Humans vs machines: the strategic balance between automation and curation in media intelligence.

 

 

Businesses rely on intelligence garnered from media sources as a foundation for critical decision-making; a fact reflected in the 2018 AMEC study[i] where more than half of companies put emphasis on insight over evaluation in their media reporting. At the same time, however, one in four media measurement companies report an increase in demand for fully automated solutions. Can solely machine-driven solutions deliver the audience insight that companies need for success, or are businesses falling for the attraction of a flashy data dashboard? In this article, we explain why it is important to strike a balance between automation and curation in media intelligence.

 

Technology has changed the face of the media industry, producing an explosion of online, social and digital channels that create and share content by the second. Similarly, it has boosted our capacity to measure and monitor these channels, with the ability to quickly and cost efficiently collect, process and display vast quantities of data much faster than the human brain. With more media content being generated in digital formats, and increasingly intelligent tools such as optical character recognition and context analysis being employed, automation can help us to make sense of it all.

 

But speed and quantity doesn’t necessarily deliver quality. The key is to understand the limitations of technology, the parameters of the media that matters, and the context of how you are planning to use the information within your business.

 

Computers operate best on the heavy lifting side of data analytics. They excel in accuracy over large quantities of data, and in counting and mapping anything that is tagged or categorised. Companies operating in formal settings, such as legal environments or on national news channels, where communication tends to remain within searchable parameters of terminology and format, may find that automation alone can deliver strong results.

 

But, if yours is one of the majority of companies today that communicate through a mix of earned, owned, shared and paid media and generate a broad variation of data – from text and video to images, emojis and GIFs – computer-based measurement on its own will not provide accuracy or insight, and rarely justifies the investment without the element of human curation.

 

Consider the difference between how a machine might code an extract of video versus how a human can also add context based on expression, gesticulation and a host of other factors. Different generations, communities, nationalities, social groups and even individuals personalise their own forms of expression and communication at such pace and complexity that computer science simply can’t keep up. Furthermore, sentiment, nuance such as sarcasm and humour, and even punctuation can dramatically influence interpretation yet fail to be registered by machine-based analytics. Data can be crunched but true insight – telling us what we don’t know that will make us do something differently – requires a human mind.

 

Furthermore, computers alone can’t say what ‘good’ looks like for a brand, what media really matters or how a phrase fits into the context of its environment and if it is true to the company’s positioning. For example, messaging delivered verbatim lacks that much sought-after quality of authenticity and fails to reflect truthfully how audiences and influencers think, feel and act.

 

Without question, automation has advanced over the years, but to date even the best computer-based sentiment analysis programmes peak at 70 percent accuracy, which means more than one in every three articles are wrongly reported in an auto-only scenario. Can solid strategic business decisions be made on data that is one-third incorrect?

 

Undoubtedly automation will continue to improve, and this should prompt more demand for qualitative and strategic insight. The two components go hand in hand. But the key to successful media measurement is defining a blended approach of machine and mind – automation and curation – that delivers the right quality and reliability of actionable insight so that your business decisions are based on the information that really matters.

[i] AMEC Global Insights Study, 2018