In November 2023, Grid Pulse recorded 38% Day-1 retention and 18% Day-7 retention in Turkey. We chose Turkey because our team was building in Istanbul, which made player feedback accessible, and because its mobile market offered the engagement conditions we wanted to test. The following four months taught us more about the limits of soft-launch data than about the game itself.
Why Turkey Works for a First Test
Turkey has a large, active mobile audience with varied genre preferences. CPI for casual and puzzle games is lower than in tier-1 markets, so a statistically useful cohort costs less and leaves more budget for development. Its broad demographics also expose differences in session frequency and play depth across age groups.
For Grid Pulse, Turkey produced day-1 and day-7 retention data quickly. Day-1 was 38% and day-7 was 18%, each measured against the acquisition cohort for that period. Both results were acceptable for a casual puzzle title, and the cohort showed the engagement patterns we expected.
These are internal Bold Games measurements from our own live titles, not benchmarks for the wider industry.
What Turkey Cannot Tell You
Our mistake, also common among early-stage studios, was trying to use Turkey to read IAP conversion. Turkish players engaged deeply with the systems, but conversion was lower than in tier-1 markets and did not predict the UAE or Saudi Arabia. We spent three weeks debating price points from Turkey data before recognizing that the dataset could not answer that question.
Turkey is useful for testing engagement mechanics, session pacing, and retention-curve shape. It is not a reliable market for calibrating willingness to pay.
The day-7 to day-30 retention slope was also steeper than our pre-launch models predicted. Drop-off accelerated in weeks two and three compared with sandbox testing. That was a valid signal, so we revised event cadence before entering MENA. That change alone justified the Turkey phase.
MENA Is Not One Market
We added the UAE and Saudi Arabia as Grid Pulse soft-launch markets in early 2024, then added Egypt a few months later. Their differences were large enough that one analytics block labeled "MENA" would have produced bad reads.
UAE players showed early-adopter behavior: they installed soon after seeing an ad, explored features deeply in early sessions, and a meaningful share made an IAP within the first week. UAE CPI was high, but the conversion signal was useful. For monetization timing, the UAE gave us a read faster than almost any other market at our budget level.
Saudi Arabia produced strong IAP volume but needed more time. Acquisition was slower, first-week sessions were longer, and conversion usually arrived later, around day 14 to 21. Reading Saudi Arabia on the UAE timeline created misleading early results. We nearly labeled the economy broken before its conversion window opened.
Egypt supplied volume at lower CPI and required localization, making it useful for testing Arabic-language UX and notification copy. IAP conversion was low, while engagement depth was real. For understanding how a large Arabic-speaking casual audience interacts with game systems, Egypt was an underrated test market.
How to Read CPI Data in Early Campaigns
During the MENA phase, we trusted early CPI too quickly. With limited budget, a new market, and an uncalibrated audience, the first two weeks mostly reflected campaign-targeting inefficiency, not just game performance. UAE CPI swung 40% between week 1 and week 3 as the algorithm learned the audience. Using week 1 CPI to judge market fit put us in the wrong frame of mind.
Our current rule is to wait for at least three weeks of spend data in a new market, with a minimum cohort of 1,500 installs before making optimization decisions. Below that size, player-behavior variance overwhelms the signal.
Why Avoid Soft Launching Entirely
Some studios we know skip soft launch because the data is noisy and the time cost is high. That argument can be valid. For a highly culturally specific game or an advertising-only monetization model, testing in markets unlike the target audience can add delay without proportional signal.
We are not saying every studio should follow a Turkey-to-MENA path. The value depends on the gap between the target market and available test markets, plus how much design flexibility remains at soft launch. If the game is locked and the purpose is only UA learning, targeted TestFlight or closed-beta cohorts may use the budget better.
For us, Turkey exposed a material retention-curve problem in Grid Pulse before wider release. Fixing it justified the entire phase. MENA showed that monetization timing was calibrated for higher-spending markets. Running the relevant tests was worth the cost, but the answer depends on which questions remain open.
What We Would Do Differently
We would split the engagement and monetization hypotheses at the outset and assign each to a market. Turkey answers engagement. UAE answers monetization timing. Egypt answers localization. Running all three in parallel instead of sequentially would have saved 6 to 8 weeks of learning time.
We would also define kill criteria before launch: if day-7 retention falls below X, pause and redesign the second-week event structure before continuing. Setting that rule in advance makes it harder to rationalize data that says to stop and iterate.