Startup founders should be looking to drive growth not just through demand generation, but through demand collection, according to ex-Netflix data engineering and analysis lead Satya Kunapuli.
Netflix has gone from a niche market disrupter to a household name within the last five years, with 100 million global subscribers and 45 billion hours of content viewed per year. But even with a marketing budget of $US1 billion to promote 400 Netflix original titles to 190 countries worldwide, the streaming giant still has data-driven analytics at the heart of its growth.
One of the key aspects of Netflix’s growth strategy is what Kunapuli describes as demand collection: analysing, segmenting and comparing the swathes of data Netflix receives to optimise its content.
Kunapuli, who only recently stepped down from his role at Netflix and has previously worked at Disney, Yahoo, Intuit, spoke at the Interactive Minds 2017 Digital Summit in Melbourne last week about how the streaming giant uses innovative techniques to measure performance.
Data driven decision making
“Even today in spite of all the tech advancements and data that’s available, the measurement techniques used to measure performance are still flawed,” Kunapuli told the summit.
According to Kunapuli, Netflix has moved away from traditional call-to-action marketing, instead operating on the “bleeding edge” of data-driven marketing development.
“The kind of measurements … [Netflix is] looking to do are radically different from the past: we are on a new measurement journey … with no ground rules to what’s wrong and right,” he said.
He advises marketers should move away from the “pure correlation” based data analysis, which is the cornerstone of traditional marketing, and instead look at “pure causation” based analytics, which are measured through incremental customer conversions from advertising campaigns.
With 400 original titles that need to be promoted and go to market in 190 countries, scaling the marketing operations globally is key to Netflix’s success.
But instead of trialling new marketing campaigns in each geographic location and seeing which yields the best returns, Netflix engages in “quasi experiments” done on the geo-regional level, said Kunapuli. These include “burst and delay tests” and “spend mix tests” to test the potential outcomes of marketing spends.
Kunapali told the summit utilising incrementally based algorithms is the key to discerning accurate growth metrics.
“There’s not point automating something based on the wrong metric — it just gets you south quicker, “he said.
“A/B testing is the bedrock of how you measure causality: it may cost a bit but it’s very much worth running A/B tests to measure incrementality and causality.”