Tired of manually modifying funnels and spending weeks on each experiment, we wanted something similar.

Back then, companies like FunnelFox were just getting started, so making something in-house sounded reasonable. I still met with the founder of FunnelFox to see what they offered (and steal some ideas, sorry), but quickly decided to stick to our own thing. I’m glad we did.

Two years in, we have an amazing builder that serves funnels across web, iOS and Android. At the time of writing, 20 different quizzes are getting traffic.

Building such a thing from scratch gave me a lot of exposure to how web funnels work and what their benefits are. So when it came to promoting Emery, my choice was simple. And while I only have a single funnel and no builder, applying this knowledge helps me get customers at half the price of the big guys.

Nowadays, whenever I consult for companies on web funnels/attribution, I find myself explaining the same things again and again.

With this article, I want to share some nuances, technical or otherwise, that are often overlooked.

But first, let’s agree on the terminology I’m going to use.

• App2App – users click on an ad and are sent directly to the app store.

• Web2App – users are first sent to a weblink, then quickly redirected to the app store. Sounds awkward, but it’s a trick used by many companies to help with attribution. Deserves a separate article though.

• Web2Web – users start their journey on the web, complete the payment, and can then be sent to the app – either the store or web version.

This naming is the cleanest I’ve seen so far, as it covers both critical checkpoints – where the user journey begins and where the payment happens.

This article is dedicated to web2web funnels for B2C SaaS. Mostly for Meta, but some of it is relevant to other ad platforms.

The funnel in this context means a quiz – a series of questions and promo screens before presenting users with a paywall.

Now, with terminology out of the way, let’s see what people usually miss when creating their first web2web funnel.

On Meta, purchase campaigns only

Many B2C apps offer a free trial and copy that offering into their first web funnel.

However, it doesn’t work with Meta ads. Meta is too good when it comes to optimising to your signals. The algorithm always takes the path of least resistance. If you ask it to bring users who activate a trial, it’ll do just that. However, most of them won’t convert. There is a certain group of people who start trials across many apps, but never pay. And that’s who you’re targeting in this case. Doesn’t matter if you ask for a credit card or not.

It’s especially true for web2web funnels. On app2app funnels, downloading an app acts as a natural filter for a low-intent audience. People who end up going through your in-app funnel are already motivated enough to pay. Besides, app campaigns are much slower to learn and adjust to a cheaper audience.

Adjust prices

The nature of Meta traffic is very different from, let’s say, Google or organic.

Most Meta ad purchases are highly emotional. Which is great, as it makes high intent optional for a sale. And bad, because you can’t expect high retention from such an audience. So you have to restructure your unit economics depending on your circumstances.

Usually, I see companies do two things.

Setting much higher prices to make the ad spend back.

And giving a discount for the first period to add urgency for emotional buyers.

It’s always a trade-off between how much you can make and how quickly. With a cold, emotional audience, speed is the deciding factor.

Offer higher prices with refund guarantees

In case of Emery, a bootstrapped project, I offer a lifetime option.

The trade is simple – as I’m scaling my ads, a quick return of spend is more important than long-term revenue. It lets me recycle money as quickly as possible, without investments or loans. And on days when someone purchases lifetime, I make the ad spend back on day 1. Results? $14k made in July while spending only $8k on ads. Not counting all the subscription renewals, just day 1 revenue.

And a web funnel is the only one that allows this. Selling a $149 option to a cold audience requires a money-back guarantee, which is not really possible if payments are controlled by the stores.

Setup CAPI and deduplication

Tracking ad performance is getting more difficult every year. Pixels are getting blocked by ad blockers and browsers. Various region-specific limitations are getting applied too.

To ensure high data quality, you need CAPI in place, where you duplicate all your key funnel events by sending them server to server. But you should be very careful and ensure proper deduplication. You don’t want to count double conversions every time someone makes a purchase. It varies from ad platform to ad platform, but they all support deduplication IDs (shout-out to Google, which managed to name them differently in different places, very helpful).

And one last thing: critical server events shouldn’t be a proxy for your client events. They should really happen on the server side – e.g. a Stripe webhook on a successful purchase. I’ve seen some bots that trigger client-side events manually (especially if Audience Network/Display Network is enabled in your ad account). You want to send data as clean as possible.

The funnel must be placed on the main domain

That’s my main problem with a lot of third-party funnel builders – they all offer custom subdomains, but I’m yet to see one that lets you create funnels under your main domain name.

Why is it important?

Whenever you add the pixel to your website, Meta tracks all visitors. And it reasonably assumes that people who already visited your website are much warmer. So, whenever you create a new campaign, it tries to go for that warm audience first and brings cheap conversions.

If your funnel is placed on a separate domain, you lose this signal. Having the same pixel on two domains doesn’t solve it. Meta structures data at the domain level.

Even if you’re a small business and Meta is your first acquisition channel, it matters. Try running some ads and you’ll see people searching for your company on Google and ending up on your website.

In my case, I place everything under one domain name – the website, the app itself and my funnel.

Set up audiences right away

One thing you want to avoid from the very beginning is to keep showing ads to people who already purchased. The main spender on Meta is ecommerce. For them, it makes total sense showing different ads to the same people again and again. But for a subscription-based business, it’s a waste.

So, make sure to create a custom audience of purchasers. Even based on Meta purchase events, which can be done in a few clicks in Ads Manager. And use it as an exclusion list across all campaigns.

Meta funnel must be optimised for small screens

Most of the time, Meta opens your website inside an in-app browser. It comes with an extra top bar and doesn’t render fullscreen. As a result, you get much less vertical space. So, all content and design should be optimised to fit into a window smaller than the regular device screen. Take your mobile design and shrink the window by 200 pixels – that’s how much space you have.

Sounds small, but I keep seeing companies squeezing in too much content that doesn’t fit vertically.

The first screen should open instantly

Again, we’re talking about a cold audience. They are not really motivated to wait for your pages to load. The quicker you engage the user, the more chances you have.

Difficult to achieve though.

Making a funnel that loads fast and has instant transitions between screens is not technically possible. You either load screens one by one and make interactions slow. Or you load all screens right away and sacrifice the initial loading speed.

In my case, I’m trying to have the best of both worlds. The very first screen is a static HTML page that loads instantly. And while the user reads the initial content and answers the first question, I load the rest of the funnel in the background. If they click too fast, they’ll see a short loader, but the rest of the journey will be instant.

Make it look nice

That was a surprise for me.

Obsessed with copy, I thought of it as the main lever. Until I added a few simple animations and it turned out to be the most successful experiment.

Which makes sense if you think it through.

People clicking on ads don’t know anything about your product. Their funnel journey is the introduction. They form an opinion on the product by how your funnel looks, how nice it is, how smooth it is.

Copy is still important, but don’t disregard the visual part.

Important note here. Don’t add animations for the sake of it. Do it strategically – e.g. draw user attention to a specific element, presenting information piece by piece. You are designing a thing that should guide user attention and how they consume your content.

Checklist

These are the technical basics. At some point, I’ll finally go through my notes and publish a piece on the creative side of high-converting funnels too. In terms of content, sales mechanics and so on. But for now, I want you to have a quick checklist covering the basics.

So, here you go:

• Does your first screen open instantly? Or does it show a loader instead?

• Does it look nice? Do you have animations? Do they actually serve some purpose?

• Does your content fit into 650 pixels vertically?

• Do you have pixel + CAPI set up? Is deduplication set up correctly (you can check in Events Manager)?

• Do you have a custom audience of purchasers? Is it used as an exclusion across all campaigns?

• Is your funnel placed on your main domain?

• Do you use trials or direct purchases on Meta web funnels?

• Do you offer a very expensive plan for the 10-15% of people who always buy the best?

It turned out Google requires a substantial investment before bringing any decent results. On an empty account with no conversion history, it brings completely random people. So, you have to spend around $10-15k without any guarantee of seeing results in the end.

Out of desperation, I decided to give Meta Ads a shot. And it started to bring some conversions right away. Same funnel, same offering. The price per conversion was too high. The performance was unstable. But it was at least something. The most challenging thing was to overcome my personal biases towards Meta. All my experience at Speechify and Bumble formed a strong belief that you need a lot of video creatives – UGC, animations, product shots. Something that a small side project can’t afford. No time, no money. But with some traction already, I had to enter that rabbit hole anyway.

What about AI?

It’s 2026 and AI is everywhere, so that could be the most natural place to start. I even took a 3-month bootcamp on AI creatives. Surprisingly, it convinced me to stay away from AI creatives. At least for now.

There are several problems that I quickly discovered.

AI is quite bad when it comes to producing statics. It’s awful at rendering fonts and following precise instructions – like making the text smaller or fitting content within a certain frame.

And when it comes to video, producing high-quality creatives still takes too much time. Models are limited in the duration they can generate. So you end up building a huge pipeline that writes the script, splits it into scenes, generates videos, stitches it all together, adds voiceover and captions. And at each level, slop creeps in unless you check and fix every step manually. So much for automation.

However, my biggest problem was ideological. If everyone produces AI slop, I simply won’t stand out. Unless I do it extremely well, which brings us back to the previous problem – time I don’t have.

So, back to manual production.

Subscribe now

Static ads to the rescue

Without time for video, static ads were the only way to go.

High-quality statics take time too, but I wasn’t going to do that. Learning from my AI experience, I decided to focus on differentiation. If everyone does AI slop, I’ll do it manually, crafting each message by hand in Figma. If everyone does colourful ads, I’ll go greyscale (that’s something I picked up from “Very Good Copy” by Eddie Shleyner). If everyone does beautiful polished ads, I’ll do ugly ones.

Over time, I developed my own set of rules:

• The ads will be static only. I banned myself from even looking at video for now.

• They should be greyscale. Sometimes I can add a colour to highlight certain pieces. Or even experiment with the full palette. But 90% of ads are still shades of grey.

• The layout should be assembled by hand. I can still use AI to generate additional assets, but I don’t trust it with text or element positioning. It takes time, but unlocks ease of translation. Generally, I can squeeze an extra 50% from successful creatives when translating them into German, French and Spanish.

• The creative is 1080x1920, but all content must be visible within the centre square. As a result, I can easily crop it inside Ads Manager instead of producing two sizes.

3 questions every ad must answer

When producing statics, content becomes the central part. The headlines, the visuals, the text – everything must be on point. I don’t have a lot of opportunities to grab someone’s attention while they’re scrolling their feed.

In the end, my checklist boiled down to 3 key questions. Whenever I create an ad that doesn’t answer one of them, I see poor results. I still break these rules from time to time, just to see if they still stand. And generally they do. I’m getting clicks with those ads, but no conversions. At least not at a reasonable price.

So, here it is:

What is being advertised?

Users think in categories. This one is a fitness product. That one is a beauty product. To help them out, I want to make it clear right away. In my case, it’s a planner. But not just any planner. It’s not a paper planner, not a kitchen renovation planner (a big thing in Germany, apparently). It’s a productivity app.

There are different ways of answering this question. Sometimes, I add store icons. Sometimes, I include keywords. But nothing beats app screenshots.

How is it different?

Once the category is clear, users should understand how this product differs from the rest of it. Not how it’s better, just different. And they can decide if it’s something they’re interested in.

Usually, it’s one of the angles of our positioning. It’s a guilt-free planner. Starts with a fresh page every morning. No overdues and no chance to mess it up. A planner that doesn’t bully you into productivity. That kind of thing.

Why should the prospect care?

In other words, what’s in it for them. Value over features. How their life will be different with my product, how they can make use of it.

Normally, it’s a natural progression from the previous question. Or sometimes even the same sentence.

Top-performing ads so far

Now, to illustrate, here are my 3 best performers.

I simply went to Ads Manager and checked the last 3 months of history. My LTV sits at $94. Acquiring paying customers at $30 allows me to break even in 3-4 months on subscriptions, and on day 0 whenever someone purchases a lifetime option. Anyway, the purpose of these numbers is to show that the maths works.

My main goal, however, is to show you don’t need an amazing ad to sell on Meta.

So, here you go:

As you can see, all of them are rudimentary. A catchy title, simple screenshots and a lot of space in between.

The only problem with the statics-only approach: creatives die quickly. You can squeeze 30-50 purchases out of a single variation. 80 if you strike gold.

But once you find 3-4 winning templates, variations are truly endless.

If you found this useful, check out my previous article where I reverse-engineer Meta ads to find the right campaign structure.

In the last 6 months, I spent a fair amount of time going through different sources. Reading all these subreddits. Watching dozens of YouTube experts, each claiming to have some secret knowledge. Attending calls with Speechify’s Meta reps and a dedicated engineer assigned to the company (not very useful, I have to say).

While I can’t point to a specific source, all these efforts started to pay off recently. One of the campaign structures worked. Then worked again. And again. The compounding effect kicked in. But as an engineer, I don’t like mysteries. So I went through the technical details to find out why this particular method worked over the others. And that was probably the most important part of the journey. Understanding the internals of Meta’s engine won’t help me win every time. But at least I’ll stop making silly mistakes.

In this post, I want to share my understanding of the Meta engine. It’s a simplification. It’s speculation. But I don’t think I’m too far off.

It might get a little technical at times, but once you grasp these few concepts, many things that looked like mysteries before will finally start to make sense. Meta will still find a way to surprise you. It’s gambling, after all. But we can stack the odds in our favour.

3 key components of the Meta engine

Whenever I open my Instagram, Meta needs to populate my feed with ads. And do it fast – under 200ms. To hit the time budget, it goes through 3 key steps:

1. First, it asks Publisher to return all the ads available for this user. Publisher uses hard checks when selecting ads for a specific user session. E.g. gender, location, age. It simplifies the next stages. But it still results in hundreds of thousands of creatives, if not more.

2. Ads received from Publisher go into Retrieval. The job of Retrieval is to shortlist the ads to a manageable number. It filters ads by matching them with the user’s interests and profile.

3. The shortlisted set is passed to Ranking. Potentially, everything could go there straight from Publisher, but Ranking is the most expensive operation in this chain, and the role of Retrieval is to lift some of that burden. Ranking chooses the ads with the highest potential of hitting the advertiser’s target for this specific user. It operates on countless parameters and, like any model, acts as a black box. Even its developers can’t tell you why a certain ad was ranked higher.

The important part that is often overlooked – this whole process is based on the user’s profile and runs entirely from a user-centric perspective. The ads are not competing in a vacuum. Every time, it’s a very different game for a specific user.

Now, where is the “learning” part in all of that?

Subscribe now

What Meta calls “learning”

I have a personal beef with the term “learning” in ad platforms. I appreciate the genius behind the naming. Whenever I create a new ad set, it starts with “learning”. It naturally nudges me to give the ad some time. It even creates an impression that the system is training for my product. In the world of marketing, full of magical thinking, I want to help the Meta God learn more from my ads and about my product. So it’ll be gracious enough to send me the right people. Meta even gives me an exact number – I have to feed it 50 conversions a week to keep the learning God happy.

However, as I mentioned before, the entire engine operates under a very strict time budget. Some sources say 200ms. With this constraint, there is no way to train anything specifically for every campaign. So the entire “learning” can be summed up as data collection. It does inform ad selection, but calling it “learning” is a bit of an overstatement. Moreover, Meta doesn’t switch to a full-beast mode when reaching 50 conversions. It just keeps collecting the data. If I had to bet, at some point Meta PMs ran some research and found that people who collected data from 50 conversions see better results. So they came up with this number and updated their dashboard to reinforce the guidance.

Now, and this is important, the data is collected on two levels: Ad Set and Ad Creative.

Ad Creative collects data on the performance of the specific ad. It also uses internal IDs based on ad content. So two identical ads placed in different ad sets can combine their data.

And Ad Set collects data on budget allocation, targeting and ad combinations. However, it’s heavily influenced by the creatives inside. After all, the entire history of ad impressions is based on the creatives.

The data collected in those two buckets is used in two places – Retrieval and Ranking.

Now, let’s see how, equipped with this knowledge, we can solve some common mysteries.

Solving mysteries

Meta sends all the budget to one creative

Very common for new campaigns. You upload 10 creatives into an ad set and Meta spends 80% of the budget on a single one. Great when you’re lucky and it’s the most profitable one. But often, it’s not the case. Usually, you see one or two creatives with much better ROI that barely get any spend. Feels broken.

But if we apply the concepts described earlier, it makes total sense. The most common reason – all the ads are too similar to each other:

1. Meta retrieves all my ads from Publisher

2. They all go through the same retrieval process and end up in Ranking, as they all match the same user profile

3. And then Ranking selects the one it predicts has the highest probability of delivering results.

Eventually, it’ll start giving other ads a shot, but the data pipeline is slow. If you have a single creative with a high ranking, it can cannibalise the ad spend for days.

Now, what’s the solution? Increase the variety within the same ad set. More formats, more angles, more messages. Don’t overestimate formats, though. The key here is to create ads that match different Instagram feeds. Using UGC instead of static is just one way to achieve it, and not the best one. This multiplies your chances of finding a banger that will produce revenue for a long time.

I disable the top-spending creative with bad ROI, and the entire ad set stops performing

Another popular one. This usually happens as an instinctive reaction to the previous issue. I don’t want to spend money on something that clearly doesn’t perform, so I stop it. But it almost never works the way it’s intended.

For two reasons:

Firstly, the data collected in the ad set was heavily influenced by that creative – it was the top consumer, after all. So the entire ad set is already way off, and it takes a lot of time to recover.

Secondly, the disabled ad was winning the ranking. So the rest of the ads are most likely already worse than the disabled one.

One way of fixing it – cloning the ad set without that ad and starting data collection from scratch. But it’s worth bearing in mind: it’s going to be an ad set of ranking losers. There is also a way to push spend to a specific creative, but the ad set data is usually biased at this stage.

The right way, in my experience, is to give up. If I treat Meta as a casino, I won’t win with a team of losers. I’d better abandon this ad set and create a new one. And never touch ads in the ones running.

A creative that tested well flops in the scaling campaign

I have a problem with the entire creative testing approach. It all sounds 2 years out of date – from before Andromeda (Meta’s current retrieval engine). It ignores the simple truth: data is collected in two places – ad level and ad set level. And they both influence each other. So a creative performing well in one ad set can be a complete waste of budget in another. And there is no way of predicting it. One is useless without the other.

Don’t get me wrong, I believe in testing to learn – which angles work, which messages resonate and drive the right audience. As a way for me to learn, not to game the Meta engine.

Learning, but not copying. With one small exception – usually I’m able to squeeze a little more juice by translating a successful creative into other languages. But that’s about it.

And if I get lucky and find something that works wonders here and now, I scale it in place, without copying it into a “scaling” campaign.

Meta doesn’t give new creatives a shot

This usually happens when I mix old and new creatives in a single ad set. Even if the new ones are more promising, the old ones come with the baggage of history. All things being equal, Meta gives preference to something that has already collected some data. And don’t forget, Meta uses internal IDs to match identical creatives by content. So even if you recreate the same creative from scratch, it’s still going to be treated as the same ad.

So, another rule – never combine new and old creatives in a single ad set. It won’t be a fair fight.

Understanding these few rules doesn’t make my campaigns bulletproof. After all, it’s still a game of chance – if the ad set picks up the right signal, if the creative resonates. There are millions of factors no one can predict. However, with these few concepts, the game becomes a little more predictable, and I start winning a little more often. I can’t say I’m becoming an expert in Meta Ads. It’s burned me too many times to get comfortable any time soon. But at least I get fewer and fewer surprises.

Lately, with Emery, I faced exactly the same problem. $5k spend, barely any conversions. The exact same funnel that performs well on Meta. The exact same angles. And search traffic is supposed to be high intent. The general consensus says you have to spend at least $10k to get anything approaching quality traffic. And that's if you already know what you're doing – campaign structure, keyword research, cadence. Fine for established businesses, but a deal breaker for a small bootstrapped project.

After some research, it looks like I finally found a way.

Since Google requires real training data and time, the question is how to feed it as cheaply as possible. Solution? Launch a campaign with branded keywords only.

In general, running branded campaigns is controversial. You pay for traffic you’d get anyway. So, overall, I’d be cautious with this approach. But as a warm-up mechanism, it does work, and here’s why:

• People searching for your product are the most likely to convert – they either saw your ad somewhere else and are doing research, or someone recommended you.

• Branded keywords are much cheaper than competitive category keywords. And even if competitors bid on your branded keywords, the auction is significantly skewed in your favour.

The only downside – many of your existing customers will be searching for the brand too, and you’ll be paying for them as well. However, it’s partially solvable with exclusion lists.

Now, here’s my checklist for pre-warming search ad accounts:

1. Ensure the tracking is in place – conversions, mid-funnel events, page views firing across all pages. I’ll write another article on this soon, it’s a big topic.

2. Go to Search Console and find all branded keywords people already use. In some cases, Google already offers a branded/non-branded query filter. If you don’t have it yet, use a regular filter and enter your brand name manually.

3. Create a search campaign targeting branded keywords. A few rules here:

   1. Location depends on your traffic source. In my case, I’m running Meta ads globally, so I go with worldwide targeting.

   2. Exact match only, despite all the warnings Google shows.

   3. Display Network and Search Partners are off. Otherwise, you’ll get a lot of spam traffic, especially on a new account. Besides, they just don’t make sense for a branded campaign.

   4. Fill in all headlines and descriptions – come up with as many variants as you can.

   5. Don’t set a target CPA yet.

   6. Set the end-of-funnel event as the conversion. Either trial or purchase. Everything else is noise.

   7. Set the budget to get at least 30 conversions a month at your CPA. If you’re just starting out and don’t have enough search volume, you can start with as little as $10 a day.

4. Set up an exclusion list (Customer Match) to minimise impressions for existing customers. It can be as simple as all your sign-ups.

Once the campaign is live, prepare to be patient and review performance regularly. Google Ads really doesn’t like frequent changes, so we need to minimise two things when scaling the campaign – the size of the changes and how often we make them. My rule of thumb: just a single change a week, be it budget, CPA or anything else.

Here’s what the timeline can look like:

• Week 1. Verifying the setup. Checking the search terms to ensure they’re all branded, checking conversion events.

• Week 2. Getting the first conversions.

• Week 3. After getting a few conversions, I set my initial CPA target. You can use your regular CPA + 20%, or just take the actual value from the previous two weeks.

• Weeks 4–8. Start gradually decreasing the CPA. No more than 10% once a week. If we’re to pay for branded traffic, we should pay as little as possible. In general, I’m ready to pay 60–70% of what I pay Meta for cold traffic. So if my regular CPA is $30, I want to gradually get down to $20 while still getting conversions.

After a couple of months of this exercise, you should be able to feed Google enough conversions and start your first non-branded campaign.