What it really costs to bring a customer back: app vs website

"People can just shop on my website."

I hear it from founders who might be on the fence about an app. And they're right. They can. Your website works, your checkout works, and nobody needs an app to buy from you.

But there's a second question that rarely gets asked: what gets them back there?

Unless you're Amazon, nobody types your URL from memory on a Tuesday afternoon. Pull up last month's returning customers and look at what brought each one back. For most brands, almost every repeat visit started with something you paid for.

This post is about the customers you already have, and the full app installed on their phone. Winning someone's first order is a different set of economics, and that's where App Clips come in. I'll touch on them further down, but the focus here is what it costs to bring people back.

Every trip back to your website has a toll

There are four main ways a past customer finds their way back to your site, and each one costs you something.

Email. Cheap per send, about a tenth of a cent. But you pay for the whole list every month, whether people open or not. A 100,000-profile list on Klaviyo runs about $1,380 a month.¹ And the inbox gets more crowded every year.

SMS. This is where the meter starts to show. A text costs about a cent once carrier fees are in, and an MMS with an image costs about 4 cents.² Say you have 40,000 subscribers and send them eight MMS campaigns a month. That's about $12,000 a month just to tap people on the shoulder.³

Paid retargeting. You already paid to acquire this customer once. Now you pay again to remind them you exist. A median Meta click for DTC brands costs about 57 cents.⁴

Branded search. Plenty of your returning customers Google your name, and plenty of brands pay to win that click so a competitor doesn't.

Some customers do come back on their own. But that's the slice you can't trigger when you need it, like the day a restock lands or a sale starts.

Here's the real issue. On the web, purchase frequency is metered. Every extra purchase occasion you want comes with a bill attached. Want customers back twice a month instead of once? Send twice as many texts. Grow your list, and your email bill grows with it. The cost of bringing someone back scales with how often you want them back.

An app removes the toll

An app changes the cost structure, not just the experience. Three things are different.

It lives on the phone. Your icon sits on a screen your customer unlocks dozens of times a day. That's a reminder you don't pay for, every time they swipe past it.

Push has no cost per send. Once someone opts in, sending one more notification costs you nothing extra. Not a cent per message, not a cost per click. You've paid for the channel already, so the tenth push of the month costs the same as the first.

Coming back takes less effort. They're already logged in. Their payment details are saved. There's no hunting for the tab or resetting a password. Less effort per repeat purchase means more repeat purchases get finished.

The result: on the web, re-engagement is a variable cost. With an app, most of it becomes a fixed one.

We can see this in how app customers actually come back. Across Reactiv apps, about two-thirds of purchase sessions start with a push tap or with no marketing link at all. Only about 1 in 20 starts with a paid ad click.⁵

To be clear, SMS and email still start about a quarter of app purchases. An app doesn't replace your retention stack. It reduces how often you have to pay to bring someone back.

And once they're back, they tend to buy more. In Reactiv customer data, a native app converts at about 2x the rate of mobile web, and an installed app lifts average order value by about 20%.

What an app doesn't do

This is the part most app vendors skip, and it's the reason a lot of smart operators don't trust app ROI claims. So let's be straight about it.

An app doesn't create demand. Your website and your app are both surfaces. They convert demand that already exists. Ads, creators, press and word of mouth create demand. If someone tells you an app will find you new customers on its own, be skeptical.

"App share of revenue" is the wrong scorecard. If a loyal customer moves their orders from your website to your app, you haven't lost anything, and the app hasn't "made" anything either. That move is how an app works, not a risk. The questions that matter are whether that customer buys more often in total, and whether bringing them back costs you less.

Watch out for the pull-forward trap. A purchase that happens sooner looks exactly like a new purchase if your window is short. A push that gets someone to buy this week instead of next month will look like lift in a 30-day report. Treat anything measured over less than about 90 days as possibly just timing.

So what does an app actually produce? Two things, and a bit of a third:

Frequency and margin. Everything an app is good at comes back to those two: more return visits, each one cheaper to drive.

There's one place the economics extend to acquisition: App Clips. A Clip is a lightweight, native version of your app that opens straight from an ad, with no download. It still doesn't create demand; your ad does that. What it changes is how much of that paid traffic converts. In Reactiv customer data, App Clips convert ad traffic at up to 3x the rate of mobile web. That's margin again, this time on the cost to acquire. (Clips run on iPhone; Android shoppers land on your mobile site.)

Clips also feed everything above. An average of 15% of Clip visitors go on to install the full app, which turns traffic you paid for once into customers you can bring back without paying per send. How Clips change acquisition costs is a post of its own (start with what an App Clip is); this one is about what happens after the first order.

The math that actually matters

Once you accept that, the question stops being a debate about causation and becomes arithmetic. An app is a fixed monthly cost, so you can work out exactly what it needs to do to pay for itself:

Here's a hypothetical. Say your app costs you $3,000 a month all in: platform plus the time your team spends running it. And say your contribution margin is $30 per order. You need 100 extra orders a month to break even.

If you have 10,000 people using the app, that's one extra order per 100 app users per month. It's a low bar, and you can check it.

Then add the margin side. Back to the 40,000-subscriber SMS list: if you moved just two of those eight monthly MMS campaigns to push, you'd save roughly $3,000 a month in sending costs.⁶ In this example, that alone covers the app before it drives a single extra order.

Your numbers will be different. That's the point: plug them in. The formula works for any brand, and the inputs are all in your own reporting.

How to test it on your own data

If you already have an app, or you're about to launch one, you don't need to take anyone's word for this, including ours. Two tests do most of the work:

Push holdout. Hold push back from a random slice of your app users for 90 days or more. Compare total spend across both groups, on every channel, not only in the app. If push is doing its job, the group that got it buys more often.

Intent-to-treat. Hold back the "download our app" prompt from a random part of your customer list. Compare total spend between the two groups, whether or not anyone installed. This shows the app's effect on your whole business, not just on the customers who were always going to install.

For both tests, measure contribution margin across all channels, not revenue the app claims credit for, and run them long enough to get past the timing trap. Channel attribution will always flatter the app. Total margin won't.

Do the math on your brand

Your website will always be a great place to shop. The question is how much you pay every time you need someone to come back to it.

Bring your contribution margin and last month's SMS and email spend. We'll work out your break-even with you, and tell you honestly if the numbers don't work. Book a call.

Related: What push notification data tells us about CRO and The BFCM checklist for brands with a mobile app.

Sources

1. Klaviyo email pricing at 95,001–100,000 active profiles, as listed on klaviyo.com/pricing (captured May 2026). Includes 1,000,000 sends a month.
2. Postscript Growth plan: $0.008 per SMS and $0.03 per MMS, plus average US carrier fees of $0.0033 per SMS and $0.008 per MMS. postscript.io/pricing, accessed October 2026. Other platforms and plans vary.
3. 40,000 subscribers × 8 sends × ~$0.038 per MMS ≈ $12,200 a month. Sent as plain SMS, it's about $3,600.
4. Median Meta CPC across 15 DTC brands, July 2025 to June 2026, all Meta traffic (not retargeting only). Top Growth Marketing.
5. Reactiv customer data, median of 14 apps with 300 or more purchase sessions, August to October 2026. A session counts as "no marketing link" when it carried no tracked campaign link. This includes customers opening the app from their home screen.
6. 40,000 subscribers × 2 sends × ~$0.038 ≈ $3,000 a month. Illustrative only.