Igor Lyubimov
CEO of web2wave
Where to hear and read about Igor Lyubimov4 episodes and tests
App MastersCopy Headway’s $100M+ Web2App Funnel Strategy (That Converts Like Crazy)
App MastersCopy This Meta Ads Strategy to Scale Your Web2App Campaigns (Part 2/2)
App MastersThe Beginner's Guide to Meta Ads for Web2App Campaigns (Part 1/2)
App MastersThe Anatomy of a High-Converting PaywallIgor Lyubimov is the CEO of web2wave, a platform for web-to-app quiz funnels, and he teaches subscription apps how to sell on the web instead of through the app stores, from quiz psychology and paywalls to Meta ads tracking and scaling. The most useful lesson from him is that the web funnel is an all-in sale: warm the user with a quiz about them, answer every doubt on a long paywall, take money early, and judge it all by return on ad spend.
People are paying not for the product, they're paying for solutions of their problem, and they want to understand when their problem will be solved.
The Anatomy of a High-Converting Paywall
Igor Lyubimov: the story and the lessons
1Igor Lyubimov and web2wave
Lyubimov runs web2wave, which powers web-to-app funnels for subscription apps. web2wave hosts quiz funnels and paywalls that sell a subscription on the web before sending the user into the mobile app. He speaks from roughly 60 to 70 client projects.
He estimates 80 to 90% of the traffic to web2wave funnels comes from Meta ads.
He sees web-to-app as a lasting market, not a passing trend. He estimates 5 to 10 billion dollars a year flows through these funnels for mobile apps and similar subscription businesses.
His evidence is client demand, conference talks and the volume of YouTube content on the subject, and he notes it has been happening for years.
He argues web unit economics beat mobile. On the web there is no 30% store fee, money arrives in two days instead of six weeks, and campaigns scale faster.
Web also allows discounts, timers, upsells and exit pop-ups that store rules forbid or penalize.
2What Lyubimov learned from Headway's quiz funnel
Copy a proven funnel structure instead of inventing a new interface. Headway, said to earn hundreds of millions a year from a simple, heavily tested funnel, tried a chatbot version of its quiz. Lyubimov saw it for about a month and never again, so he suspects it failed.
He advises spending testing effort on copy, offers and flow. "Do not try to invent the wheel and try new UI for the funnels."
The first screen sells the outcome, not the app. Headway opens with "become the most interesting person in the room" rather than "welcome to Headway", labels it a three minute quiz, then asks something easy like age.
He recommends a question on the first screen over a bare Get Started button.
The quiz should talk about the user, not the product. Most of Headway's questions, on self-doubt, decisions and work-life balance, have nothing to do with book summaries. Every answer gets a supportive reply, so the quiz feels like a conversation with a psychologist.
Identity questions like "big picture or detail oriented" or "leader or follower" trigger a labeling effect, and the next screen changes with the answer. His sample reply to self-doubt: if you don't make mistakes, you don't make anything.
Imply features through questions instead of claiming them. Asking whose success inspires you (Steve Jobs, Richard Branson, LeBron James) made Lyubimov himself assume the app had their content. "Do you like to learn by listening?" hints at audio without selling it.
Reward screens and filled loaders keep people answering. Bright congratulation screens give "a small dose of dopamine" between dull questions. His clients always saw higher drop-off on progress-bar-only screens than anywhere else.
He fills loaders with a review, a content peek or a small question, and reassures users on the email screen with awards or a result preview and a no-spam promise.
The more spontaneous the purchase, the longer the funnel. Headway's quiz runs about 50 screens, and people who finish long funnels are more likely to pay. Utility subscriptions use short funnels.
Education and fitness buyers need warming up, with their problem made vivid and their answers heard.
Social proof runs throughout, but only claims you can back. Headway shows "40 million people", press logos and "5 million users in your location". Lyubimov warns many copycat funnels show the same logos and are lying.
3How Lyubimov builds a web paywall
A web paywall is an all-in last chance to sell. A mobile paywall has about five blocks because the user already installed the app. A web paywall needs about 14 blocks, as leaders like BetterMe use, because an unsold visitor is probably gone for good.
His test: screenshot your paywall and "if they don't have like 14 blocks then add more." He treats the last two or so quiz screens as part of the paywall.
He reflects quiz answers so the plan feels made for the user. The paywall repeats age, gender and goals, and can swap graphics and copy by answer, such as a matching graphic for someone who picked "improve self-belief".
Promise a concrete result by a date and show it. An example commitment is losing 10 kilograms by September, backed by a progress graph and before and after visuals.
Illustrations with real people convert better than abstract graphics.
Explain the product only at the paywall, with trust built in. Users who arrived from a reel minutes ago have seen no app reviews, so the FAQ explains what they get, how charging, cancelling and refunds work, and shows a 30 day money back guarantee in several places.
"On the web, they know nothing about your payment methods."
He benchmarks paywall conversion at 6%. Average paywall to active subscription is 6%, with a range of 2 to 15%, against about 2% install to paid on mobile. Losing 70 to 80% of visitors during the quiz is fine.
Some web2wave projects reach 15% on the paywall with only 20 to 30% finishing the quiz.
Fix a weak paywall in a set order. Content first, then the payment form: drop the payment terms line at launch to get events faster and add it back if chargebacks grow, and put Apple Pay right after the prices.
Then check Stripe approval rates, especially cross-border. He suggests a legal entity in the users' country, looser fraud rules, or orchestration like Primer.io, or Paddle, which handles taxes but is less customizable.
4Lyubimov's pricing rules: big discounts, paid trials, upsells
Open with at least 50% off, because small discounts are ignored. His experiments showed that the bigger the discount, the better, with 50 to 60% working well. One six month plan showed a $200 discount and a $66 total.
He admits inflating the comparison price is a gray area but says it converts. Timers work on web and on mobile, especially Android.
He does not recommend free trials on the web. Free trials convert to paid two to three times worse than on mobile, and optimizing Meta on the trial event brings people who never pay. He prefers a paid intro such as 9 cents for three days.
Headway sells $19.98 with the price rising after three months. One client's 99 cent first week renewing at the weekly price works very well. His rule: "start selling and getting money as soon as possible."
Show per-month prices when per-day looks too small. $11 a month is a normal price, so he would show it monthly rather than daily.
Per-day pricing can confuse buyers into thinking a six month plan never renews.
Upsells, not the first offer, produce the profit. Stripe keeps the payment token, so each upsell charges in one click. He says upsells can raise LTV by 20 to 30% easily, and the goal is to break even on the initial offer.
Each upsell needs a different price so users don't think they were charged twice; Headway's are 14.59 and 9.59. Offers can chain: pricier after a yes, cheaper after a no.
Send checkout abandoners to the same paywall at a lower price. When a user closes the payment form, a pop-up offers a bigger discount and leads to the same paywall with an extra 25% off applied. On one app with about $15,000 MRR, a downsell added $1,000 to $2,000 a month in its first month.
He says the gain is smaller than from upsells, but most big apps use it.
Charge in-app purchases with the card saved on the web. In the US, an app can now charge the web-saved payment method inside the app with one API call, avoiding the 30% App Store fee. web2wave built this into its SDK.
Headway shows another upsell when the app first launches.
5How Lyubimov recovers and keeps buyers after the funnel
Payment should land the user in an already activated app. A deferred deep link on the final screen opens the app with the subscription active, with no password or copying.
Ask any extra questions, like confirming the email, before the final screen. "You want it to be like one click flow."
Eight winback emails with climbing discounts chase quiz leavers. Headway emails everyone who left an email but did not pay, eight times, never more than two or three days apart, with discounts up to 90%. Lyubimov says email usually brings 10% of sales or less, and 20% is a miracle.
Since payment odds fall over time, a 90% or 95% discount costs little.
Each email carries one message and one button. The first asks to confirm registration, with an emoji subject and a button to the paywall. The next just says the plan is ready, up to 70% off.
Later emails look like a personal note rather than a designed newsletter.
He tracks whether buyers actually open the product. Buyers who never start using the app are far more likely to request a refund or file a chargeback.
6How Lyubimov sets up Meta tracking for web-to-app
Send every event from the web funnel, never from the app. Many new clients want to send purchases from the app. In his flow the deferred deep link (Adjust, AppsFlyer, Branch) passes the web user ID into the app and RevenueCat syncs the subscription.
Optimize on standard events, mainly Purchase. Meta has billions of standard events and knows what users do before buying. Only one or two of his 60 to 70 clients chose a custom event.
"The algorithm will be way more better when you optimize onto standard events." Custom events serve as side metrics.
Send events from both the browser pixel and the server. Ad blockers, private relay and some VPNs block the pixel, so the Conversion API fills the gap while browser events arrive faster.
Setup is one dataset, a script on every page and an access token.
Deduplicate and enrich every event. Send the same event ID, FBC and FBP from browser and server, or Meta counts two purchases and misallocates budget. Add email, name, location, gender and external user ID.
If the quiz never asked for a name, take it from Apple Pay. Currency and amount can go on pre-payment events like initiate checkout.
Tag links and verify events before spending. More UTM tags give results down to the creative level, and the Meta Pixel Helper Chrome extension shows exactly what each event sends.
Expect about 2% of visitors to buy. About 50% leave on the first screen, 20 to 30% finish the quiz, and 4 to 8% of paywall viewers buy, sometimes 10 to 12%.
These steps are proxies for diagnosis. "You always have to focus on return on ad spend."
7How Lyubimov launches and scales Meta campaigns
Start on a cheap proxy event, then switch to Purchase. On day one the pixel doesn't know who buys, and purchase optimization can lose hundreds or a thousand dollars. About $50 a day for 7 to 10 days on an earlier event trains the pixel.
Proxy events give about 10 times more signals and reveal bugs and drop-offs on a small budget.
Test 15 to 30 creatives at about $100 each. Build them from 2 to 4 audiences, 1 to 3 hooks each and 1 to 3 formats per hook. Usually 60 to 70% fail and the top 20 to 30% carry the account.
Give each creative at least three days, cut ads that spend without purchases or get no spend, and retest the losers every couple of months.
Make about 70% of new ads variations of winners. Vary top performers by idea, format, color or structure, and add new concepts every couple of months as audiences tire.
Scale slowly and duplicate rather than edit. Raise budgets 20 to 25% every few days, never two or three times at once. A copy of a winning campaign in the same account performs like the original.
Editing targeting restarts learning, so start in the United States and copy the campaign to new countries.
Leave low-budget campaigns alone on bad days. Under about $1,000 a day, bad days happen. Stop a campaign only after two or three weak days, and never scale on a bad day.
Judge everything on ROAS, using predicted LTV early. Without day-one data, estimate LTV, for example 50% of monthly subscribers renewing in month two and about 25% in month three, then track revenue retention by channel.
Lookalikes come only with a strong customer base, and he masters Meta before adding TikTok or Google.
Copy Headway’s $100M+ Web2App Funnel Strategy (That Converts Like Crazy)
A step by step teardown of Headway's web to app funnel, covering the quiz psychology, paywall, downsell, upsells, deferred deep link and eight email winback sequence. Headway is said to earn hundreds of millions of dollars a year from this simple funnel, so it is a model to copy.
Copy a proven funnel structure instead of inventing a new interface. Headway, said to earn hundreds of millions a year from a simple, heavily tested funnel, tried a chatbot version of its quiz. Lyubimov saw it for about a month and never again, so he suspects it failed.
He advises spending testing effort on copy, offers and flow. "Do not try to invent the wheel and try new UI for the funnels."
The first screen sells the outcome, not the app. Headway opens with "become the most interesting person in the room" rather than "welcome to Headway", labels it a three minute quiz, then asks something easy like age.
He recommends a question on the first screen over a bare Get Started button.
The quiz should talk about the user, not the product. Most of Headway's questions, on self-doubt, decisions and work-life balance, have nothing to do with book summaries. Every answer gets a supportive reply, so the quiz feels like a conversation with a psychologist.
Identity questions like "big picture or detail oriented" or "leader or follower" trigger a labeling effect, and the next screen changes with the answer. His sample reply to self-doubt: if you don't make mistakes, you don't make anything.
Imply features through questions instead of claiming them. Asking whose success inspires you (Steve Jobs, Richard Branson, LeBron James) made Lyubimov himself assume the app had their content. "Do you like to learn by listening?" hints at audio without selling it.
Reward screens and filled loaders keep people answering. Bright congratulation screens give "a small dose of dopamine" between dull questions. His clients always saw higher drop-off on progress-bar-only screens than anywhere else.
He fills loaders with a review, a content peek or a small question, and reassures users on the email screen with awards or a result preview and a no-spam promise.
The more spontaneous the purchase, the longer the funnel. Headway's quiz runs about 50 screens, and people who finish long funnels are more likely to pay. Utility subscriptions use short funnels.
Education and fitness buyers need warming up, with their problem made vivid and their answers heard.
Social proof runs throughout, but only claims you can back. Headway shows "40 million people", press logos and "5 million users in your location". Lyubimov warns many copycat funnels show the same logos and are lying.
Explain the product only at the paywall, with trust built in. Users who arrived from a reel minutes ago have seen no app reviews, so the FAQ explains what they get, how charging, cancelling and refunds work, and shows a 30 day money back guarantee in several places.
"On the web, they know nothing about your payment methods."
Open with at least 50% off, because small discounts are ignored. His experiments showed that the bigger the discount, the better, with 50 to 60% working well. One six month plan showed a $200 discount and a $66 total.
He admits inflating the comparison price is a gray area but says it converts. Timers work on web and on mobile, especially Android.
He does not recommend free trials on the web. Free trials convert to paid two to three times worse than on mobile, and optimizing Meta on the trial event brings people who never pay. He prefers a paid intro such as 9 cents for three days.
Headway sells $19.98 with the price rising after three months. One client's 99 cent first week renewing at the weekly price works very well. His rule: "start selling and getting money as soon as possible."
Upsells, not the first offer, produce the profit. Stripe keeps the payment token, so each upsell charges in one click. He says upsells can raise LTV by 20 to 30% easily, and the goal is to break even on the initial offer.
Each upsell needs a different price so users don't think they were charged twice; Headway's are 14.59 and 9.59. Offers can chain: pricier after a yes, cheaper after a no.
Send checkout abandoners to the same paywall at a lower price. When a user closes the payment form, a pop-up offers a bigger discount and leads to the same paywall with an extra 25% off applied. On one app with about $15,000 MRR, a downsell added $1,000 to $2,000 a month in its first month.
He says the gain is smaller than from upsells, but most big apps use it.
Charge in-app purchases with the card saved on the web. In the US, an app can now charge the web-saved payment method inside the app with one API call, avoiding the 30% App Store fee. web2wave built this into its SDK.
Headway shows another upsell when the app first launches.
Payment should land the user in an already activated app. A deferred deep link on the final screen opens the app with the subscription active, with no password or copying.
Ask any extra questions, like confirming the email, before the final screen. "You want it to be like one click flow."
Eight winback emails with climbing discounts chase quiz leavers. Headway emails everyone who left an email but did not pay, eight times, never more than two or three days apart, with discounts up to 90%. Lyubimov says email usually brings 10% of sales or less, and 20% is a miracle.
Since payment odds fall over time, a 90% or 95% discount costs little.
Each email carries one message and one button. The first asks to confirm registration, with an emoji subject and a button to the paywall. The next just says the plan is ready, up to 70% off.
Later emails look like a personal note rather than a designed newsletter.
Copy This Meta Ads Strategy to Scale Your Web2App Campaigns (Part 2/2)
Explains how to run and scale Meta Ads for web2app funnels: start on a proxy event, move to purchase optimization, test creatives in bulk, then scale by duplicating campaigns. Gives concrete budgets, thresholds and pacing rules, such as $100 per creative and 20 to 25% budget raises every few days.
Start on a cheap proxy event, then switch to Purchase. On day one the pixel doesn't know who buys, and purchase optimization can lose hundreds or a thousand dollars. About $50 a day for 7 to 10 days on an earlier event trains the pixel.
Proxy events give about 10 times more signals and reveal bugs and drop-offs on a small budget.
Test 15 to 30 creatives at about $100 each. Build them from 2 to 4 audiences, 1 to 3 hooks each and 1 to 3 formats per hook. Usually 60 to 70% fail and the top 20 to 30% carry the account.
Give each creative at least three days, cut ads that spend without purchases or get no spend, and retest the losers every couple of months.
Make about 70% of new ads variations of winners. Vary top performers by idea, format, color or structure, and add new concepts every couple of months as audiences tire.
Scale slowly and duplicate rather than edit. Raise budgets 20 to 25% every few days, never two or three times at once. A copy of a winning campaign in the same account performs like the original.
Editing targeting restarts learning, so start in the United States and copy the campaign to new countries.
Leave low-budget campaigns alone on bad days. Under about $1,000 a day, bad days happen. Stop a campaign only after two or three weak days, and never scale on a bad day.
Judge everything on ROAS, using predicted LTV early. Without day-one data, estimate LTV, for example 50% of monthly subscribers renewing in month two and about 25% in month three, then track revenue retention by channel.
Lookalikes come only with a strong customer base, and he masters Meta before adding TikTok or Google.
The Beginner's Guide to Meta Ads for Web2App Campaigns (Part 1/2)
A beginner's guide to the technical setup of Meta ads for web-to-app funnels: which events to optimize on, how to send them from pixel and server, and what funnel conversion rates to expect. The payoff is a clear benchmark chain (about 2% visitor to purchase) and a tracking setup that avoids double-counted events and weak optimization.
Lyubimov runs web2wave, which powers web-to-app funnels for subscription apps. web2wave hosts quiz funnels and paywalls that sell a subscription on the web before sending the user into the mobile app. He speaks from roughly 60 to 70 client projects.
He estimates 80 to 90% of the traffic to web2wave funnels comes from Meta ads.
He sees web-to-app as a lasting market, not a passing trend. He estimates 5 to 10 billion dollars a year flows through these funnels for mobile apps and similar subscription businesses.
His evidence is client demand, conference talks and the volume of YouTube content on the subject, and he notes it has been happening for years.
He tracks whether buyers actually open the product. Buyers who never start using the app are far more likely to request a refund or file a chargeback.
Send every event from the web funnel, never from the app. Many new clients want to send purchases from the app. In his flow the deferred deep link (Adjust, AppsFlyer, Branch) passes the web user ID into the app and RevenueCat syncs the subscription.
Optimize on standard events, mainly Purchase. Meta has billions of standard events and knows what users do before buying. Only one or two of his 60 to 70 clients chose a custom event.
"The algorithm will be way more better when you optimize onto standard events." Custom events serve as side metrics.
Send events from both the browser pixel and the server. Ad blockers, private relay and some VPNs block the pixel, so the Conversion API fills the gap while browser events arrive faster.
Setup is one dataset, a script on every page and an access token.
Deduplicate and enrich every event. Send the same event ID, FBC and FBP from browser and server, or Meta counts two purchases and misallocates budget. Add email, name, location, gender and external user ID.
If the quiz never asked for a name, take it from Apple Pay. Currency and amount can go on pre-payment events like initiate checkout.
Tag links and verify events before spending. More UTM tags give results down to the creative level, and the Meta Pixel Helper Chrome extension shows exactly what each event sends.
Expect about 2% of visitors to buy. About 50% leave on the first screen, 20 to 30% finish the quiz, and 4 to 8% of paywall viewers buy, sometimes 10 to 12%.
These steps are proxies for diagnosis. "You always have to focus on return on ad spend."
The Anatomy of a High-Converting Paywall
Web-to-app paywalls work best as long, personalized, objection-handling pages with about 14 blocks, unlike short mobile paywalls. Average paywall-to-subscription conversion is 6% on web, roughly three times the ~2% install-to-paid rate on mobile.
He argues web unit economics beat mobile. On the web there is no 30% store fee, money arrives in two days instead of six weeks, and campaigns scale faster.
Web also allows discounts, timers, upsells and exit pop-ups that store rules forbid or penalize.
A web paywall is an all-in last chance to sell. A mobile paywall has about five blocks because the user already installed the app. A web paywall needs about 14 blocks, as leaders like BetterMe use, because an unsold visitor is probably gone for good.
His test: screenshot your paywall and "if they don't have like 14 blocks then add more." He treats the last two or so quiz screens as part of the paywall.
He reflects quiz answers so the plan feels made for the user. The paywall repeats age, gender and goals, and can swap graphics and copy by answer, such as a matching graphic for someone who picked "improve self-belief".
Promise a concrete result by a date and show it. An example commitment is losing 10 kilograms by September, backed by a progress graph and before and after visuals.
Illustrations with real people convert better than abstract graphics.
Explain the product only at the paywall, with trust built in. Users who arrived from a reel minutes ago have seen no app reviews, so the FAQ explains what they get, how charging, cancelling and refunds work, and shows a 30 day money back guarantee in several places.
"On the web, they know nothing about your payment methods."
He benchmarks paywall conversion at 6%. Average paywall to active subscription is 6%, with a range of 2 to 15%, against about 2% install to paid on mobile. Losing 70 to 80% of visitors during the quiz is fine.
Some web2wave projects reach 15% on the paywall with only 20 to 30% finishing the quiz.
Fix a weak paywall in a set order. Content first, then the payment form: drop the payment terms line at launch to get events faster and add it back if chargebacks grow, and put Apple Pay right after the prices.
Then check Stripe approval rates, especially cross-border. He suggests a legal entity in the users' country, looser fraud rules, or orchestration like Primer.io, or Paddle, which handles taxes but is less customizable.
Open with at least 50% off, because small discounts are ignored. His experiments showed that the bigger the discount, the better, with 50 to 60% working well. One six month plan showed a $200 discount and a $66 total.
He admits inflating the comparison price is a gray area but says it converts. Timers work on web and on mobile, especially Android.
He does not recommend free trials on the web. Free trials convert to paid two to three times worse than on mobile, and optimizing Meta on the trial event brings people who never pay. He prefers a paid intro such as 9 cents for three days.
Headway sells $19.98 with the price rising after three months. One client's 99 cent first week renewing at the weekly price works very well. His rule: "start selling and getting money as soon as possible."
Show per-month prices when per-day looks too small. $11 a month is a normal price, so he would show it monthly rather than daily.
Per-day pricing can confuse buyers into thinking a six month plan never renews.
Upsells, not the first offer, produce the profit. Stripe keeps the payment token, so each upsell charges in one click. He says upsells can raise LTV by 20 to 30% easily, and the goal is to break even on the initial offer.
Each upsell needs a different price so users don't think they were charged twice; Headway's are 14.59 and 9.59. Offers can chain: pricier after a yes, cheaper after a no.
Send checkout abandoners to the same paywall at a lower price. When a user closes the payment form, a pop-up offers a bigger discount and leads to the same paywall with an extra 25% off applied. On one app with about $15,000 MRR, a downsell added $1,000 to $2,000 a month in its first month.
He says the gain is smaller than from upsells, but most big apps use it.