When did you first become involved in the ‘social intelligence’ sector?
We have been actively involved in social data since the earliest days. We were formed in 2001 initially as the ‘Alternatives Channel Group’ supporting part of WPP (although fully independent). That was the term we used before the term ‘social media’ emerged in 2003.
It was the era of the ‘Cluetrain Manifesto’ (still a required read) when the technologies that eventually become social media emerged. We had strong success – Digiday named us Social Media Agency of the Year in 2008 – for example and we helped a lot of big organizations structure their first social programs. 2008 was also the year that we then pivoted towards building our software.
Because we had so many people analysing social data for our campaigns, we inadvertently ended up with what is now referred to as ‘training data’. We began working with Dr. Philip Resnik, a pioneer in in the application of machine learning to natural language processing, leveraged data we had to train models and were off and running. So, this year represents the 11thyear that we’ve been actively been building machine learning technology for social data analysis and obviously have learned through hard won experience what works and what doesn’t and how it can be practically applied.
Today we have powerful programmatic integrations with a growing range of industry partners, such as Brandwatch, where we provide advanced, sophisticated NLP models programmatically and help leading brands apply this advanced data into areas including CX, brand guidance, brand advocacy and trend discovery.
Even after all this time in the space, we see the most exciting days ahead and powerful new AI powered NLP technology is able to classify language “like humans do,” with great precision and scale, going far beyond just sentiment, and finally help organizations truly drive value from this massive unstructured data resource. Even today, according to Forrester, brands are only processing about 20% of this data set (in large part because of its complexity). But those same brands are hungry for real time (and predictive) consumer insights, so the opportunity the industry is enormous. About half our business is direct with brands and half through ecosystem partners.
What did you want to be when you were growing up?
I’m a bit of a dichotomy. I was an English lit major, and for a period of time flirted quite a bit with becoming a veterinarian. I’m actually licensed as a professional horse training professional (with some wins to my name) and worked with some of top vet clinics during summers in college. I wasn’t a big fan of chemistry though and now fully recognize the irony of that since I’m now knee deep in data science and natural language processing on a daily basis.
My literature background took me into advertising where I became head of the Innovations Group at a division of Young & Rubicam and was member of the WPP.com board before I took the plunge with Converseon. I was always enamoured with language and always energized to be around computational linguists, advanced analytics experts and strategists to are exploring the incredible language data for the wants, hopes, needs, desires and experiences embedded within this data.
What, in your opinion, are the key benefits of social data analysis for organisations?
They’re myriad. Let’s first recognize two things – unstructured data is growing at approximately 60% YOY and is forecast by some to make up 80-90% of available data to consumer-obsessed large organizations in the coming years and, secondly, it remains largely untapped. Forrester says that organizations are processing only about 20% of this data.
The good news is that we have been able to show that the effective processing, filtering and classification of this data – and human level performance – is able to effectively tap into this data resource and help transform market research/insights, customer care and more.
The key drivers are simply ‘better, faster, cheaper’ than traditional approaches, many weighted down by challenges of latency, actionability from pure survey-based techniques. We have this ‘always on’ data set that is providing to have strong predictive power. For example, third party studies have shown that highly performing language models are predicting the outcomes of survey-based brand tracking results up to 12 weeks in advance and are highly actionable.
In other cases, we are discovering the hot new flavours and trends long before traditional trend spotting techniques. We are working with the co-founder of Satmetrix on new ways to model social data to address many of the challenges with traditional Net Promoter Scores which will not only provide ‘better, faster, cheaper’ approaches but also help brands discover what we call the ‘knowable unknowable’: topics, trends, attributes and discoveries that emerge organically from these conversations that couldn’t have been spotted by any other means.
For uber, we were able to set up a global brand tracking nerve center using our advanced models for nuanced brand attributes in more than 8 languages (including Hindi, Arabic etc) that were able to let the organization understand (and respond) at the speed of social. That process also helped redesign what global brand guidance looks like in the this new ‘always on’ world. The key issue is getting this machine learning technology into the hands of the true subject matter experts (not just analysts or data science teams) to build and deploy their own models. The closer the model is to the expertise, in general better the model will be.
Brands today are in greater need of insight than ever before. The market research industry has been a bit slow to change but I believe social ‘voice of customer’ data together with advanced language analysis (with a layover of advanced analytics) will be the game-changer that will help transform this $80 billion industry.
What aspects of your work do you most enjoy?
Our ability to innovate. We’re always challenging our assumptions and have been continuously reinventing ourselves to stay ahead of the curve for more than a decade. We’re agile enough to keep adapting – whether it’s new technologies, such as deep learning, or new use cases for this data. I love working with a cross-functional team that ranges from some of the best minds in natural language processing to strategists who uncover meaningful insights, resulting in better products and more socially responsible organizations.
We have a great team and great technology. For anyone who has built solutions in this space, it can be harrowing at times. The fact that our technology is now established, scaling and being adopted deeply, is obviously something I enjoy these days. We’re running a profit and growing fast, so I’m sleeping much better at night than I perhaps did during some of our initial pivots.
What are the biggest challenges in your day-to-day work?
Definitely the balance between being reactive and proactivity. It’s easy to get lost in the specific needs of the day which can be a serious obstacle on doing the proactive development necessary to evolve offerings and solutions that we know are going important tomorrow. Managing that balance between the here-and-now while investing in the future is always a bit of a tug of war but I think we’ve done a pretty nice job of making work.
Are you worried about regulatory restrictions impacting on your work in the future?
Unfortunately, social data and insights have become polarized in some ways that are unfortunate. There have indeed been missteps by some of the platforms and technology providers, and misuse of the data in other ways, but, in my experience, social data and insights has done exponentially more ‘good’ for society than bad.
You simply need to look at the recent Business Roundtable announcement in the US where 181 leading CEOs signed a document declaring their intention to serve society more broadly in part because (and they said this) of the pressures driven through social media for greater corporate responsibility. Forbes magazine called it perhaps the most important announcement for business in two decades. Of course, the proof will be in the pudding.
You can’t manage what you can’t measure. And in that spirit, we will be announcing some important work to hold these brands accountable to the declaration probably in November. Stay tuned.
The other thing we are doing is applying our technology to other forms of related data such as review sites, mainstream media, call center transcripts, long form survey verbatims. We see social data as a subset of ‘voice of customer’ data and there is a massive need to make sure these voices are accurately and effectively heard.
Which social data analysis tools do you use? Do you have a favourite?
We have integrations with a wide range of platforms and are pretty familiar with the strengths and weaknesses of them all. We will be making more announcements in the partnership area soon. We do have quite a strong partnership with Brandwatch, where we work closely together with some of their larger clients who are using this data in pretty advanced ways, including IBM, Walmart and Uber. In our view, data is going to be the critical bridge to overcome the scepticism towards this data among market research and data science pros who are highly data driven and will have high expectations for data performance and classification. That’s what we are focused on day in and day out.
Do you think the ‘social intelligence’ community needs its own professional body?
Yes, on several levels. It should advocate for clear standards around data performance. There are too many product claims that are creating confusion in part because inconsistent methodologies on how to test performance. There is also inconsistency just on some basic definitions like ‘sentiment’. The term ‘AI’ is thrown around too loosely and do a disserve to many brands who want to make better use of this data but have a hard time delineating between what’s real and what’s marketing hype. If you noticed in the last Forrester Wave, the analysts called this out specifically and warned brands to not take claims at face value but instead test solutions. We can all do a better job here.
There also needs to be advocacy for this data with the CX and insights folks about why and how this data is powerful and compelling. We’ve made strong headway here but there are more joint academic, solutions, practitioner opportunities for studies and partnership. We live in a world today of data science and our approaches and solutions need to withstand rigorous review and evaluation by these teams who live in world of model evaluation. Avoiding AI bias, and creating ‘ethical AI’ will be crucial. We’re rapidly moving into a new era where this data will transcend traditional social analyst usage.
And, of course, we need to make sure political and regulatory bodies understand that analysis of this data can be done ethically and to everyone’s benefit. It’s important it remains a mirror of ourselves. If we push this opinion underground, into dark corners, it will have a negative impact on everyone.
Name a book you would recommend to others.
It’s old but I still love the idealism of Cluetrain Manifesto (and it was the inspiration for our name). But we’ve been reading a wide range of books lately and have partnered and collaborated with many of the authors including Marketing in the #Fake News Era with Peter Horst, our friend Shiv Singh’s Savvy: Navigating Fake Companies, Fake Leaders and Fake News in Post Trust Era, and more recently Chris Burggraeve’s Marketing is Finance is Business. Analysis is only useful if it’s meaningful. And of course, as someone who remains wary of how language and speech can be manipulated, I recently reread George Orwell’s 1984. It’s incredibly timely as we all consider how to better apply this data for the common good and to enhance individual freedoms. Language is power.
Which person, living or dead, would you most like to follow you on social?
I’ve always been a fan of the beat writers; traveling the world in search of love, hope, music, and inspiration, and some of those writers really helped me first fall in love with language. Jack Kerouac for one, would be one of those. When he would write, “There was nowhere to go but everywhere, so just keep on rolling under the stars…” I can’t help but think that would make a great tweet, no? And I’d love nothing more for social media to express a bit more optimism, less politics and reflect more of the poetry of life.