Voodoo UA Playbook 2026: Paper.io 2 and Hole.io Are Not Being Advertised — They Are Being Harvested

Voodoo has four games on the US download chart, and the striking thing about all four is how little advertising is behind them. Across Paper.io 2, Hole.io, Castle Busters and Mob Control, 557 measured ad instances account for 129,381,048 impressions built from just 263 distinct video assets.

To put that in scale: measured over the same 30-day window, those four titles together carry 129,381,048 impressions and generate 1,904,459 monthly US downloads. A single competing puzzle title measured the same day — Oakever's Meowdoku! — carries 1,786,574,888 impressions for 2,158,898 downloads. Voodoo takes 88% of the downloads on 7.2% of the impressions.

And the two classic io titles are the emptiest of all. Only 9.8% of Paper.io 2's window impressions come from creative first seen in the last 90 days. Its single largest ad instance launched on 2025-12-31. Voodoo is not defending these games with creative. It is collecting what they still produce.

#Executive summary (TL;DR)

• Four titles, 129 million impressions. Paper.io 2 carries 80,205,165, Castle Busters 18,658,566, Hole.io 18,194,182, Mob Control 12,323,135. The entire measured US portfolio is smaller than a single week of output from high-intensity puzzle publishers.
• The io classics run on old creative. Paper.io 2's oldest live instance was first seen 2024-10-28 and has been running 661 days. Average instance age across the fleet is 182.2 days. Only 30.3% of instances are less than 90 days old, and they carry just 9.8% of window impressions.
• The creative team has moved. Castle Busters — Voodoo's newest chart title — accounts for 120 of the portfolio's 263 video assets despite carrying 14.4% of its impressions. 96.5% of its instances are under 90 days old, against an average lifespan of 21.5 days.
• Extremely thin delivery. Paper.io 2 runs at 43 impressions per placement and Castle Busters at 16. For comparison, Oakever's Amaze GO! runs at 2,291. Voodoo is buying enormous placement counts at very low density.
• Two publisher IDs, two spellings. Voodoo's iOS titles report under ID 262118821 as "Voodoo"; its Android bundles report under 1731657538 as "VOODOO". On Paper.io 2 the Android side carries more impressions than iOS (44,839,770 vs 35,365,395) from fewer instances.
• Two name traps in one portfolio. Castle Busters' ad creative is titled "Castle Clashers!", not Castle Busters. And Hole.io's highest-impression instance in the raw data is a Lingokids ad — a mis-attribution that would have added a bogus 2.19 million impressions and a "#1 LEARNING APP FOR KIDS" hook to a black-hole game. We excluded it; details in the methodology section.
• Mild declines where spend is newest. Mob Control is down just -4.37% and Castle Busters -11.41%, against Paper.io 2 at -27.74% and Hole.io at -30.03%. The two titles receiving fresh creative are the two holding up.

All figures are AdMapix third-party estimates for the United States, pulled 2026-08-19.

#The publisher at a glance

Read the last three rows together and the portfolio splits cleanly in two. Paper.io 2 and Hole.io are old fleets with long-lived assets and high live rates. Castle Busters is a young fleet with a 21.5-day average lifespan and three-quarters of its instances already stopped. Those are not the same activity.

Paper.io 2's single largest measured ad instance: 12,450,275 impressions across 238,412 placements, 30 seconds, first seen 2025-12-31 and still running after 232 days.

#The scale problem: what "not advertising" looks like in numbers

Publisher-level impression totals are hard to interpret in isolation. Here is Voodoo's entire measured US portfolio next to the two titles from another publisher we pulled on the same day, in the same country, with the same method.

Every figure below is the same 30-day window — 2026-07-20 to 2026-08-19 — for every title, so no normalisation is needed or applied.

Voodoo's entire four-title US portfolio carries 7.2% of the window impressions that one Oakever puzzle game carries, and delivers 88% of that game's monthly downloads.

Per thousand impressions, Voodoo converts at 14.72 downloads against Meowdoku!'s 1.21 — 12.2x the yield. Its weakest title on this measure, Paper.io 2 at 7.14, still converts nearly six times better than the best-converting Oakever title.

Two readings are possible and both are probably partly true.

Reading one: brand equity. Paper.io 2 and Hole.io are seven- and eight-year-old franchises with enormous historical install bases and permanent presence in the "io games" search space. A large share of their 1.07 million combined monthly downloads is organic — store search, category browse, word of mouth, cross-promotion inside Voodoo's own network — and requires no media at all.

Reading two: hypercasual economics. Voodoo built its business on titles that monetise through ad inventory rather than in-app purchase. In that model the ceiling on user acquisition spend is very low, because the lifetime value of an installed user is measured in cents. You cannot buy your way to scale on a 30-cent LTV; you can only ship a lot of games and let the winners run cheaply.

The data cannot distinguish between these. What it can show is where Voodoo's remaining creative effort is going, and that answer is unambiguous.

#Paper.io 2: 62% of the portfolio, 58 videos, and almost nothing new

Paper.io 2 is comfortably Voodoo's largest measured US ad presence, and it is running on a museum.

Every one of the top ten was first seen before 2026-03-19. Not one of Paper.io 2's ten largest assets was made in the last five months. Eight of the ten have been running for more than 150 days, two for more than a year, and all ten were still live on the pull date.

Notice the fourth column. Every single instance delivers exactly 52 impressions per placement. That uniformity across ten assets with wildly different ages, durations and formats is not a creative property — it is a property of the inventory. Whatever Voodoo is buying for this title, it is a single, extremely consistent, extremely thin delivery channel.

The cohort view makes the stagnation explicit:

Eight instances launched in December 2025 still carry 31.5% of everything this title has ever delivered. Twenty-two launched in August–September 2025 carry another 30.3%. Together, thirty assets from two windows nearly a year ago account for 61.8% of Paper.io 2's lifetime measured impressions.

Meanwhile the 21 instances launched across July and August 2026 carry 5.4%. Voodoo is still producing creative for this title — it just is not putting weight behind it.

The playable end card shared by Paper.io 2 instances #1, #2 and #3 — one asset, first seen 2025-09-06, still attached to creative delivering in August 2026.

Format-wise, Paper.io 2 is a two-length account: 51 of 99 instances at 60 seconds and 22 at 30 seconds, with a scatter of 26s, 29s and 59s cuts and a single 90-second outlier. Seventeen distinct playable assets serve 47 instances — again, a small shared component set rather than per-creative builds.

Paper.io 2 instance #5 — 60 seconds, 4,580,748 impressions, first seen 2025-08-19 and still recording placements after 366 days. The video behind it is shared with instance #6, which has been live for 380 days.

#Hole.io: the smallest fleet with the longest memory

Hole.io carries 18,194,182 impressions across 142 instances — 14.1% of the portfolio, on a title doing 499,514 monthly downloads. Its oldest instance dates to 2024-12-24, a 604-day run.

The interesting structural fact is that Hole.io looks more active than Paper.io 2 by instance count and less active by weight. 73.9% of its instances are under 90 days old, but they carry only 50.3% of impressions, and the title's largest genuine asset — after the exclusion described below — delivers just 1,207,814 impressions. That is 6% of what Paper.io 2's champion delivers, and less than one-hundredth of what a top Oakever creative does.

Instance #7 is the one worth flagging: a playable-first unit first seen 2025-04-05 that ran for 483 days before being switched off on 2026-07-31. That is the clearest genuine retirement decision anywhere in the Voodoo portfolio, and it took sixteen months to arrive.

Hole.io's duration profile is also the most fragmented of the four titles: 33 instances at 38s, 30 at 31s, 26 at 30s, 14 at 60s, and a tail running down to 17 seconds. There is no house format here in the way Paper.io 2 has one.

Hole.io instance #2 — 31 seconds, 1,138,938 impressions across 26,952 placements, first seen 2025-12-29 and still live after 234 days. Its underlying video is shared across eight instances carrying 2,732,528 combined impressions.

Hole.io instance #1 — 38 seconds, 1,207,814 impressions in 26 days, first seen 2026-07-25. The same video appears across four instances totalling 2,027,056 impressions, making it the title's largest video asset by a narrow margin.

#Castle Busters: where Voodoo's creative team actually lives

If you only looked at impressions, Castle Busters would be a footnote — 18,658,566, or 14.4% of the portfolio. If you look at production, it is the whole story.

Castle Busters has produced more distinct video assets than both io classics combined, on a title with a fifth of their impressions. Its July 2026 cohort alone is 140 instances carrying 53.0% of the title's impressions — more new creative in one month than Paper.io 2 has produced in its entire measured history.

And it churns. Average instance lifespan is 21.5 days. Nearly two-thirds of its instances had genuinely stopped before the pull date, excluding the observation-lag window. This is a title in active testing, being run the way a hypercasual publisher runs a title it is still trying to find product-market fit for.

The delivery pattern inside the fleet is also violently bimodal:

Instance #1 delivers 353 impressions per placement. Instances #2 and #3, launched a day apart in late July, deliver four and five. Those two assets between them account for 614,902 placements — more than half the title's entire placement count — and produce 3.16 million impressions. That is the signature of a very broad, very cheap reach buy sitting alongside a much denser one, and Voodoo is running both simultaneously on the same title.

Castle Busters instance #1 — 60 seconds, 7,134,466 impressions across just 20,176 placements, first seen 2026-05-22. At 353 impressions per placement it is 88x denser than the title's second-largest asset.

Castle Busters instance #2 — 29 seconds, 1,676,248 impressions across 351,912 placements. Four impressions per placement, first seen 2026-07-29. The broadest, thinnest delivery in the entire Voodoo portfolio.

Castle Busters is also the only Voodoo title in this set carrying a real ad-copy narrative. Ninety-eight instances run the line "Enter high-stakes 1v1 PvP battles in Castle…", and creative titles include "Build Smarter. Win More", "Erase the enemy" and "Bricks Don't Lie". Compare Paper.io 2, which runs two distinct copy strings across 99 instances, and Mob Control, which runs one.

#The naming traps in this portfolio

Two separate identity problems appear in Voodoo's data, and both would have corrupted this analysis if taken at face value.

Trap one: Castle Busters advertises as "Castle Clashers!" The App Store lists app ID 6746328263 as Castle Busters, and it appears on the US download chart under that name with 434,022 downloads. But the most common creative title across its ad fleet is "Castle Clashers!", and the dominant copy string references "Castle…" without ever completing the product name. Anyone searching creative databases for "Castle Busters" by name would find a fraction of this fleet.

Trap two: Hole.io's biggest ad is not a Hole.io ad. The single highest-impression instance returned for app ID 1389111413 carries 2,193,435 impressions, a 60-second duration, a playable end card, and the following ad copy:

"#1 LEARNING APP FOR KIDS — Spark success…"

Its creative title is "Lingokids - Play and Learn" and its playable asset filename contains the string m_en_lilo_. This is a children's education app that has been attributed to a black-hole physics game. It is a clear mis-attribution in the upstream data.

We excluded it. Every Hole.io figure in this article — the 18,194,182 total, the 142 instance count, the 1,207,814 top creative — is calculated with that instance removed. Including it would have inflated Hole.io's total by 12.1% and, worse, produced a confident paragraph about Voodoo testing an education-style hook on a casual game.

The general rule this illustrates: a single mis-attributed row at the top of an impression-sorted list does maximum damage, because that is exactly the row an analyst is most likely to feature. Always read the ad copy of your top result and check that it describes the product you think you are analysing. If the copy and the app disagree, trust the copy.

#Mob Control: the healthiest title, and the smallest fleet

Mob Control is down just -4.37% month over month — by far the mildest decline in the portfolio, against -27.74% and -30.03% for the io classics. It also has the smallest ad fleet: 90 instances, 31 video assets, 12,323,135 impressions.

But it is the one title where Voodoo's delivery looks conventional. At 223 impressions per placement it runs 5x denser than Paper.io 2 and 14x denser than Castle Busters. Its top three instances all launched on the same day — 2026-06-11 — and all three are still live 70 days later:

Three assets, launched together, at 30s / 40s / 39s, delivering within 17% of each other. That is a coordinated batch launch with near-identical outcomes — the cleanest example of a controlled multi-length test anywhere in this dataset, and it produced 5.2 million impressions, or 42.4% of the title's total.

Mob Control also has the highest playable attachment in the portfolio at 71.1%, served by just 12 distinct playable assets.

Mob Control instance #1 — 30 seconds, 1,944,044 impressions across 8,529 placements, first seen 2026-06-11. One of three assets launched the same day, all still live 70 days later.

A Mob Control playable end card, one of 12 serving 64 of the title's 90 instances — the highest playable attachment rate in Voodoo's US portfolio at 71.1%.

#Two publisher IDs, and an Android side that outweighs iOS

Voodoo appears in the creative data under two distinct publisher identifiers with two different spellings:

The split matters more than the spelling. On Paper.io 2, the Android side carries more impressions than iOS from fewer instances:

Android instances on this title average 2.1x the impressions of iOS instances. Hole.io runs the opposite way — 108 iOS instances carrying 16,290,769 impressions against 34 Android instances carrying 1,903,413, an 8.6x iOS skew.

Neither Castle Busters nor Mob Control returned any Android-side creative in this pull at all. That may be a monitoring coverage gap rather than a real absence, and it is the main reason we treat all portfolio totals here as lower bounds.

#What the format data says

Across the four titles, Voodoo's duration profile is the most conventional element of the whole portfolio.

Three of the four titles anchor on the near-minute format, and Hole.io is the only one that does not — it sits in the 30–38 second band. Castle Busters is the most format-locked of the four: 140 of 226 instances at 59 or 60 seconds, or 61.9%, with a small 15-second cluster underneath.

The playable numbers show the same component-library pattern that Oakever runs, at a much smaller scale. Castle Busters supports 62 playable-carrying instances with eight distinct playable assets — roughly one build serving eight placements' worth of creative. When you are testing 120 videos in eleven months, the interactive layer is the part you cannot afford to rebuild each time.

#How to reproduce this teardown

1. Group the portfolio by publisher ID, then verify by display name. Voodoo spans 262118821 ("Voodoo") and 1731657538 ("VOODOO"). A case-sensitive name match would have split it; an ID-only grouping would have too.
2. Never join creative data to ranking data by app name. Match on app ID or bundle ID. This portfolio contains one title advertising under a different name (Castle Busters as "Castle Clashers!") and one mis-attributed foreign ad at the top of an impression sort.
3. Read the copy on your top result before you write about it. The Hole.io / Lingokids mis-attribution is only detectable by reading the ad text. Impression counts, dates and formats all looked perfectly plausible.
4. Separate instance counts from asset counts. Voodoo's 557 instances resolve to 263 distinct videos and 61 distinct playables. Instances measure media weight; fingerprints measure production capacity. Castle Busters looks small on the first metric and dominant on the second.
5. Always compute a cohort view. Paper.io 2's top-ten list looks like a healthy fleet. Its cohort table shows that 61.8% of lifetime impressions come from two windows nearly a year old.
6. Compute impressions per placement. It is the single fastest way to tell what kind of inventory an advertiser is buying, and in this portfolio it ranges from 4 to 353 within a single title.

#What transfers to your UA program

If you manage a mature catalogue: the Paper.io 2 pattern is a legitimate strategy, not neglect. A title with strong organic demand and low per-user monetisation should not be defended with fresh creative every month. The correct behaviour is to keep a small set of proven assets running indefinitely, spend nothing on iteration, and move your creative team to the title that still has an unanswered question. Voodoo has done exactly that.

If you are deciding when to stop investing in a title: the signal to watch is not download decline — all four Voodoo titles are declining. It is the share of impressions coming from recent creative. Paper.io 2 is at 9.8% and Castle Busters at 93.7%. Those two numbers tell you where the organisation's attention actually is, regardless of what the roadmap says.

If you run reach-oriented campaigns: Castle Busters' bimodality is worth studying. One asset at 353 impressions per placement and two at 4–5, running simultaneously on the same title, is a deliberate barbell. Very cheap broad reach and much denser targeted delivery are not competing strategies here; they are two lines in the same plan.

If you benchmark creative volume: be careful comparing across monetisation models. Voodoo's 263 videos across four titles would look like underinvestment next to a puzzle publisher shipping 461 videos for one title in 91 days. But Voodoo's titles monetise through ad inventory rather than IAP, which caps acquisition spend structurally. Creative volume benchmarks are only meaningful within a monetisation model.

If you are competing with Voodoo: Paper.io 2 and Hole.io are undefended in media terms and defended by brand. You will not outbid them, because they are barely bidding — you would have to out-rank them in store search and category placement. Castle Busters is the opposite: actively contested, high churn, and still looking for its format.

#A falsifiable prediction

Prediction: by 2026-10-31, Paper.io 2 will still have no single ad instance first seen after 2026-08-19 carrying more than 2,000,000 impressions, and its share of impressions from creative under 90 days old will remain below 20%.

The basis: over the last four months Paper.io 2's newest cohorts have delivered progressively less — 9.0% from April–June, 5.4% from July–August — while the December 2025 and August–September 2025 cohorts continue to carry 61.8% of lifetime volume. There is no sign in the data of a planned relaunch, and the title's -27.74% download decline gives Voodoo no obvious reason to start one.

If instead a new Paper.io 2 asset breaks two million impressions in that window, it means Voodoo has decided the franchise is worth re-funding — which would be a genuine strategy change and would likely show up first as a jump in impressions per placement, not in instance count.

We will re-pull and report the outcome either way.

#Methodology and data notes

• Source: AdMapix creative search and ranking APIs, queried 2026-08-19, country filter US, sorted by impressions descending.
• Sample definition. Creative sets were retrieved by keyword and filtered to instances whose primary app identifier matched the target title. Paper.io 2 (99 instances), Hole.io (142 after exclusion) and Mob Control (90) returned complete retrievable sets; Castle Busters returned 226 instances within the retrieval depth used. Portfolio totals should be treated as lower bounds.
• One instance was excluded. The highest-impression row returned for Hole.io (app ID 1389111413), carrying 2,193,435 impressions, has the creative title "Lingokids - Play and Learn" and ad copy beginning "#1 LEARNING APP FOR KIDS". It is a mis-attribution in the upstream data and is removed from every Hole.io figure in this article. No other exclusions were made.
• Instances versus assets. The API returns individual measured ad instances, each with its own impressions, placement count and date range. Multiple instances frequently share one underlying video or playable. Voodoo's 557 instances resolve to 263 distinct video fingerprints and 61 distinct playable fingerprints. Impressions are reported per instance and do not double-count.
• Publisher attribution was verified on both the ranking and creative sides by app ID and publisher ID. Voodoo appears under 262118821 (iOS, "Voodoo") and 1731657538 (Android, "VOODOO"). These are AdMapix publisher identifiers, not App Store artist IDs.
• Stop dates carry a scan-lag caveat. The most common "last observed" date other than the pull date is the day immediately before it across every title, which reflects observation cadence rather than synchronised stops. Figures quoted as "genuinely stopped" exclude any instance last observed within two days of the pull date.
• ⚠️ Impressions are a 30-day window value, not a lifetime total. With no explicit date range the API returns a rolling 30-day window, verified here as 2026-07-20 to 2026-08-19. Every impression figure in this article is that window, for every title, which is what makes the cross-publisher comparison valid without normalisation. A different window returns different values, and window impressions must never be divided by a campaign age.
• Cross-publisher comparisons with Oakever Games use figures pulled on the same date with the same method and country filter. Both use the identical 30-day window, so the comparison is valid for relative volume; it is not a spend comparison, and the two publishers operate different monetisation models.
• Impressions, placement counts and download figures are third-party estimates, not publisher-reported and not spend.
• Store listings: Paper.io 2, Hole.io, Castle Busters and Mob Control on the App Store. Company information is on Voodoo's corporate site.
• Coverage is limited to creatives observable through third-party monitoring.

#Frequently Asked Questions

How many ads is Voodoo running in the US? Across its four US chart titles, Voodoo has 557 measurable ad instances built from 263 distinct video assets and 61 distinct playable assets, totalling 129,381,048 impressions as of 2026-08-19. Paper.io 2 accounts for 62.0% of those impressions on its own.

Is Voodoo still advertising Paper.io 2? Barely. Paper.io 2 has 99 measured ad instances, 80.8% of which were still live on 2026-08-19 — but only 30.3% were first seen in the last 90 days, and those carry just 9.8% of the title's lifetime impressions. Its largest ad launched 2025-12-31 and its oldest has been running since 2024-10-28. Voodoo is keeping proven assets in market rather than producing new ones.

What is Voodoo's best-performing ad creative? A 30-second Paper.io 2 video with a playable end card, first seen 2025-12-31, carrying 12,450,275 impressions across 238,412 placements over 232 days. It is 5.8x larger than the next-biggest asset in the portfolio, Castle Busters' 60-second lead creative at 7,134,466 impressions.

Why do Paper.io 2 and Hole.io have so few impressions compared to other chart games? Voodoo four titles carry a combined 129,381,048 impressions in the 30-day window, against 1,786,574,888 for a single competing puzzle title measured on the same date — 7.2% of the volume — while producing 88% of that title monthly downloads. Two factors explain most of it: these are long-established franchises with substantial organic demand, and hypercasual titles that monetise through ad inventory rather than in-app purchase cannot support high acquisition spend.

Which Voodoo title is getting new creative? Castle Busters. It accounts for 120 of the portfolio's 263 distinct video assets despite carrying only 14.4% of its impressions, 96.5% of its instances are under 90 days old, and its July 2026 cohort alone contains 140 new instances. Its average instance lifespan is 21.5 days against 182.2 for Paper.io 2.

Does Voodoo use playable ads? Yes, on all four titles, but as a small shared component set. Mob Control has the highest attachment at 71.1% served by 12 distinct playables; Castle Busters supports 62 playable-carrying instances with just 8 distinct playable assets. Across the portfolio, 61 playable builds serve 557 ad instances.

Is Castle Busters the same game as Castle Clashers? The App Store lists app ID 6746328263 as Castle Busters, and it appears on the US download chart under that name with 434,022 monthly downloads. However, the most common creative title across its ad fleet is "Castle Clashers!". Searching creative databases by the store name alone would return only part of the fleet.

---

Want to know whether a competitor is defending a title or quietly harvesting it? Cohort-level first-seen dates, per-instance impressions, placement density, video and playable fingerprints, and ad-copy strings are all live in AdMapix, filterable by country, publisher and app ID. Create a free account to run this teardown on any publisher in your category, or see plans for full historical coverage.

Data source: AdMapix third-party ad intelligence estimates, United States, retrieved 2026-08-19. Impression, placement and download figures are modelled estimates for comparative analysis and are not publisher-reported. Figures change as new creatives enter rotation.