Senior Director, Data Science & Engineering and Head, AI, Meltwater
B2B marketers are seeking inspiration from B2C strategies and campaign management ideas to win more customers. It is only a matter of “when” B2B marketers start connecting with their audience using brand stories, based on media intelligence platforms. Last month, media intelligence giant, Meltwater, came up with a unique report on the social media buzz created by brands to win over customers leading up to Singapore’s 52nd birthday. The study came close on the heels of the company acquiring Cosmify, a knowledge discovery engine.
We spoke to Aditya Jami, Senior Director, Data Science & Engineering and Head, AI, Meltwater to understand how B2B marketers could use predictive media intelligence and data science capabilities to connect to audiences.
MTS: How do you enable customers to use media intelligence to connect to audiences especially in the B2B segment?
Aditya Jami: B2B companies face a unique challenge. Their audience is small but highly targeted and valuable. We want to empower B2B companies with the tools to identify their audience of journalists/influencers and potential clients.
With AI and sophisticated boolean searches, Meltwater helps reduce false positives and ensures B2B companies are able to find the stories that mention the brand — rather than generic terms or beats. They are also able to keep track of known competitors or new market entrants.
Trends and insights on hiring, ad spend, media coverage and external data are only valuable when it’s in the right hands. With the Meltwater Mobile application, clients can reduce the time to share these insights internally.
MTS: Would you share a customer story on how they used Meltwater to track media coverage patterns?
Aditya Jami: Gogo’s rapid growth triggered a need to track, identify, and understand market trends. As they transitioned from being a private company to a public company, Gogo required insights about what the investment community was saying.
Reliable metrics, clear analysis of competitors’ coverage, and the impact of industry trends could provide perspective and enable more informed decision-making. The quality of sources, ability to filter searches, and sharp focus on chosen industries and the global competitive landscape make Meltwater the ideal platform for Gogo’s fast-paced growth.
MTS: How do you see AI/ML proliferating further into B2B software platforms, by 2020?
Aditya Jami: In the beginning, consumer facing internet companies were able to leverage AI/ML and create the most impact. Examples include: search ranking, revenue bearing systems and recommendation engines. We have seen the rise of practical AI/ML solutions surfacing in the last decade due to the explosion of data generated on the web, mobile, and IoT, with advancements in distributed high-performance computing optimised at the chip-level.
With the evolution of Big Data technologies, enterprise companies have invested heavily in making business workflow processes very data oriented and organized. With the democratisation of scalable AI/ML methods by big companies and academia, you can see an explosion in the number of B2B start-ups (and groups within big companies) solving complex problems in different functional categories including Business Intelligence, Customer Management, B2B Sales & Marketing Automation, Consumer Marketing, Digital Commerce, Finance & Operations, Security & Risk and Productivity.
MTS: How does Meltwater use AI and Data Science?
Aditya Jami: Simply put, our platform is about converting unstructured data from web (anything behind a customer’s firewall) into a structured source to enable Search/Alerting/Analytics capabilities for all our Media Intelligence products. We are using AI and Data Science at several stages:
- Ingestion: Our crawling engine is a fully automated unsupervised full website extraction system. Based on a few samples, it figures out how to navigate web content, what that content is about, and then how to extract structured attributes from the data and send it to further downstream processing. In total, our ingestion engine brings in data from tens of millions of sources across online news, broadcast, social media, reviews, forums and corporate websites. Our recent acquisition Wrapidity has played a critical role in this sophisticated automation.
- Natural Language Processing: We analyse billions of datapoints every day and in order to extract meaningful insights, all our data is constantly augmented with very rich annotations tasks including Semantic (Named entity recognition, sentiment analysis, relationship extraction, topic modelling), Syntactic (POS tagging, brown clusters, and lemmatisation), Discourse (Automatic Summarisation, Coreference resolution) across 20 different languages. We built most of these methods in-house using traditional natural language processing methods but started seeing good advancements for some of these tasks using deep learning approaches.
- Knowledge Base construction: We extract factual knowledge around entities (companies, products, brands, influencers) from all our data sources to continuously update our knowledge graph. We use rule mining and link prediction techniques to ensure high quality of the graph.
- Insights Generation: The insights we compute and disseminate across different product line are driven by statistical and machine learning models (Data Science heavy) on top of our vast Outside Data Repository.
- Active online learning: These are techniques that continuously refine and personalise higher order insights like sentiment (Document level, entity level and aspect level), news feed generation, search ranking etc. based on customer feedback.
MTS: To what extent is Meltwater’s media intelligence platform predictive? Would Cosmify add predictive and prescriptive capabilities to the media intelligence suite?
Aditya Jami: Encore Alerts (a company we acquired last year) brought in their proprietary anomaly detection engine which is now fully integrated into our platform and producing actionable real-time alerts that our customers are using to stay on top of emerging crises, important events, and critical trends. We are definitely excited about companies working in the space of trend predictions and information propagation analysis.
Cosmify, on other hand, is a knowledge discovery engine that can analyze very large and dynamic data lakes to produce vectorised information, and when fed into our existing Data Science platform, automates generation of highly customised outside insights.
MTS: Could you provide a sneak preview into how customers use Meltwater to measure PR ROI with AVE and track social conversations?
Aditya Jami: An interesting story that I’ve heard is from our Singapore office. A local consumer brand is keeping an eye on all conversations happening across their social media platforms like Facebook and Twitter as well as local forums to ensure that all customer complaints are dealt with quickly. With Meltwater, they are able to track product suggestions made by the public and respond in real time to show their customers that they are actively listening. We have seen a significant spike in their fan growth and engagement on social channels since using Meltwater.
Our solution also allows communications teams to measure their success comprehensively. Many of our clients are analysing their share of voice against competitors and tweaking their media strategy based on insights they have garnered from Meltwater. It’s also really important for our clients to take a look at the tonality of the articles and conversations, and our software is designed to take the headache out of manual sentiment rating.
Meltwater is always adding different PR metrics into our product. We have just included metrics for measuring the reach of social media conversations, as well as an integration with Google Alerts to track web traffic. This is definitely an active area of investment within our company that we will be keen to share once we start rolling out new features in our roadmap.
MTS: Thanks for chatting with us, Aditya.
Stay tuned for more insights on marketing technologies. To participate in our Tech Bytes program, email us at firstname.lastname@example.org