Skip to main content

Harnessing the power of data network effects

(Image credit: Image source: Shutterstock/Carlos Amarillo)

Everywhere you look today, companies are trying to replicate the strategies and systems that underpin the runaway success of tech titans like Amazon, Google, Facebook and others. From fledgling tech startups pioneering new and emerging technologies to the age-old industry stalwarts on the path to digital transformation, everyone is after the same superpower: data network effects.

Similar to the traditional concept of network effects, data network effects take the phenomenon and apply it to data. That is, the more data added to – or consumed – by a product or service, the better it gets.

Take Google Maps, for instance: as more people use it to navigate, the more data Google has available to refine key features like route suggestions and estimated arrival times. The improved features then attract more users, which in turn drives more data into the system, creating a virtuous cycle that continues to accelerate over time.

It’s no surprise, then, that today’s most successful products and services are all powered by data. In fact, the impact of data network effects on the user experience is so powerful that consumers and enterprise users now expect personalised experiences in their regular product interactions.  

In many industries today, leveraging data to optimise product and service experiences remains a strategic differentiator for companies that get it right. But, as we reach the inflection point where the technologies required to create data network effects become available, affordable, and easy to use – which in some areas is happening right now – this will change. In five to eight years, data-driven experiences will be table stakes and companies won’t be able to compete effectively without them.

I recently came across this company, Naked Labs (opens in new tab), which produces a full body scanner. It’s a straightforward concept, though very innovative because it builds on the value and power of the data. They mention that the more often you scan yourself, and the more users they have, they can provide more and better insights. This is not just the story of a product that leverages data for rich experiences, but one that is essentially a data product; it provides benefits and incentives rich enough to encourage you to provide data about yourself.

So how do you build data network effects into your products and services? Here’s what nearly 20 years mastering them at Microsoft and Google – and now Snowflake – taught me about doing it right:

Pick the right technology stack

One of the core tenets of working at Google is the idea that teams should focus on impact. This is a simple principle that easily applies to every company. An implication is to understand where existing technology can help you.

For most businesses, building out all of the technical resources necessary to drive data network effects doesn’t make sense. Instead, look to the growing ecosystem of third-party vendors to put together the right technology stack. That way, you can focus on how best to put the technology to work growing your business, rather than wasting time trying to figure out how to make the technology work.

At the highest level, you need to create a system that can handle processing large volumes of data very quickly. More specifically, there are four key components of the stack:

  • Data ingestion: Collecting the data you need and ingesting it into your analytics platform.
  • Data ingestion: Collecting the data you need and ingesting it into your analytics platform.
  • Data analytics platform: Developing models from the data and extracting insights.
  • Machine learning platform: Many data-driven experiences leverage descriptive insights, but increasingly they’re leveraging machine learning. Such models can help offer recommendations or tailor the experience based on past behaviours and/or predicted intent.

Take a more open approach to data

We’ve all heard the phrase “data is the new oil” before. It’s a catchy trope – but it’s misguided. Data’s power doesn’t come from its scarcity (like oil), but rather from its abundance and its ability to encode patterns.

When building data network effects, don’t over-index on your own proprietary data. You want to enrich your models with additional data feeds. There’s a likelihood that the more data sources you have – and the broader they are – the more accurately you can model the world and make predictions.  

For instance, if you have a fitness app and are trying to provide “workout of the today” types of recommendations, having regular ingestion of weather data can help improve the quality of the recommendation. Other data sources, like local events information, can further enrich the experience you provide to users.

Get data to close the loop - Measure!

In addition to collecting data on the world surrounding your products or services, you also need to collect data on how users are experiencing the products or services you are delivering, especially relative to what you may have modelled as expected behaviour.

Do not skip this step. It’s easy to overlook – but in the end, closing the loop is what builds the data network effect. Here’s how it works: when you collect data and apply it to your product or service, you are increasingly making predictions about what users want. In that sense, the system is constantly trying to predict what users want and suggest or customise as needed. The only way you can improve outcomes, and therefore increase value, is knowing whether previous predictions were correct or not.

If you miss this part, you will not benefit from data network effects. Worse, it can lead you down the wrong path altogether. Having well-defined metrics is critical here. As you change the data and models that inform the experiences for your users it is be critical to be able to measure and assess the positive or negative impact. In the end this represents a data-driven approach to build data-driven experiences.  

It’s easy to see that the most successful products and services are the ones that close the loop. There are many examples of purchase recommendations of the form “customers also bought...” that lead to sales increases, or “users also watched…” that lead to increased watching.

In today’s hyper-competitive and highly data-driven business environment, the winners are those who are always testing, measuring and iterating. Similarly, the most satisfying experiences will be those that are tapping into the value of data.

Christian Kleinerman, VP of Product, Snowflake Computing (opens in new tab)
Image source: Shutterstock/Carlos Amarillo

Christian Kleinerman is the Vice President of Product at Snowflake Computing. Christian served as General Manager of the Data Warehousing product unit at Microsoft and recently worked at Google leading YouTube’s infrastructure and data systems.