Data analytics - vitamins to boost enterprise heath or just pain killers?

I recently met up with Thoughtspot CEO Sudheesh Nair, shortly before the firm’s recent Beyond conference in Dallas. The meeting was a follow-up to the announcement that the company had taken on a funding round that gave it an additional £200 million, topping up available funds for future investments to a total of $350 million. The discussion agenda, at least in part, was ‘re-defining enterprise analytics’, a somewhat more-than-moderate challenge that seemed worthy of, well, some analysis.

The solution will almost certainly involve some acquisitions along the way, which is why the extra money is useful. One underlying drive here, however, will not just be to just get Thoughtspot’s corporate hands on specific technologies that fit with the development plan. More important to Nair is getting hold of new talent:

I want to make sure that the best entrepreneurial talents in our space have an opportunity to come here.  Some of them will come as employees that we have hired, but we have to be open for talent acquisition.

As yet, he has little idea which companies might become targets or which direction some of the future will follow. Much will depend, he said, on what the customers want. But by the same token there is an undercurrent in his thinking which suggests a degree of education may be required to get users thinking along the right lines. He sees analytics as analogous to vitamins for the enterprise, while many enterprises tend to see it as a pain killer:

Today you cannot sell analytics as vitamins,  you have to turn that into pain killers because there are real pains that can be solved with analytics. To change that you have to fundamentally change the decision-making pipeline inside the enterprise business community.

Do you have a relationship with your data?

To do this means having a much deeper relationship with the data – Nair calls it being able to talk to it, ask it detailed, interconnected questions. Basic numbers can tell you how well a product is doing,  but a much deeper understanding can be gained from knowing a richer context in which those numbers sit: how long do they use it, why did they buy it, what they like or dislike about it, what is their age group and/or gender, do they use it on mobiles, what would they like added/dropped or modified? There can be a long list of possible questions which, with the coming of IoT and mobile systems and applications, is only ever going to get longer.

In Nair’s view, the problem with the decision-making pipeline now lies with the content speak that data language, or SQL. So while there  is ever-more data, and more processing power, that pipeline has become so small that is it now a major pain point for business. In his view, it is also a pinch-point that can now be reamed out and opened up.

There is an argument here that part of the problem is that analytics is essentially high-grade mathematics. Somebody has to decide what question gets asked, where the data is, what the algorithm or formula is in which it has to be analysed. End users then get an answer, because there is no place for any dynamically shifting contextual input, look at the answer and wonder what it means.

It is certainly Nair’s position that, by coming at it through the search argument, it is possible perpetually to refine the search arguments, pulling the subject backwards, forwards and sideways at almost any point, ending up with the ability to come up with answers in a much more direct way. That, he suggests, is the way to solve the business problems that are worth solving, and doing it in away that maps onto the way humans learn and form opinions:

I believe as a species the way we understand things hasn’t changed. So what we’re trying to do here, if we pull this off right, is that our brain’s ability to have divergent thinking is now completely complimented by the datasets that we run with. The API of our eyes, ears and mouths, our sensory perceptions are under sensory overload because TVs, tweets and the rest are constantly flowing, and that has become the limitation.

That is the reason why Elon Musk now has a company called Neuralink, where he’s trying to bypass all of this and go directly to the brain. I think the better approach is to provide access to data in a way that is conducive to how we understand it. Most organisations that claim to be data-driven are usually making decisions and then finding data that justifies the decision. The reason why ThoughtSpot is so relevant here is because we are proposing a radically different approach to the market. We are saying that data in its raw form should never go into a human.

The company has a simple philosophy for managing this, known by the acronym, STAR. First of all it has to be Secure. Next it has to be Transparent, what Nair calls `a glass box, not a black box’. Users need to know what the analytics are doing. Third up is the need for it to be Accurate, otherwise users will rapidly lose trust in it as a tool, while the last part is that the results have to be Relevant:

We want the customers to use that to measure every other AI vendor in the space, to see if they check all of the STAR boxes.

What’s here for now

The main announcement at the Dallas conference demonstrated some specific movements forward in the company’s aspirations here. This was a major move towards creating analytics tools that are increasing autonomous.

This is part of the latest update to its eponymous product, ThoughtSpot Version 6, which includes the addition of ThoughtSpot Monitor, which is being made available in beta in Version 6. This allows users to instruct ThoughtSpot to monitor specific datasets, charts, or KPIs and have it continuously analyse the underlying data. It can then alert users to changes and new trends.

It also adds a new mobile application aimed at allowing users to search on their own terms, wherever they can get a signal, and now sets out to help users find the right questions with Answer Explorer, which serves up suggestions on a pinboard of terms frequently used with similar data sets throughout a user’s organisation. A suggested answer can be followed up by simply clicking on it

Another important update has been made to the Embrace service, which runs search and AI-driven analytical workloads directly in existing databases. This now covers the increasingly common clouds data warehouse, Snowflake. In addition, it has added support for Google Cloud Storage with plans to extend working with BigQuery from Google Cloud Platform and Amazon Redshift from AWS, with other data services in the pipeline.

The company has also extended user options in building their own analytics solutions with the launch of Atlas Marketplace, a directory of solutions and technologies largely built by Thoughtspot partner companies. With these users can build their own turnkey analytics solutions from pre-built and tested components and subsystems, with a view to getting to valuable results and business insights as quickly as possible.

My take

Perhaps the most interesting thought to take away from this is the observation by Nair that data in its raw form should never go into a human. It does sometime seem that the data analytics business is determined to squeeze out ever more granular drops of raw data from which business value might be extracted, and then come up with ever-more devilish tools to be bought that may help to persuade some of the value out into the open. It does increasingly beg the question of whether there is a different, hopefully better, way.