Inside SciPlay’s ELT strategy: more, more, and more data

What can you learn from SciPlay’s ELT strategy?

Every major app publisher that is serious about growing has a Pokémon Go approach to data: catch them all. Every source. Every kilobyte of information. But most are limiting themselves to ad network data and their own first-party in-app data. 

The reality is that there’s more. Much more.

Accessing it is the key to significantly better insight into your growth levers, competitive positioning, campaign incrementality, and CAC.

And a smart ELT strategy can unlock all of that for you.

Check out my recent conversation with SciPlay’s director of ad product, Gal Karniel:

SciPlay’s ELT strategy

When you market at SciPlay scale, you’re stitching together 20 to 50 data sources across ad networks and app stores … and more.

Karniel says the only way to make that useful is an ELT layer that’s flexible, observable, and built to be customized. That’s why his team adopted Singular’s ELT product, Extract, to ingest hard-to-reach APIs, including Apple’s new App Store APIs, pull parallel datasets to resolve platform limitations (like Meta’s dimension conflicts), and enrich core metrics with contextual metadata. 

The payoff: deeper analysis down to the ad placement level, faster troubleshooting, fewer custom-built data pipelines to maintain, and a clearer line of causality from “what changed” to “what moved performance.”

Much more data…

You desperately need all the data Singular has traditionally supplied from your ad networks: the cost data, the deliverability data, the results data. And you need it all combined. And enriched with your own first-party in-app data.

But there’s more available now with the right ELT strategy.

Think App Store and Google Play performance tracking:

• Downloads & deletions
  Validate install numbers, monitor churn trends
• Ratings & reviews
  Surface user sentiment, detect product issues, and feed insights into product/ASO teams
• Purchases & subscriptions
  Get the most accurate revenue and refund data directly from the stores
• Crashes & ANRs
  Track app stability issues that impact retention and ratings
• Engagement data
  Measure user actions in the app store to understand intent
• Acquisition sources
  Identify where installs come from (search, browse, referrals, territory)

Think social organic data:

• Page posts engagement
  See which content drives the most organic traction
• Followers statistics
  Understand organic audience growth and align with paid UA targeting
• Comments data
  Capture unfiltered feedback and sentiment at scale

There’s app store APIs, ad networks additional data, social data, CDP data, CRM or liveops data, and much more than you can pull in. And think about competitive data: top lists on the App Store and Google Play that you can automatically query and bring down to your BI systems for analysis.

Another key part of SciPlay’s ELT strategy: grabbing multiple datasets from ad networks like Meta because they only expose incompatible dimensions and don’t allow you to pull geo and placement together, for instance. Now you can pull both, and while you can’t join the tables due to the lack of a shared primary key, the additional insight still exposes more opportunities.

The hidden opportunity in app store data

As I mentioned above, most marketers limit their data ingestion to ad networks: Meta, Google, AppLovin, Unity, TikTok, and so on. That gets you spend, clicks, and installs, and you can add revenue for the full monetization picture.

But crucially, it misses the contextual layer of what’s happening in the app stores themselves.

In other words, the entire app ecosystem at large.

That’s 1 of the reasons why SciPlay ingests Apple’s new App Store APIs and Google Play data directly into their warehouse. Now they can correlate campaign performance not just with in-ad metrics, but also with and much bigger picture:

• App store events
  Featuring, rankings, reviews, updates
• Metadata shifts
  Creative assets, descriptions, screenshots, categories
• Market context
  Competitive placement and store algorithm changes

This adds a bigger, broader lens.

Instead of only asking if your advertising campaigns moved the needle on your growth, you can also ask what else happened in the store that might explain this. The result is faster root-cause analysis, fewer false attributions, and a stronger feedback loop between UA, product, and ASO.

Because your app doesn’t exist in a vacuum. Other publishers kick off marketing campaigns. Some apps get featured by Apple or Google. External events like movies, holidays, weather, and sporting events influence consumer behavior.

The right ELT strategy, therefore, helps you see much more.

Why Extract?

A perfectly valid question, of course, is why use Extract for your ELT strategy?

The answer: it’s an optimal solution at an amazing price.

Karniel’s team wasn’t actually hunting for a pure data-funneling tool … they were being pitched bundles and dashboards, and considering whether to build a tool for themselves. Extract’s focus on data movement and easy configurability was the perfect answer.

Extract gets SciPlay the data they want quickly, easily, and at low cost. It doesn’t require huge technical chops to run, so product managers and ops leaders can use it themselves. It offers full end-to-end visibility, and it has industry-best pricing.

(Get more on what Extract can do here.)

Key things to remember as you build your ELT strategy

If you’re serious about growth and looking for all the data sources that will enable your performance marketing team to achieve it quickly and efficiently, great.

Here’s a few key things to keep in mind based on SciPlay’s experience:

1. Design for parallelism
   When a platform won’t return all the dimensions you want simultaneously, split your data streams and decide per use case which one you need
2. Don’t stop at ad networks
   App store APIs and contextual signals are underused gold mines … bringing them in-house tightens the loop between UA, ASO, and product
3. Prioritize context, not just KPIs
   Installs/spend/revenue are necessary but insufficient: add store, market, and creative metadata to interpret install and monetization shifts
4. Buy flexible, not rigid
   Out-of-the-box is great until it isn’t, so choose tools with configuration and customization options so you can do what you need to do, how you want to do it
5. Make observability a requirement
   Full logs, timestamps, and run details build organizational trust in your data (and make it easy to debug something when a flow gets interrupted)

Over time, you’ll want even more data sources. The good news is that Extract is continually adding more connectors, so you’ll have more and more simple options to add and further improve your data models.

Much more in the full podcast

As usual, check out the full podcast. There’s much more about SciPlay’s ELT strategy that you’ll find interesting and useful as you boost the signals you’re acquiring.

What you’ll find:

• How SciPlay manages 20–50 different data sources for marketing
• How SciPlay balances in-house solutions vs. third-party tools
• How Extract solved the challenge of Apple’s 50 new App Store APIs
• How Extract can create parallel datasets for greater depth
• Why visibility, logs, and transparency matter for trust in data pipelines
• How Extract simplifies data enrichment pipelines and reduces maintenance
• How SciPlay built faster access to insights, better targeting, and better data for smarter decisions

And don’t forget to try Extract for free…