Campaign Management
Gaming & Apps

Predicting UA campaigns’ future to maximize ROI at KSG Mobile

Equipped with daily LTV predictions, KSG Mobile optimizes UA campaigns’ ROI and easily demonstrates marketing ROI to stakeholders.

Predicted LTV within 5% of the actual value
Pecan’s predicted LTV for 2023 campaigns was within 5% of the actual value, reflecting Pecan’s models’ accuracy
Mobile app publisher
Company Size
Creators of over 200 games; millions of downloads on Android and iOS
Predictive LTV for marketing campaign management
Data Stack
BigQuery, AppsFlyer, Looker Studio
“We know we’re making good decisions about where we’re directing marketing dollars.”


Mobile game publisher KSG Mobile, Inc. (KSGM), was ready to ramp up their marketing spend and make a bigger bang with their social casino games. But with added investment came a stronger need for user acquisition managers to make optimal, data-driven campaign decisions and verify marketing performance for finance and executive stakeholders.

“We wanted to be better at articulating what was happening and what these results meant in the short and long term,” says Mike Eisele, vice president of business development for KSGM. Mike makes budget decisions for KSGM marketing and user acquisition and reports to the C-suite on his team’s results.

The user acquisition team had a reasonable idea of how their campaigns performed in terms of users’ lifetime value and return on ad spend (ROAS), based on data collected in AppsFlyer. They had used rule-based calculations to estimate user LTV, but these weren’t as accurate as they wanted, and they didn’t provide the confidence to make significant campaign decisions.

Additionally, they didn’t have user-level predictions that they could aggregate in multiple meaningful ways (Geo, Channel, Campaign, etc.). But they wanted deeper understanding and more certainty about their plans and daily campaign decisions — advantages that Pecan’s predictive analytics solution could offer.


With Pecan, KSGM now uses four predictive lifetime value (pLTV) models built on the user, engagement, attribution, and revenue data. These models project the revenue they’ll gain from the users acquired through different channels and campaigns. Each of the models looks at those users’ and channel/campaigns’ future within specific time frames: from Day 1 to 7, Day 7 to 30, Day 30 to 180, Day 60 to 360, and Day 60 to 720.

“We’re a relatively small, lean team, so outsourcing LTV prediction was prudent, and would help us accomplish what we needed to accomplish,” Mike says.

The pLTV predictions for specific campaigns serve two main purposes:

  • Shorter-term predictions help KSGM’s user acquisition team understand if specific campaigns will likely provide desired results, providing early indications of campaign success or failure. Campaign budgets and creatives can be adjusted based on early predictions. 
  • Longer-term predictions show how different marketing channels are paying off (or not), informing future budget planning.

KSGM sends Pecan’s predictions — refreshed daily — into a Looker Studio dashboard that informs both retrospective reporting of results and day-to-day marketing decisions.

Pecan can see around the corner to those campaigns that may be outperforming longer term, without having all the actual revenue signals there early on.”

— Mike Eisele, vice president of business development, KSG Mobile, Inc.


KSGM has found both reporting and campaign decision-making are more informed and confident as a result of working with Pecan.

The user acquisition team found they were able to identify underperforming campaigns significantly earlier than with their previous campaign management approaches, leading to direct savings.

On a daily basis, the UA team can consult their prediction-enriched dashboard to examine their campaigns’ projected results and take informed action fast. For example, the team might see a campaign that appears to be underperforming based on Day 1 to Day 7 pLTV predictions. But with Pecan’s trustworthy predictions, they can feel confident that by Day 30, that campaign will hit its targets. 

“These are signals that go beyond just looking at Day 3 as you normally would,” Mike says. “We know we’re making good decisions about where we’re directing marketing dollars.”

Reporting on marketing performance and user acquisition is made easier by being able to validate previous forecasts with real results that are very close to anticipated returns, thanks to Pecan’s reliable predictive models. 

“We’ve seen very accurate projections in the longer term. We’re considering even longer-term models to see what our users are doing after Day 360,” Mike says. “As a social casino app, that long tail matters a great deal. We have to feel confident that Day 30 looks good, and that we’re going to get where we need to be at Day 180 or Day 360. Pecan’s predictions give us a lot of comfort in knowing that’s going to happen.”

Overall, Pecan’s predictions make it possible for the KSGM user acquisition team to understand, validate, and improve marketing performance, providing a vital long-term perspective even in the early days of campaigns.

“Pecan can see around the corner to those campaigns that may be outperforming longer term, without having all the actual revenue signals there early on,” Mike says. “We’re more comfortable both with seeing which campaigns are out- and underperforming, but also validating that long-term LTV curve.”

Get the full story