Cutting events from 30 days to 14 raised IAPs by 50% — longer events aren't better

In one line: The casual game Mergic replaced one 30-day battle pass per month with two 14-day passes, and in-app purchase revenue rose about 50%. The reason wasn't selling more often — it was that for half of those 30 days, players had already finished, and the event was an empty shell.

Terminology follows references/glossary.md.

The numbers first

Mergic is a Merge-2 casual title developed by Diesel Puppet and published by Azur Games. Its problem was a familiar one: good retention, flat revenue. No meaningful DAU growth, no install surge, no market tailwind — but a highly loyal base, with over 60% of players active in the game for more than two months.

Its event structure was a standard battle pass (the source calls it a Season Pass):

• Each season ran 30 calendar days
• Players completed a fixed set of tasks inside that window for a high-value reward
• Average playtime was around 45 minutes per day

Then they noticed something awkward: at normal play intensity, the pass was finished in two weeks — ten days for fast players.

The knock-on effects were consistent. Once tasks were cleared, in-event activity "dropped sharply" and progression slowed markedly. Some players abandoned the event midway, not because it was hard but because it was too long. Meanwhile, in-app purchases spiked on the first day of every season — the source says daily purchases could jump tenfold.

That last detail is what actually hurt. When revenue can jump tenfold on a single day, you cannot compute a break-even point for user acquisition.

So they changed one small thing: the October update replaced the 30-day season with two 14-day seasons — the 1st to the 15th, then the 16th to the 30th.

The result: within two months, IAP revenue rose 50%, with a comparable increase in IAP conversion. In-app purchases account for roughly 70% of the project's total revenue, so the magnitude matters.

Why an over-long event is dead time

Start by correcting the intuition. Most people assume a longer event is more forgiving and therefore better for players. Mergic's data says the opposite: when event duration substantially exceeds the time players need to consume the content, the extra days aren't a benefit — they're dead time.

Three steps in the causal chain:

Step 1: front-loaded rewards make the participation curve an inverted V, not a flat line. Rewards are spread evenly across the event (the source stresses Mergic has no PvP pressure, distributes rewards throughout, and lets players set their own pace). Players don't behave evenly — they clear everything in the first 10–14 days. Once cleared, the event is worth zero to them.

Step 2: during that dead stretch, the event does nothing but occupy space. Players didn't churn (retention stayed high), but the event generated no incremental engagement. So a "30-day event" was really 14 effective days followed by 16 days of shell. Worse, the stretch is actively negative: some players quit midway because it felt too long, meaning duration itself was manufacturing churn.

Step 3 (the key one): a long cycle compresses purchases into very few moments. One season per month means one moment worth buying at — day one. Everyone who wants to buy piles into the same day, producing a tenfold spike, with flat days either side. That is terrible for the P&L: UA payback models assume revenue is predictable, and this pulse structure is precisely what destroys predictability.

(Steps 1 and 2 are source claims — the article's own observations and attribution. The framing in step 3 is this card's inference; the source only says the day-one spike made break-even planning difficult, without abstracting it into "long cycles compress purchase moments.")

Shortening the cycle fixed both problems at once: it removed the dead time and turned one purchase moment per month into two. The source frames it as testing a hypothesis: "if we shorten the event duration, will players still be willing to buy shorter passes? They were."

Why they could do it: loyalty is the precondition, not the result

This part is the most portable, because it draws the boundary.

Before acting, Azur checked three things, all pointing the same way:

1. Pass repurchase rate was around 60%. Players weren't sampling; they were repeatedly buying the same thing. The source gives an example: a clover event run every three months that many players buy again and again.
2. Over 60% of players had been active for more than two months. These people wouldn't leave over a cadence change.
3. Payment in this genre is impulsive. Merge-2 and Match-3 players respond well to countdown-timer offers — buy now, or the opportunity and the bonus rewards disappear together.

In other words: shortening event duration is a lever that only works under high repurchase, high retention, and impulsive payment. It doesn't create demand; it releases already-proven demand twice a month instead of once.

Where those three conditions don't hold, the same move becomes straightforwardly harmful — see Scope and limits.

How to use this mechanism

When an event's duration substantially exceeds the time players need to complete it at normal intensity, compress the duration to "just enough to finish, plus a few buffer days" and schedule a second event in the freed time — because the surplus days generate no incremental engagement, only dead time and pulsed revenue.

To test whether duration is "substantially" over: one ratio. Divide the average days needed for full completion by total event days. Mergic was 10–14 against 30 — under 0.5, a textbook case of length roughly doubled.

Transferability:

• ✅ Battle passes and seasonal events — the most direct fit, since their value depends entirely on whether the progress bar still moves
• ✅ PvE-only casual genres with no PvP pressure — players don't lose their goal by finishing early
• ⚠️ Events with PvP or leaderboards — duration carries the competition cycle rather than content consumption; shortening changes matchmaking and season settlement. Don't copy directly
• ❌ Content-driven products (open world, MMO) — duration is the content; shortening means cutting it

Mergic's specific conditions (don't mistake these for a general rule): narrative is the game's strongest differentiator, rendered directly on top of gameplay with no screen switching, so players are held by the story rather than the event. Shrinking the event is safe because retention is doing the work somewhere else.

Three things you can do tomorrow

1. Compute one ratio: days to full completion ÷ total event days. If you can't, pull the last three events and find the average day on which players who reached 100% completion got there. Above 0.7 means your duration is reasonable; below 0.5 means half your event window is a shell.

1. Plot last month's IAP revenue by "day of season." If day one dominates and the rest is a flat line, you're running Mergic's pre-change pulse structure. It also explains why your UA payback model keeps being wrong.

1. Check your pass repurchase rate. Below 60%, leave duration alone — fix why players buy once and never again first, because shortening the cycle multiplies purchase frequency, and it can only amplify a repurchase habit that already exists.

Appendix

Source list

• We cut the duration of events in a casual game in half — and increased IAPs by 50% — first-party account (vendor), publisher Azur Games' official blog, 2026-02-26. The sole source for every figure and case detail in this card: the 30-day season, 45-minute daily playtime, 10–14 day completion, tenfold day-one IAP spike, 60% repurchase rate, 60% of players active over two months, the October switch to two 14-day seasons, +50% IAP, ~70% IAP revenue share
• Mergic on the publisher's games page — used to confirm genre placement as casual merge (not itemised against the source)

⚠️ Single-source warning: this card rests entirely on one article published by Azur Games itself, with no third-party corroboration. That is the direct reason for evidence tier A rather than S, and confidence medium.

Evidence and sources

• Source type: first-party account. Azur Games publishes Mergic and the blog also serves to attract developers to submit games — it has a stake, and successful cases are almost certainly over-represented.
• Evidence tier A, confidence medium: specific figures and a specific decision process, but no control group and a single source.
• Measurement definitions:
• "+50% IAP revenue" — the baseline is unstated (month-over-month before the change? year-over-year?) and no absolute figure is given
• "Tenfold day-one spike" — the comparison day is unstated (against a normal in-season day, or against the same day last month?)
• "60% repurchase rate" — the measurement window is unstated (per season or per quarter?), as is the denominator (pass buyers or all active players)
• Every figure is a single-case observation, with no confidence interval and no significance test
• The most important methodological flaw: this is a pre/post comparison, not a controlled experiment. The source explicitly notes "in the fall, the project saw an unexpected surge in in-app purchases," then attributes it to the shorter events. Autumn carries its own seasonality, and no control arm was retained (for example, keeping 10% of players on the 30-day season). So the article cannot say how much of the "+50%" came from event duration and how much from seasonality.

Scope and limits

• Genre: casual / merge / Match-3 — high retention, impulsive payment, no PvP pressure. Heavy SLG, competitive titles, and MMOs follow different logic.
• Player base: requires both high repurchase (about 60% here) and high long-term retention (60% active over two months). For new titles or UA-driven products with fast user turnover, this lever doesn't hold and may reverse.
• Region: no regional split is given. Differences in paying behaviour (Japan's high ARPPU, lower payment rates in emerging markets) directly affect whether more frequent purchase moments are accepted.
• Industry stage: applies to mature products in long-term live operations where content supply lags consumption.
• Compliance not covered: a shorter cycle means denser payment touchpoints. Paid design aimed at minors is explicitly restricted in several jurisdictions, and a denser event cadence increases exposure there (see today's OPS-9 card on the EU KIDS Act draft).

What would falsify this

1. If products appear with a duration ÷ completion ratio below 0.5 where shortening the event lowered revenue → the "dead time" explanation fails and something else is doing the work
2. If a product keeps a 30-day season but lifts revenue by the same magnitude purely by inserting a mid-season reward node → the problem is reward distribution, not event duration, and this card's attribution needs rewriting
3. If the same product repeats the change out of season (no holiday, no year-end) and revenue no longer improves → part of the autumn surge was seasonal and "+50%" is overstated

The weakest link in the chain: attributing the entire post-change revenue gain to the shorter event. With no control group, seasonality, concurrent changes, and general autumn market conditions cannot be excluded. This is what most needs external verification.

What data would settle it: a control arm held on the 30-day season (even 10% of users), run for two full cycles. Or repeat the change on a structurally similar product and see whether the result reproduces.

Unverified

• The comparison baseline and absolute value behind "+50%"
• What the reference day for the "tenfold spike" was
• The measurement window and denominator for the 60% repurchase rate
• Whether D1 / D7 retention changed after shortening (the source reports revenue only — if retention suffered, this is not a net gain)
• Whether content cost rose, since denser seasons mean denser content production
• How the approach performs on products with low repurchase rates