iGaming funnels that convert: a teardown with math

Two affiliates promote the same casino, in the same geo, from the same traffic source. One turns 2% of clicks into registrations. The other turns 8%. Neither has a secret offer — iGaming funnels are where that difference is created, and it is almost always visible in the first three screens.

The useful way to think about this is that a funnel is not persuasion. It is subtraction. Between the click and the first deposit there is a chain of small refusals — a slow page, an unfamiliar payment logo, a form asking for a postcode, a bonus term that appears for the first time at the cashier — and each one removes a fixed slice of the remaining audience. Your job is to find the slices and cut them.

What follows is the full chain with arithmetic: what each step is worth, what an extra screen actually costs, how proof elements have to change by market, why mobile constraints decide more than design taste does, and how to diagnose which specific screen is leaking rather than guessing.

iGaming funnels: the click-to-deposit chain

Start by writing the whole chain down. Most affiliates track two numbers — clicks and deposits — and then wonder where the money went. There are at least five steps in between, and every one of them multiplies.

Here is a representative paid campaign. Assume 10,000 clicks at a $0.60 CPC, so $6,000 of media.

Click-to-registration is 9.3%. Registration-to-deposit is 22%. Cost per first-time deposit is $6,000 / 205 = $29.27.

Now the part that changes how you prioritise. Because the steps multiply, the value of fixing a step is proportional to the relative improvement available there, not the absolute percentage points.

Ten percentage points on the pre-lander is worth more than everything else on the list, because 38% is a low base and ten points is a quarter of it again. Five points on a 58% step is almost nothing. This is the single most common prioritisation error in funnel work: people attack the step with the worst-looking number instead of the step with the most headroom relative to its base.

The three-screen rule and what each screen owes you

A user should reach the registration form within three screens of the click. Not because three is magic, but because of what each additional screen costs, which we will price in a moment. Within those three, each screen has exactly one job.

Screen one: earn the second click

The pre-lander's job is not to sell the casino. It is to convert an ad-click impulse into deliberate forward motion, and to shed the users who were never going to deposit before you pay to render a heavy page for them.

One idea, one action. The strongest formats we ship are a single question with two answers, a short qualification step that feels like a choice rather than a form, or a concrete claim tied directly to the ad creative. The absolute requirement is continuity: whatever the ad promised must be the first thing on the screen, in the same words. Message mismatch between creative and pre-lander is the most reliable way to convert a 45% pre-lander into a 20% one.

What does not belong here: game libraries, provider logos, terms, licence text, anything requiring scrolling to understand. All of that is screen two's problem.

Screen two: make the bonus concrete

This screen answers "what exactly do I get, and what is the catch". Vague generosity performs worse than specific modesty. "100% up to 500 plus 50 free spins on a named game, 35x wagering, minimum deposit 20" outconverts "huge welcome bonus" in every market we have tested it in, because the reader has already assumed there is a catch and you are either the site that told them or the site that hid it.

This is also where proof lives: the licence, the payment methods they recognise, withdrawal timing, and any local trust signal. Proof is geo-specific and is covered below.

Screen three: get out of the way

The registration form introduces no new information. Every field is a refusal opportunity. Ask only what the operator requires to create an account, defer everything else to KYC, keep the bonus visible while the user types so the reason for the effort stays on screen, and make sure the keyboard type is right for each field — a numeric keypad for phone numbers, correct country prefix pre-filled by IP.

If the operator's own registration is bad and you cannot change it, that constraint should influence which operators you promote. A great funnel into a hostile signup form is a wasted funnel.

What one extra step actually costs

Affiliates add screens for reasonable-sounding reasons: an interstitial to warm the user, a second offer page, an age gate that could have been a checkbox. Price it.

Insert one additional screen at a generous 75% pass-through into the baseline funnel above. FTDs fall from 205 to 154. Cost per first-time deposit rises from $29.27 to $38.96 — a 33% increase in acquisition cost, from a screen that on its own looked like it retained three quarters of its traffic.

A screen that passes 75% is not a good screen. In a multiplicative chain it is a 25% tax on everything downstream of it. This is the entire justification for the three-screen rule, and it is why "let us add a quiz to increase engagement" needs to clear a much higher bar than it usually gets.

The mirror image is also true: removing a redundant step is often the highest-return change available, and it costs nothing to test.

Localization is not translation

Running the same page through a translation layer produces a page that is technically readable and emotionally foreign. The elements that create trust are not sentences. They are logos, currency, numbers, and tone.

Proof elements by geo

These are patterns we see repeatedly, not universal laws — verify against your own data before betting a budget.

• Brazil. The instant-payment logo above the fold does more work than any headline. Portuguese from the domain and URL slug onward, not just body copy. Amounts in reais with a realistic minimum deposit. Withdrawal speed stated plainly.
• India. UPI availability is the make-or-break signal, followed by rupee display and a genuinely low entry deposit. Cricket context outperforms generic casino imagery for sportsbook-adjacent offers.
• Japan. Restraint converts. Aggressive urgency and exclamation-heavy copy actively suppress conversion. Correct script usage, yen amounts, bank transfer and convenience-store payment options, licence stated soberly.
• Canada and the Nordics. Regulator prominence and responsible-gambling messaging read as competence rather than as friction. Withdrawal timelines are the most-scrutinised claim on the page.
• Nigeria and Kenya. Mobile money support is the first question. Page weight is a conversion factor, not a technical nicety — a heavy hero image is a lost user.
• Germany and other tightly regulated EU markets. Deposit limits and permitted product scope constrain what you can advertise at all. The compliant page is often the only page.

Two things travel badly across all markets: humour and social proof invented by you. Testimonials that a local reader can tell are fabricated cost more than they earn, and in several regulated markets they are also a compliance problem.

The practical implication is that one template plus four translations is not a localization strategy. Four pages that share a structure and differ in proof, payment and tone is.

Mobile-first is not a design preference

Most iGaming traffic in most geos is mid-tier Android on a variable connection, arriving through an in-app browser. That single sentence dictates more funnel decisions than any design principle.

Weight. Every unnecessary kilobyte shows up as a lower render rate, which then multiplies through the whole chain. Compress and modernise image formats, avoid render-blocking web fonts, inline critical CSS, and be ruthless about third-party scripts. Cutting page weight is the cheapest render-rate improvement available and it needs no creative approval.

In-app webviews. Traffic from social apps often opens in an embedded browser with restricted storage and inconsistent redirect behaviour. Client-side attribution can silently fail here, which is why server-to-server tracking is not optional in this vertical — the mechanics are covered in the S2S postback guide.

Thumb reach and sticky CTAs. The primary action must be reachable without a grip change. A persistent bottom CTA usually beats a hero button, especially on longer proof screens.

Form mechanics. Autofill, one-time-password auto-read, sensible password rules, and a country prefix that is already correct. Registration completion moves several points on these alone, and none of them are visible in a desktop preview.

App handoffs. If the operator pushes users into a native app, test whether attribution survives the install. Assume it does not until you have proven otherwise with your own device.

Diagnosing which screen leaks

You cannot fix a funnel you can only see the ends of. Instrument every step, break every step by source, creative and lander, and read the pattern. Sub-ID depth is what makes this possible at all — the sub-ID tracking approach is the difference between "the campaign is down" and "creative 4 on placement B stopped rendering in Brazil".

The last row deserves emphasis because it is the most expensive and the most misdiagnosed. When deposits lag, affiliates rewrite copy. Copy is rarely the problem. Deposit-step failures are almost always about money mechanics: the user reached the cashier, did not see the method they use, and left. No headline fixes that.

One more diagnostic worth running: compare deposit rate by traffic source at identical registration volume. If one source registers well and deposits badly, you are buying curiosity rather than intent, and the fix is upstream in targeting, not in the funnel.

Testing discipline for iGaming funnels

Most funnel testing in this vertical is theatre, because the volume required to read a result honestly is larger than people assume.

A workable rough rule for sample size per variant is n ≈ 16 × p × (1 − p) / δ², where p is the baseline rate and δ is the absolute lift you want to detect.

Testing the pre-lander at a 38% baseline for a 10% relative lift (δ = 0.038): n ≈ 16 × 0.38 × 0.62 / 0.038² ≈ 2,610 sessions per arm. At an 88% render rate that is about 2,970 clicks per arm, or roughly 6,000 clicks total. Affordable.

Now test the deposit step at a 22% baseline for the same 10% relative lift (δ = 0.022): n ≈ 16 × 0.22 × 0.78 / 0.022² ≈ 5,670 registrations per arm. At 9.3% click-to-registration that is over 60,000 clicks per arm — around $73,000 of media to read one test.

The conclusion is not "do not test". It is that upper-funnel steps are testable and the deposit step generally is not, at normal affiliate volumes. So:

1. Test where the volume is, one variable at a time.
2. Always read the downstream deposit number even though you cannot test it. A variant that lifts pre-lander clicks by 25% and drops deposit rate by 30% is a loss, and you will only notice by looking.
3. Run tests in whole weeks. Day-of-week effects in gambling traffic are large enough to invent and destroy winners.
4. Do not stop early because a variant is ahead. Early leads on small samples reverse constantly.
5. Keep a changelog with dates. Half the "the funnel broke" incidents resolve to a change someone forgot they made.

The economics that decide how good your funnel needs to be

How much funnel work is worth doing depends on your deal structure, which is why the two questions should be answered together.

On a flat CPA, funnel improvements are worth exactly the marginal deposits they create, once. On revenue share, the same improvement compounds — those extra depositors keep paying, and the cohort curve does the rest. That asymmetry means a revenue share deal justifies substantially more investment in the funnel, which is one of the more practical arguments in the revenue share versus CPA comparison and in choosing between iGaming deal types.

There is a second-order effect that catches people out. Funnel changes alter player quality, not just player quantity. A pre-lander that leans harder on bonus size will produce more registrations and a bonus-hunting mix — which is fine on CPA and can be actively harmful on revenue share, particularly under a negative carryover arrangement where a lucky cohort can put you in a hole. Judge funnel variants on the metric your deal actually pays on.

Where funnels fail after they are built

They rot. Payment methods change, operators redesign their registration, a licence badge becomes outdated. Re-walk your own funnel monthly from a real device in the target geo.

They get copied badly. A lander that works in Brazil gets duplicated for Mexico with a currency swap and nothing else. It underperforms, and the conclusion drawn is "Mexico is weak" rather than "the page is Brazilian".

They stop being measured. Tracking breaks quietly. If a step's number is suspiciously stable while everything around it moves, suspect the instrumentation before the market.

They win on the wrong metric. Optimised to registrations because that is what the dashboard shows, then quietly delivering worse depositors every month.

They outrun their traffic. A perfect funnel on unqualified traffic still converts badly. When every screen is clean and the numbers stay poor, the problem has moved upstream — targeting, source, or intent. That is where SEO for casino affiliates and source selection start to matter more than any further page work.

Working with Shazam on iGaming funnels

We build geo-specific landing pages and full funnels for partners at no cost, because a generic template applied across four markets is the most common reason good traffic underperforms. Tracking runs server-to-server and down to sub-ID depth, so partners see click-to-registration and registration-to-deposit rates per source, per creative and per lander — which means when a funnel leaks we can point at the screen instead of debating the offer. Direct relationships with operators also let us fix things on the other side of the handoff, like registration friction and payment coverage in a specific geo, rather than routing around them. If your gambling traffic is converting below the ranges above, the funnel is the first place to look and rebuilding it is part of the deal. There is more on how the partnership works on the main site.

Send the Telegram bot your geo, your traffic source and your current click-to-registration rate, and we will tell you which screen we would rebuild first. Partners compare funnel numbers in the community chat.