The point we’ve reached, this fast-paced world we live in where 2.5 quintillion bytes of data are produced every single day, needs fast-paced intelligence that befriends a couple of chummy and efficient robots. Only in 2020 —the unforgettable year of the pandemic— each of us generated, on average, 1.7MB of data every single second.
Asking any of us to work hand in hand with robots seems like a dream stolen from a 70s movie script, but now it’s perfectly possible. Ultimately, swift real time data analysis needs more than the human brain and artificial intelligence (AI)is an invaluable and almost unskippable partner.
The AI working with real time data requires, in turn, to be strengthened by machine learning so it’s capable of “learning” and building models once it finds patterns out of the huge amount of data it’s analysing. Data that, by the way, you and we all produce.
“What is all this explanation for?” you may ask, while reading this article on your smartphone.
Well, it comes down to a simple fact: real-time data, when cleverly used, improves business decision making, timely and as accurately as possible that can help organisations stand out from the competition or, better yet, be more relevant to the customers and users they serve.
Ongoing digitisation and automation, that the pandemic has only accentuated, has increased n-fold the need for businesses to find ways of retrieving data faster and get smart insights out of it.
Today, things quickly change, news—fake or not— can go viral in minutes, brands get backlashed or are greatly celebrated in quick time lapses. And all this, often, is the result of what some like to call “the press release of the XXI century, a simple tweet that unleashes the rage of Internet users.
In this article, we build the case for real-time data and why we all need to be using it to make our organisations more relevant.
Stored data vs real-time data, which one AI prefers?
If AI can learn from data sets collected in the past —or historical data— one may wonder what the advantage of using real time data is.
Although historical data is, by all means, relevant, the models that AI creates based on them are more useful in a context or reality that’s similar to the one it learnt from. However, and we purposely have mentioned it lines above, the world is in a “fast pace” mode. Because input data are changing more rapidly, patterns follow this trend as well, and models are less and less likely to be repeated in the future.
We want to share with you a vivid example that relates to that. Let’s recall the COVID-19 pandemic once again. By February 2020, we were finalising details for the launch of a publication that we had been preparing for several months. Then, suddenly, within a couple of weeks, the pandemic hit, and all the work, data and findings were (most likely) no longer relevant and had to be redone.
Although real time couldn’t have fully “saved” the launch, the use of it would have helped us update the data and related findings quicker. The result would have been a more nuanced publication, with up-to-date information that could have reflected the reality we were living.
Real-time data makes AI more powerful. To models built upon the historical data, AI adds information that’s contextual and reflects reality (model calibration), so the likelihood of getting more powerful insights for business decision making in terms of brand reputation, product marketing, sales strategy, etc is high.
A little parenthesis here, but we think it’s important. When one reads about real-time data, the expression “streamed data” pops us frequently. This is, simply, data that flows continuously and has no end. Part of them are stored for later analysis but they can also be used “on the go”. That’s real-time data analysis.
Quicker isn’t weaker: building the real time data case
What better evidence of the need for quicker, more real-data data than our own life. The last 15 months have just shown us to what extent there is an urgent need for quicker, real time data. News and events around the pandemic happened overwhelmingly frequently and, even today, news on the vaccination process and issues with the vaccines sometimes arrive several times a day.
Alternative metrics —data coming from social media, web scraping, search trends, mobile technology, wearables, IoT devices, satellite images, geolocation, etc— are valuable to capture the voice of the users and consumers directly, correct mistakes or omissions quicker and, ultimately, improve the performance of the business.
We don’t want you to get the facile idea of what’s at stake when using real-time data, however. It needs an ecosystem where both AI developers and the infrastructure that supports machine learning work tête-à-tête. By infrastructure we mean the tools to gather real time data (streamed data), to train AI, manage it, and to deploy the models resulting from the machine learning capabilities.
The value that alternative metrics bring is twofold:
They enable us to take a quicker and more frequent “pulse” or “screenshot” of the situation, i.e. public perception around a matter, awareness of a phenomenon, etc. With this information, decision-making, operational efficiency and personalisation are reinforced.
They push organisations to reevaluate and adapt their strategy more frequently; even the readiness to adjust business models for the new realities can increase. It’s like running the business always in beta mode. We foresee that traditional, well-known procedures to collect, analyse and distribute information are poised to change. For instance, moving gradually from the use of slower data like surveys or national stats —that are only available annually and aren’t always straightforward enough to take business decisions based on it— to using more real-time sources.
Already in 2018, according to a Forrester survey, fast data solutions were the answer to cope with an increasingly complex reality, but many missed the mark. In the same study almost 90% of organisations reported that “they typically require their data to be ingested and analysed within one day or less, and that 88% need to perform analytics in near-real time on stored streamed data”.
And Covid-19 didn’t exist yet (or at least that’s what we think).
Real time, when is more commonly used, so far?
Although the advantages of using real-time data can be harnessed in every single industry —providing that all of them are, to a greater or lesser extent, already digitised— certain industries are one step ahead. Or, to be more precise, certain practices within those industries.
One of them is the financial sector. For instance, the analysis of real-time transactions by machine learning algorithms allows officers to fight fraud. AI identifies patterns that help to detect fraudulent transactions or even stop them. The same goes for cybersecurity. The technology fighting cyber-attacks works in tandem with people: the more informed and up-to-date they are, the smarter the cyber-defense decisions they make.
On the market side, financial operations are being upgraded with the use of real-time data. By detecting errors with real time analytics, financial statements are more accurate which, in turn, reduces operational risks. And stock market traders use data from assorted sources —being social media and weather forecasts among them— to make smarter decisions. Credit scoring, although not yet 100% accurate (this podcast tells you why) is also using read-time data to approve (on not) credit or subsidies.
Also, the use of wearable and mobile technologies are advancing personalised healthcare. They collect, analyse and share information on a person’s health situation, helping them to keep track on their health, make better decisions and even save their lives. Smartwatches, for instance, include health apps more and more. By being connected to mobile networks or to home’s wi-fi, they can share data in real-time.
The aviation industry uses information that sensors collect to improve flight safety and also to reduce maintenance costs. In the extractive industry —the ones using technology to drill, pump and mine— are also using real-time data to make their heavy machinery more efficient and environmentally friendly.
This non-exhaustive list can not exclude online marketing, being social media the most conspicuous and well-known example of how timely monitoring of the “happening now” information is key to improving engagement, advertising and selling, preventing reputational risks and offering better customer service.
Some tangible use cases today
We wanted to go a little further with the examples on how the use of real-time data is helping businesses and organisations to differentiate and offer better services. In some cases, we’ve made the example anonymous. However, the use of the data is what we like you to take away.
The first use case we wanted to talk about is a solution that combines social media’s big data that connects, in real time, to a brand’s KPIs. It’s called Real-time Brand Valuation. For instance, this solution is supporting a business in the automotive sector to assess brand strength in real-time across three continents. It is providing valuable insights of changing consumer behaviour due to the pandemic, and allowing the business to adapt both messaging and offering, replacing traditional customer surveys.
A well-known online retailer uses real time data to detect transactions whose likelihood of being fraudulent is high. Also, it optimises customer offering and sales with the use of real-time data.
An Asian car-maker giant is tracking information on product preferences (car category, colour and mode) to understand demand and make better decisions for each local market.
A leading firm in the extractive industry is using real time data to foresee when one of the machine parts may fail so operations don’t get disrupted.
What we think
Benedikt Jonas, Director at PwC Luxembourg
Empowering business users with real-time data and clear insights allows them to follow market, brand and customer data on the go. This leads also to a faster response to grasp new market opportunities and mitigate reputational risks.
