In 2021, Apple significantly limited the data that mobile app developers could collect about users. This led to a sharp decline in the effectiveness of advertising campaigns that directed traffic directly to app pages.

The solution became web funnels, which are not subject to Apple’s restrictions. This means that instead of directing traffic to the app page, it is sent to product landing pages, whose task is to onboard the user, convert them into a purchase, and only then direct them to the app.

I spoke with FunnelFox founder Andrey Shakhtin about how and why web funnels emerged and gained popularity, who uses them, what products they work for, what their advantages and disadvantages are, and what is needed for their effective launch.

About the guest

Andrey Shakhtin is the CEO and co-founder of FunnelFox, a tool for product growth through creating and optimizing web funnels.

The context

Web funnels attract users to a mobile app by purchasing traffic not directly to the mobile app, but first to the web version of the product’s onboarding.

Here’s how it works:

  • The developer attracts to the web version through ads
  • The user lands on a page where they go through onboarding
  • In some cases, they can immediately purchase a subscription to the product in the web version
  • The user downloads the app from the store, logs in, and either purchases a subscription in the app or starts using the product right away if they purchased a subscription on the web.

There are two main reasons why app developers are switching to attracting traffic through the web:

— In 2021, Apple effectively broke traffic attribution when it changed how IDFA (the unique device identifier) worked, giving users the ability to opt out of tracking and not share the device identifier with developers. This meant that mobile app developers could no longer evaluate the effectiveness of advertising channels with sufficient accuracy.

— If the purchase happens on the web, there’s no need to pay a ~30% commission fee to app stores like the App Store or Google Play on each purchase and only need to pay the payment processor fees (~5%).

Web funnels preserve traffic attribution capabilities and allow developers to avoid app store commissions.

You can check out how web funnels work with these app examples (these are just a few screens from many). We will discuss them in more detail later in the interview.

Noom

Headway

Betterme

The origins of web funnels for apps

Key points:

  • There are two types of web funnels: Web2App and Web2Web. In both cases, traffic is directed to a web page. In Web2App, the user buys a subscription in the downloaded app, while in Web2Web, the purchase is made directly on the web.
  • Experiments with web funnels began in the mid-2010s: e-commerce products and dating services that didn’t have mobile apps were first to adopt them. A notable example of successful web funnel use was the weight loss service Noom: it used them starting in 2016, initially operating exclusively as a web service and only later launching a mobile app.
  • After changes to IDFA in 2021, mobile developers lost the ability to attribute traffic that they were buying for their apps. To “fix” analytics and attribution, developers began to more widely use web funnels, effectively creating a landing site and directing traffic to it instead.
  • The popularity of web funnels is also growing for another reason: selling a product on the web eliminates the need to pay app store commissions. Therefore, the Web2App practice spread to Android apps as well.
  • Web funnels are more commonly used in products from Eastern Europe. In Western markets, this practice is much less widespread

***

— Tell us the backstory: when did web funnels appear? Who started using them? How did they spread?

— Actually, this is quite an old practice; it just took on a new form and new application. Previously, web funnels were mostly used in e-commerce and arbitrage. Later, these practices transitioned to mobile apps.

Of the classic mobile apps, Noom was probably one of the first to actively use web funnels. Noom is a weight-loss app where you track your diet and exercise. They started as a web service but later created an app and began migrating users brought in through the web. This was around 2016 or 2017.

In general, apps had been experimenting with web funnels for a long time: some succeeded, others didn’t. But in 2021, the release of iOS 14.5 and all the changes to IDFA triggered a wave of Web2Web and Web2App adoption.

— Can you explain the difference: Web2Web, Web2App?

There are two types of web funnels:

  • Web2App — traffic is initially directed to a product’s web page, which can either be a simple landing page or an onboarding flow. The user then clicks the button to install the app and purchase the subscription within the app.
  • Web2Web — traffic is also directed to a web page, but the user purchases the subscription on the web, after which they install the app and gain access to the service.

— Do they do this on Android as well? Does traffic attribution work there too or does the store commission still play a significant role?

— Yes, the 30% commission fee plays a major role. Plus, sometimes it happens that the audience doesn’t convert in the app on Android, but they do on the web, and they convert just as well as on iOS.

— Are web funnels more common in Eastern Europe and haven’t yet caught on in the West?

— Yes, it’s mostly Eastern Europe. In Western companies, it’s much less common. In the U.S., web funnels are just beginning to be applied. The main blockers are the lack of quality information about this approach and the high entry barrier, especially if you’re building the infrastructure yourself.

How web funnels are structured

Key points:

  • A web funnel consists of four elements:
    • — user acquisition (channels/creatives)
    • — onboarding (quizzes or other formats)
    • — monetization and payment flow
    • — infrastructure: marketing tracking, analytics, subscription synchronization with the app, A/B testing, and so on.
  • It might seem that a quiz distances the user from the product’s value. But with the right design, a quiz can deliver value to the user and create an aha moment, which encourages them to make the purchase. This can be a certain result that the user receives based on their data (e.g., “you have good compatibility with Capricorns” or “you will reach your desired weight in seven months”). Or it could be useful information, like important concepts about healthy eating.
  • Another way to boost user engagement is by building trust. For example, this can be strengthened through the use of infographics with links to studies, reviews, and other tools.

***

— How are web funnels structured, what are the most important elements, and what should we pay attention to?

— The first and most important things are the creative and the channel, followed by the payment flow. Only after that comes onboarding.

Let’s go step by step.

Creatives. You need a lot of them, sometimes we’re talking about hundreds of new creatives per week if we’re talking about static creatives. Dozens, if we’re talking about video. The more and more actively you test creatives, the more chances of success in the ad channel.

Channels can vary. There’s Facebook, TikTok, and some even succeed with Pinterest. Google, for example, also has many configuration options. You can run ads in search, target keywords, or use display ads. There are also affiliate networks.

The payment flow, meaning how seamless the payment process is. Currently, 50-60% of payments are made via Apple Pay, cards account for about 20%, PayPal another 20%, and Google Pay makes up about 1%. Previously, Apple Pay’s share belonged to PayPal.

Apple Pay only started working at the end of 2022 with native Instagram or Facebook browsers, which account for a significant share of traffic. Apple Pay requires far fewer steps to complete a payment than PayPal. As a result, it has captured over half of all payments.

For a web funnel, a seamless payment flow is crucial, with no barriers for the user, an optimized checkout, and support for flexible subscription plans: a week-long trial, intro offers, upsells, subscription upgrades, and so on. All of this needs to work very smoothly. And after that, you can start focusing heavily on onboarding, which in our case is a quiz. And that’s the next growth point.

Onboarding (quiz or survey). Here, a lot depends on the product and how deeply you can tap into the user’s initial inner motivation to purchase. If you dig into psychology and the deep motivation of the person, you can uncover significant potential to drive conversion through the quiz.

— Why does the quiz work, even though intuitively it seems like it only distances the user from the product’s value?

— In fact, a quiz is a very good filter. And in the end, you have a highly motivated audience that not only buys but is also ready to keep using the product. Essentially, web onboarding can be an activation element within the product that helps people better understand the product’s value. Moreover, the quiz allows you to learn more about the product’s audience.

When people talk about removing barriers for the user, they usually mean shortening the path to the product’s value. The magic in creating web funnels happens when the team realizes that the quiz can also deliver value! For many products that use web funnels, one of the aha moments occurs precisely during the quiz stage.

Take Noom’s funnel, for example. They were the first to come up with a weight-loss predictor: you enter your current weight, how much you want to lose, and they tell you that you’ll reach your goal by a certain date. And this is the aha moment for the user.

— Can you give more examples from other products where users experience a “wow” effect?

— It could be any feedback loop where you input information about yourself and get a result. For example, in astrology, it could be compatibility with a particular zodiac sign. Fitness products might assess your fitness level based on your parameters.

Obviously, not everyone does it well, but it’s definitely possible to do it right.

In Noom’s onboarding, for example, there are other things that provide value. They ask you a question: what’s a better snack? Raisins or grapes? You might choose raisins, and then they explain why that’s not correct: grapes have more water, so you’ll eat less of them and consume fewer calories. Along with this, they explain the concept of calorie density. This, again, is an example of value for the user.

— What other tools help improve quiz effectiveness?

— In the quiz, it’s important to first engage the user and then build trust. To engage, you can use the following tools.

For example, you might start with a question like: “How old are you?” Or, “How much weight do you want to lose?” These are usually very simple questions that don’t require much thought. They’re easy to answer, so they help engage people in the funnel.

You also often see questions that help clarify the user’s JTBD (Jobs to Be Done) and guide them on a more relevant path in the quiz. In this context, data analysis from quizzes is very useful, as it allows you to identify different user segments with different needs. Then you can tailor the experience for each one, reducing time-to-value and making the funnel experience more relevant.

If the key JTBDs have already been identified, you can create separate web funnels for the largest ones and target the right audience through creative ads. To do this, you create creatives that filter the user’s intent and allow you to attract the right people.

There was also an interesting example where you give the user a choice in the creative, and after the click, you show a loading screen. This gives the person the impression that the product is being customized for them in real-time. Most people don’t really know how the creative works, and it’s a bit misleading, but this loading screen helps smooth out conversion drops early in the funnel and engages the user in completing the quiz.

We’ve engaged the users, and they’ve started taking the quiz. The next step is to build trust.

Infographics, reviews, and links to studies work very well for building trust. For example, “This is how you’ll lose weight without us, and this is how you’ll do it with us, and here’s why.”

It also works well when you give the user a customized response or offer based on the information you’ve gathered from them earlier.

— Let’s say we have a team that has never built such an acquisition mechanism before. How hard is it for them to create a good quiz on the first try? How many iterations will they need? How should the work on the quiz be structured?

— It’s tough to do from scratch, but it’s doable. Studying competitors helps a lot. Plus, ChatGPT has really simplified quiz generation.

For example, you can just ask ChatGPT, “As a fitness coach, create a 20-question survey that you would use to select workouts for a client.” You make a few variations, then select the best ones.

I’ve never been 100% satisfied with ChatGPT’s results, but it always gives a great foundation to build on.

So even if the team isn’t very experienced, they can try making a quiz on their own with these tools.

After that, the iterative work begins: refining the quiz based on users’ JTBD, optimizing specific steps to reduce user drop-offs, improving the flow, and experimenting with all the key steps.

Which products benefit most from web funnels

Key points

  • Web funnels work effectively for products that involve personalization. For example, it will work for products in the health & fitness category, where the user receives personalized assessments and recommendations during onboarding. Or in the fashion category, where, based on a person’s preferences and characteristics, the product can select a suitable style and provide recommendations.
  • There are exceptions to this rule. Even though entertainment products also have personalization, the value can only be realized within the product itself, and it’s hard to provide that during onboarding.

***

— Which products do web funnels work and don’t work for? Where do they definitely work, where do they have a chance, and where do they definitely not work?

— Let’s start broadly. Web funnels work where you can offer some level of personalization based on the user’s answers — for example, through the same quiz.

Personalization already provides value to the users. In the same quiz, for instance, you can give feedback based on their responses. And that’s already value. To sell, you need to give value, so if there’s no personalization, the web funnel works poorly.

Take weight-loss or fitness products. You can ask the user questions and, based on their answers, provide them with an assessment, recommendations, and feedback. This way, they’ve already received value and it’s easier for them to decide on a purchase even before they’ve used the actual product.

But something like a phone cleaner has only one use case, and you can’t provide any value based on personalization. Such products don’t perform well in web funnels.

— Let’s go into more detail about categories.

— Web funnels are most common in the Health & Fitness category. There are many JTBD, hooks, and opportunities to provide feedback and personalize the offer.

For example, do you want to do exercises? What exercises can you do? Do you have eating disorders? These things need to be considered during onboarding because it’s not just about personalization, but also about risks. For instance, pregnant women shouldn’t fast, and that risk needs to be filtered out during onboarding.

The horoscope category also works very well. Again, there are many use cases and personalization opportunities: when will I get married, when will I get rich, and so on. These types of products were among the first to effectively use web funnels to sell.

Another category is fashion. Everyone have their unique features: eye color, skin tone, hair color, body type, and measurements. Accordingly, the product can give you recommendations, both in terms of color palette and specific clothing items. You can personalize based on age, size, lifestyle, price, and brands.

Educational products also work well. This includes languages and microlearning platforms like Headway (they create book summaries).

— What about games?

— There’s definitely interest in this channel, and the Web2App format with landing pages is working right now. Most games are free to play with in-game purchases. This means you must port part of the game into the web funnel, which complicates both launches and subsequent optimizations.

— Are any major players in entertainment trying this? Who might find it relevant?

— When thinking about formats in entertainment that could try this, streaming apps in the short drama category come to mind. This is a new format: short series, packed into one-minute episodes. They are currently distributed as free-to-play: you start watching a few available episodes, and then monetization kicks in. But there’s definitely an approach with content personalization, and overall, they can show videos directly in the funnel.

What team and resources you need to create web funnels

Key points:

  • To launch a web funnel, you need a dedicated, focused team, which includes development, QA, a UA manager, a producer, a designer, and a product marketing manager.
  • The analytics stack for the funnel includes an integrated pixel that sends events to the ad system and product analytics tools like Amplitude or your own BI.
  • It’s important to build a growth model to understand the product’s economics and target CPA and marketing cost recovery.

***

— Suppose you have a team developing a mobile product, and they’ve decided to experiment with acquiring users through a web funnel. What kind of team do you need, who’s on it, and where are their efforts focused?

— It’s important to note that web funnels often start as a side project within the existing product team. This usually dilutes focus and doesn’t lead to success. It’s better to dedicate a separate team focused solely on the web funnel.

Now, about the people on the team.

  1. Developers: at least one front-end developer, one back-end developer, an analyst, and a data engineer if you need your own BI.
  2. A QA team, at least a minimalist one (there are many browsers, and they constantly change, breaking funnels).
  3. A team for working on the marketing product: this includes a UA manager, a creative producer, and a designer.
  4. And, of course, a product marketing manager who will design the quiz and oversee all the other parts of the process.

Our product FunnelFox allows you to launch and optimize web funnels without the need for coding, which reduces the cost and time to launch. And a selection of best practices helps you launch funnels with good metrics right from the start.

— So, you’ve gathered such a team and built the first version of the funnel. What next? What does this team do in a typical week?

— First, you need to establish a creative production pipeline. There’s a separate creative team: a creative producer, designer, and possibly external agencies. They should have a process where they generate hypotheses and launch creatives for testing weekly. This work gets quickly templated. There are different phases for testing creatives, and so on. The UA manager is also involved in this, as is the product that’s building the funnel.

QA needs to make sure everything works across different browsers. It’s not obvious, but most traffic will come through built-in browsers from Instagram and Facebook apps, which also get updated frequently. Many teams forget this and only optimize for Safari and Chrome.

The creative team’s KPI is built around the number of tested creatives and the success rate of those creatives. These KPIs must be maintained.

Now, down to the UA Manager, who works on the channel level. The work and testing process depends heavily on the main growth channels. The KPIs are usually the volume of acquisition at the target return on campaigns.

The product marketing manager’s KPI is the number of successful experiments they run in the funnel. Experiments are conducted at different levels: they should have separate backlogs for monetization, the quiz, and so on.

If the funnel starts to gain traction and scale, you can recruit multiple product managers to focus on different aspects, such as monetization, the quiz, and so on.

To sum up, everyone involved in this project is responsible for launching tests and experiments with a clear rhythm. Usually, this means one or two validated hypotheses per week at each funnel level.

— When you were working, how many experiments were being run at each level per week?

— We had two to three experiments running in the funnel itself. The creative process and channels are separate — each campaign and creative is essentially an experiment.

Most of the tests in the funnel were related to the quiz because that’s where there were more things to test.

There were fewer tests around monetization, paywalls, and checkouts. We might have tested those once a quarter. We might have seen significant gains there, but they would not be continuous. It might have given us a 20% boost, and then went quiet.

However, working on the quiz level can give you a month-to-month growth of 10% as you optimize for JTBD, implement local optimizations, and so on.

— What analytics stack is needed for such a team?

— First of all, attribution. You need to integrate a pixel, and you need to send key funnel events through the pixel.

Since a web funnel is technically a website, you can embed an ad pixel for attribution. A pixel is basically a script that sends frontend events to the ad system.

Some ad networks provide the ability to send events from the backend since cookies can be erased by browsers. Meta, for example, has the Conversion API (CAPI).

Key events include successful registration, purchase initiation, and successful purchase. But teams can customize this based on their needs and add more funnel events.

After that, it all depends on how you want to build optimization. Most people optimize for conversion. For that, you just need to send purchase events: the purchase of the main subscription, upsell purchases — all of that should be sent.

You can go deeper and optimize for value, where you send both the purchase event and its price. But this is a more complex approach and requires more data.

For product analytics, most use Amplitude or their own BI. Amplitude captures events, screen views, and all quiz answers. You can build the same in your own BI if you have a large volume of data and Amplitude becomes too expensive.

You absolutely need to build a growth model to understand payback periods and the overall product economics. This model should give you your target CPA, and you give this target CPA to the UA manager so they can buy traffic with that target.

For A/B testing, teams used to rely on Google Optimize. It’s now closed, but there are paid alternatives like Convert or Optimizely. At FunnelFox, we also have built-in A/B testing: you can create funnels and immediately run tests.

— How can you track subscription revenue?

— You can track revenue in analytics systems like Amplitude. Adapty also has good analytics for revenue and it integrates well with the web.

In general, it’s best to handle all the financial calculations and analytics in-house. It’s worth measuring everything several times.

If you’re selling through Stripe or PayPal, you can reconcile with them: track events with them and send them through your backend.

— What payment options can be set up in a web funnel?

— The main options include Apple Pay, credit cards, and PayPal. More local and exotic options are Venmo, Cash App, Sepa/Direct debit. The process is implemented via integration with payment providers like PayPal, Stripe, or Paddle.

— What are the differences between payment providers?

— Here’s how it usually works. For example, if you’re using Stripe, you’re the main seller, and Stripe acts only as a payment provider. This means you’re fully responsible for sales, taxes, invoicing, and so on.

But there are also intermediaries on the market, known as Merchants of Record. Essentially, the user is buying the subscription not from you, but from them. They take a cut, around 5%. Incidentally, Stripe also ends up with around 5% after all fees.

Merchants of Record handle taxes, send invoices, and manage notifications for you. However, they tend to perform worse in terms of conversion. If you’re willing to deal with invoices and such from the start, I’d recommend Stripe.

It’s also better to have multiple payment systems because you might spend $100,000 on ads, and then Stripe blocks your account. It’s worth keeping that in mind.

— What percentage of users actually download the app after purchasing a subscription?

— About 90-95% if everything is done well!

— Why do apps exist in this model? Why not just stay on the web? Or is it an expectation from users that there will be an app? It’s extra code for the developer.

— That’s an interesting question. We have clients who launched on the web and are already driving a lot of traffic, but are just now developing an app. I think it’s a growth-hacking way of figuring out what you need to build.

In reality, if your product is content-centric you might not need an app at all. You can deliver the content through the web or even just email newsletters or PDFs. But these options work best at an early stage, and for a better user experience, it’s worth creating a mobile or web product.

I think the main argument in favor of having an app is user engagement because it’s always visible, accessible with one touch, and push notifications make it easy to remind users about you.

On the future of web funnels and their impact on product managers’ work

Key points:

  • Teams that actively use web funnels are directing fewer resources toward creating value within the product and more toward optimizing user acquisition.
  • Competition in Web2App is becoming more aggressive, which is driving up auction costs.
  • Apple is closely monitoring the trend but hasn’t intervened yet. Two possible scenarios could occur: either Apple will take targeted action against select publishers or they will integrate into the process and find a compromise with developers.

***

— I have a philosophical question for you. Considering that such a funnel essentially locks the user into a purchase within the funnel itself, it feels like the product becomes less important. The user still accesses it after the purchase. How accurate is that idea? Have you noticed any trends in the industry where resources that used to go toward improving the product and user experience are now flowing toward optimizing acquisition processes?

— There’s definitely a shift happening. I’d say the product needs to have a basic level of quality to avoid a lot of refunds. If you promise one thing in the funnel, but the product doesn’t deliver, it won’t work. There will be many complaints, etc. So, the product needs a baseline level of quality that matches what you promised during user acquisition.

But pushing metrics further through product improvements is hard. And there’s definitely a trend where no matter how much the product improves, it doesn’t impact business metrics. I see this very often.

— Are there examples of products that have nailed both the product experience and user acquisition?

— I think there are almost none. Even though the funnel itself is partially a product, it’s more about product marketing. It’s rare to find a team that has both strong product skills and a solid product vision. But I believe it’s possible to combine them.

In fact, there are apps with good retention that could effectively use web funnels. Duolingo, for example.

Duolingo used to be very performance-driven. They sold subscriptions during onboarding and did it quite aggressively. At some point, they strategically shifted: they brought in people from the gaming industry and completely restructured the product to a free-to-play model. Then they invested in branding and social media.

Now, Duolingo does have a web funnel, but it’s not their main or only acquisition channel.

— You’ve been watching this space for a long time. Can you paint a broad picture of how it’s changing and where it’s headed? What are the main vectors of development?

— Competition is leading web funnels to become more and more aggressive. At first, they were just selling subscriptions. Then they started offering discounts on the first purchase. Then upsells were added. Over time, they started raising the price of subscriptions.

This isn’t great because not everyone is willing to act aggressively, but they still have to fight for ad auction space. Even apps with valuable products are starting to use similar mechanisms: offering upsells, discounts, timers, etc.

At the same time, interesting onboarding processes and quizzes are emerging that genuinely offer value. Overall, I think this is a good tool that gives developers a fair shot at growth, providing more freedom to operate without being limited by the sometimes unreasonable demands from Apple.

— It seems like this whole web funnel trend is bad for Apple. They’re losing money, their platform is becoming less central. Moreover, Apple originally created the review system to control the user experience, but their own IDFA initiative has led to the opposite result, where key steps in the funnel have moved to the web. Do they know about this issue, and are they doing anything about it?

— Yes, they know and definitely want to address it, especially if the trend reaches game developers. And there are two potential approaches: either they will try to suppress this practice, or they will find a way to integrate it into the process and control it.

They might come up with some legal justifications to get involved in this flow and set a 10-15% commission, which would ultimately be a compromise for everyone. The industry has long been moving toward alternative payment methods, and it’s clear that Apple doesn’t quite know how to enter this space without losing money.

Regarding the web, at the global level, I think there will be more regulation. It’s already happening. For example, the Federal Trade Commission (FTC) recently required major financial companies to ensure that online sellers don’t engage in price discrimination.

Right now, there’s virtually no price discrimination in the market because it’s well known that you can get into serious trouble for that. But the very fact that regulators are paying attention indicates that subscription sales on the web will be subject to more regulation.

— What do you think about Apple deciding to remove apps that actively use web funnels? I understand the formal reason is that people get confused about where they subscribed: they don’t understand what’s an app and what’s the web. How common is this reason? How high are the risks that those using web funnels lose their app?

— There’s currently no widespread practice of Apple removing apps. And there’s no official justification that it’s due to web funnels. Sometimes they may notice user complaints where people can’t cancel their subscriptions through the app. If there are many complaints, Apple checks the app and sees that not many subscriptions went through them. In that case, they just ask for fixes.

There have been controversial cases of apps being removed from the App Store, but there hasn’t been a clear reason stating it was due to external payments. It could be because of them, but the formal justification is always something else.

I don’t think Apple will start removing apps en masse: it’s risky for them and they might face a new class-action lawsuit. If that happens, developers will split into two camps:

– Those building venture-backed businesses and focusing on capitalization will turn off web funnels and stay within the apps.

– Those looking to make money will simply ditch the apps, create web apps, and keep driving traffic to them because growth through the App Store is becoming increasingly limited and harder to achieve.

— For those using web funnels, does it even make sense to experiment with regular campaigns that lead directly to the app? 

— It’s always a question of resources. You have a team, especially in the early stages, and you can’t do everything. From a realistic perspective, I wouldn’t recommend focusing on optimizing the app and would prioritize web funnels. But if you have the resources, then of course, you should also run campaigns for the app, because there’s potential there too.

— Great. Let’s wrap up the conversation.

To end on a positive note, let’s put it this way: web funnels are a new growth trend for mobile apps. This approach involves a lot of nuances and subtleties but definitely offers a ton of advantages and levers for growth.

Illustration by Anna Golde for GoPractice