another piece of software to learn.

Cal AI's useful feature is photographing food. PictureThis identifies plants. ReciMe imports recipes. Nobody needs to understand the models behind those actions.

We can see the opposite problem with broad assistants. A tool that opens with an empty text box quietly transfers part of the product-design problem back to the customer. The user has to decide what the tool can do, invent the request, explain the context and judge whether the answer is useful.

Normal consumer products generally work better when the interface already knows what is supposed to happen.

A photo becomes a calorie estimate. A receipt becomes a warranty reminder. A school newsletter becomes calendar events and tasks. A picture of an appliance becomes its model number, manual and maintenance schedule.

The technology can be sophisticated underneath. The promise on the App Store should probably be almost stupidly simple.

Are boring household problems becoming a real consumer AI market?

Yes. Household administration is one of the clearest consumer AI opportunities developing because several companies are independently converging on the same problem.

Cozi is useful evidence because it existed long before the current AI boom. Its standard Gold family organizer costs $39 per year. Cozi Max now costs $79 per year and adds AI tools that turn emails and photographed flyers into calendar events, create recipes and build meal plans. Cozi is effectively charging established family-organizer customers twice as much for automation.

Ohai goes further. Its household assistant processes emails, PDFs, screenshots, school schedules and calendars, then turns the information into events, tasks and reminders. Premium plans currently start at $9.99 per month. Ohai also says it can connect to around 60,000 public school calendars and keep changes synchronized automatically.

Then Fambot arrived. The company recently launched publicly after testing the product with more than 1,000 families and raising $3.5 million in pre-seed funding. Parents connect email, calendars and family communication sources, and Fambot finds deadlines, school events, things to bring and other tasks before sending a daily plan. The product is free during beta, with a paid plan expected later.

Three companies reaching roughly the same product from different starting points is more convincing than one flashy launch.

The opportunity comes from the messiness of household information. Schools send PDFs. Coaches use apps. Parents get emails. Events change. Someone sends a WhatsApp message. A paper flyer comes home in a backpack. Traditional calendar software waits for a human to clean all of that up manually.

AI can finally do a decent amount of that cleanup itself.

Is family admin already getting too crowded?

No, but we should stop pretending this is undiscovered territory. Family AI has moved from an interesting idea to an emerging category.

Cozi already owns a large installed base. Ohai is building a broad household manager. Fambot is attacking school and family logistics directly. Skylight is adding more software and AI around its physical family calendars. Apple, Google and general AI assistants can also absorb pieces of the workflow over time.

That changes what we should build.

A broad “AI assistant for families” feels too vague now. The product would enter against companies that are already building calendars, task lists, meal planning, email parsing and shared household coordination.

A sharper wedge is still available. School administration is particularly attractive because the pain repeats constantly and the incoming information is unusually messy. Parents receive newsletters, permission slips, sports schedules, birthday invitations, parent-teacher conference notices, payment requests, dress-up days and last-minute schedule changes.

Fambot's recent launch is important precisely because it confirms that other smart teams see the same opening. It also gives us a warning: if we want to build here, we should go narrower or execute faster.

“Never manually process another school message” is much clearer than “run your family with AI.”

Can a tiny team still build a huge consumer app?

Yes, although current data suggests that tiny teams win by being unusually good at product and distribution, rather than simply shipping faster.

Cal AI became a business with more than $30 million in annual revenue before its acquisition while remaining remarkably small. MyFitnessPal said seven employees joined through the deal. That is an extraordinary amount of revenue per employee for a mainstream consumer app.

Opal offers another recent example. The screen-time app had reached around $5 million in annual recurring revenue when it moved from a hard paywall toward a much more generous freemium model. According to CEO Kenneth Schlenker in a RevenueCat interview, paid conversion dropped from around 20% to 9%, but daily active users eventually climbed beyond one million and ARR passed $10 million.

Structured started as one developer's side project. The company said nearly nine million people used its daily planner during one recent twelve-month period.

But we should put these successes next to the RevenueCat base rate. Only 4.6% of newly launched subscription apps reach $10,000 in monthly revenue within their first two years. Meanwhile, nearly 70 cents of every subscription-app dollar still goes to products launched before 2020.

A small team can absolutely build something huge. “Small team” is no longer much of an advantage by itself because everybody can now build with a small team.

Do normies need to use an app every day for it to work?

No. A consumer app can work very well without daily engagement, as long as the reason to return keeps coming back.

Rocket Money is useful because bills, subscriptions and spending keep changing. Life360 remains relevant because families keep moving around. ReciMe accumulates recipes that people want again later. PictureThis becomes useful every time somebody buys a plant, sees an unfamiliar one or has a plant that starts looking sick.

The subscription data supports this distinction. RevenueCat finds that yearly plans retain roughly 20% to 40% of subscribers after a year across categories, compared with around 6% to 14% for monthly plans and only 1% to 2% for weekly plans.

Travel apps show what happens when urgency exists without enough recurrence. RevenueCat finds that Travel has the highest median trial-to-paid conversion rate at 43.5%, yet it also has the lowest median monthly revenue one year after launch, at around $35. Consumers happily pay when a trip is approaching and then stop caring.

So we do not necessarily need daily engagement. We need repeated value, accumulated personal information, useful alerts, shared household data or something else that makes returning natural.

Forcing a weekly subscription onto a product someone naturally needs four times per year will not fix weak recurrence.

Is saving money the easiest way to get normies to pay?

Saving money is extremely strong, but saving annoying work can be just as compelling when people can feel the difference immediately.

Rocket Money gives us the cleanest money example. Subscription revenue rose from $179 million in 2023 to $267 million in 2024 and $351 million in 2025. In two years, the subscription business almost doubled. Customers can easily understand why spending a few dollars to find waste across a much larger household budget might be worth it.

Life360 shows another route. Its paying customers are buying reassurance and coordination rather than a direct financial return. Life360 ended 2025 with about 95.8 million monthly active users and 2.8 million paying Circles. Subscription revenue grew 33% in a year.

Food apps sell reduced effort. Calorie logging becomes a photo. Recipe collection becomes a share button. Grocery planning comes out of meals already saved in the app.

The common thread is that the value can be explained with a concrete sentence.

“I found $80 of subscriptions I forgot about.”

“I know where my daughter is.”

“I photographed lunch instead of entering six ingredients.”

“All the dates from this school newsletter are already in my calendar.”

That kind of promise is much easier to sell than “organize your life better.”

How do you actually get a consumer app in front of normies now?

We should choose the distribution mechanism before finishing the product, because consumer discovery is currently harder than consumer app development.

Short-form video remains unusually powerful when the product produces an obvious visual transformation. Cal AI is almost designed for a TikTok: point a phone at food and watch the calorie estimate appear. Before the acquisition, its founders had built a large influencer operation around exactly that demonstration.

Some products spread through the people who use them. Life360 naturally brings several members of one family into a Circle. A shared family organizer can spread from one parent to another. ReciMe sits downstream from social recipe content, so every recipe video on TikTok or Instagram can create the need for the product.

Search works when the consumer already knows the problem. Someone can literally search “what plant is this,” which helps explain the durability of plant-identification apps. A person worried about forgotten subscriptions can search for a subscription tracker. Those are much easier acquisition moments than trying to convince somebody that they need a new abstract category of software.

Freemium can also become distribution. Opal deliberately gave away more of its core product, saw download-to-paid conversion fall sharply, and still doubled ARR because free users drove much greater adoption.

What we should avoid is a product with no obvious path besides buying Meta ads forever. With nearly 15,000 subscription apps launching in a typical month now, “we'll figure out marketing after launch” is close to saying we do not have a business yet.

Which consumer app categories are actually worth building in now?

Family administration, food, money-saving utilities and specific hobbies look much better than generic productivity or generic AI.

The big markets are attractive only when we can find a narrow entrance. Health and fitness already generates billions of dollars, but another broad fitness tracker has little reason to exist. Personal finance has proven willingness to pay, although banking integrations and consumer trust make the market harder. Hobbies can look small until we find an activity where people repeatedly need identification, recommendations or record keeping.

Category	What the evidence looks like now	A realistic small-team wedge	Our view
Family and school admin	Life360 has 95.8M MAUs; Cozi, Ohai and Fambot are all pushing further into household automation	Turn one messy stream of school or family information into actions	Best emerging opportunity, but moving quickly
Food and groceries	Cal AI passed $30M annual revenue; ReciMe has 10M+ users	Remove one repeated step such as logging, importing, planning or avoiding food waste	Excellent
Personal finance	Rocket Money generated $351M of subscription revenue in 2025	Find one recurring leak or expense people can measure	Excellent demand, harder product
Health and fitness	Around $4.5B of annual in-app spending and still growing double digits	Make one painful tracking behavior almost disappear	Huge but crowded
Gardening and hobbies	PictureThis has 50M+ Android installs and substantial current app-store revenue estimates	Identify, value, maintain or organize something hobbyists already own	Very attractive if the hobby is large enough
Family safety	Life360 generated $369M of subscription revenue in 2025	Solve one precise coordination or safety problem	Proven willingness to pay, harder trust requirements
Travel	Very high trial conversion but weak revenue one year after launch	Charge around one urgent trip problem instead of forcing permanent SaaS	Good transactional opportunity
Generic productivity	Enormous supply and weak differentiation	Only build when the “productivity” label hides one very concrete job	Mostly avoid
Generic AI	Huge usage but only a small percentage of consumers currently pay	Hide AI inside another category	Avoid as the main positioning

So what should you build and sell to normies?

We would build a very narrow household utility that takes messy real-life information and turns it into completed actions, with school administration as the first wedge.

The simplest version would handle the messages parents already hate processing. A parent connects an inbox or forwards an email, PDF, screenshot, flyer or sports schedule. The app finds the events, deadlines, payments, forms, things to bring and tasks. It then puts the right items onto the family calendar and reminds the right person at the right time.

There is now enough evidence to take this idea seriously. Cozi is charging $79 per year for a premium tier containing AI event import. Ohai starts its premium offering at $9.99 per month and is connecting directly to tens of thousands of school calendars. Fambot has just entered the market after testing with more than 1,000 families and raising $3.5 million.

As seen above, Fambot also means we should narrow the idea further than we would have a year ago. A generic family assistant already has competition. We would start with one promise: every useful thing buried inside school communications gets handled automatically.

The product could expand later. Once a family trusts it, the same system can understand summer-camp emails, sports schedules, receipts, birthday invitations, travel confirmations, household documents and maintenance reminders. But none of that belongs in the first pitch.

There is a broader pattern behind this choice. Cal AI turns a meal into a calorie log. ReciMe turns scattered recipes into something people can actually cook from. PictureThis turns a plant into an identity and a care plan. The next good consumer utilities will probably keep doing this with other pieces of ordinary life: take something messy that already exists and turn it straight into the thing the person needed.

If we are trying to sell to normies today, that is where we would look. Build something boring enough that millions of people already have the problem, narrow enough that the value is obvious in five seconds, and useful enough that nobody needs to care which AI model is underneath.