Your Enterprise MUST have a Text Analytics Strategy. The time is NOW. Here’s a 5-step guide to designing your firm’s Text Analytics Strategy.
Most progressive enterprises around the world – startups, growth companies and large corporations – certainly have a data analytics strategy. But, most of this revolves around big data models designed around quantitative data. Is that enough?
The short answer is NO.
It is a no-brainer for enterprises to build analytical models around text data. Qualitative data from emails, consumer chat interfaces, tweets, websites and images can be a game-changer to draw strategic insights if one can design an automated process for converting unstructured text data into strategic insights.
After all, as is often repeated by decision makers and leaders, the numbers don’t tell the entire story.
Enter the world of Text Analytics.
Here’s a 5-step guide to help your business build a robust Text Analytics strategy that can act as a strategic game-changer.
A Multi-disciplinary Team
W. Edwards Deming, the world renowned expert on Total Quality Management once said: “In God we trust; For everything else bring data.” Most leaders and decision-makers around the world use a combination of gut and data to aid the decision-making process.
But where is this data coming from?
We believe leaders must assemble a multi-disciplinary team from different business units and functions to provide their inputs – both qualitative and quantitative. This input can be in the form of trend reports, competitor analysis, customer feedback documents, social media summaries, etc.
For a successful text analytics process, it is crucial to find a varied collection of data sources and that’ll come from a multi-disciplinary team.
At teX.ai, we’ve worked with a range of clients on automating their text analytics process. We’ve seen a wide range of use cases: a fintech company analysing credit reports to identify lending-related metrics, a real-estate company spotting trends on where consumers are looking to buy, an automobile company taking corrective action from consumer feedback, etc.
One thing is for sure: Goal Clarity is absolutely critical in any text analytics engagement. The client needs to be certain on what key questions it is looking to answer. Text Analytics can then be used to spot trends, conduct sentiment analysis, understand key areas / topics of importance, etc.
A fintech company was looking to create a set of metrics to automate their lending-decision process. This was a company that served both SMEs as well as individuals. For both these customer sets, there were some standard inputs we used: credit reports, bank statements, income tax returns, etc. We also used the inputs from a registration form. teX.ai could run a robust text analytics process to cull out a collection of insights. These insights not only helped with the lending decision-process but also enabled the creation of personalized marketing campaigns and ideas for new product features.
But there was something extremely important to make this happen.There was absolutely clarity on what inputs were needed for each of these insights. The input data set for conceptualizing new products, for example, included specific entries filled in during registration. It is absolutely crucial to plan this out right. A small tweak – even as simple as adding a text field to a form – can be a game changer.
The Approach: A Clearly Defined Process for Text Analytics
At teX.ai we’ve a three step process for any text analytics engagement. This includes – Text Extraction, Text Summarization and Text Categorization.
Text Extraction is simply this: Automate text extraction from PDFs, images and websites to structure the unstructured data. We automate the process of gathering data from pdf files, websites and third party APIs. The extracted data is then output to a XLSX, CSV, JSON, XML file or written to a database.
The next step, Text Summarization, is about creating a library of bite-sized summaries; These could be from books, articles, journals, reviews, tweets, comments, legislative docs, etc. The summaries could aid several decision-making processes: marketing decisions around consumer reviews, key pain points a product helps solve, key trends, etc. teX.ai can also ensure that the output is easily integrated to a document feeder API for downstream summarization.
The third aspect of teX.ai is Text Categorization. All the short summaries and insights drawn from the summary process can be categorized. teX.ai allows even upto 10,000 categories depending on the segment. For example, if you’re a large e-commerce company running a text analytics process for consumer reviews across the entire website, it’ll be critical to categorize the reviews by category to pass on to respective operations and marketing teams. For starters, good and bad reviews can be segregated and used as needed. Text Categorization can also deliver an automated process for creating labelled text data to serve as an input to your ML model.
At teX.ai, we recently completed an engagement with a world-leader in e-commerce. The company had 100mn+ products listed on its website. This included a range of retailers, brands and categories. They came to us with a clear goal: improve search on the website so consumers would find what they were looking for faster. The idea was that this would improve conversion ratios. teX.ai was deployed and we used the Naive Bayes algorithm to classify the 100mn products to 3400 Google Shopping categories. It worked straight away. The e-commerce player reported a 5x increase in conversions thanks to highly improved search results.
In the case of another client, a market research firm, teX.ai turned out handy for an entirely different purpose. They wanted to check the efficacy of the graphs, charts and data points in their reports by comparing and analysing this data with those on trusted third party websites like the World Bank’s and WHO’s. We developed an NLP-based scraping tool to create an input data collection from third party websites and then ran a comparative analysis. The entire process was automated, saving crucial man-hours for the client.
The point is – the future of Text Analytics is now. Whether you’re in e-commerce or manufacturing, healthcare or luxury, there is a clear need to aid decision-makers with both qualitative and quantitative data. Text Analytics is at the core of deriving qualitative insights. And, just quantitative metrics are a thing of the past.
Give us a shout. teX.ai, a product by Indium Software, can certainly catalyze your strategic decision-making process by automating your text analytics process.