Man and machine: how Palantir is winning in the analytics wars

We are messy, unpredictable and expensive when we process complex information. No surprise then that insurtech is dominated by technology that aims to squeeze out the human element: machine learning, predictive analytics, AI, chatbots, smart contracts. A handful of new entrants launch each week. What does it take to make money and survive in this increasingly crowded space?

With $2bn in funding and at least two of the largest global insurance companies as clients, one analytics company has discovered a strategy to maximise value and differentiate itself from its competitors. It doesn't just build great tools, it deploys its engineers and analysts on client sites as an essential part of its value proposition.

Backed by the CIA

California based Palantir has been secretive about its work and its clients. Some of its earliest engagements were with the NSA and the FBI. The CIA was an early investor through its In-Q-Tel venture fund, according to Forbes Magazine.  Six years ago it sold exclusively to government agencies. Today 75% of the revenue comes from private companies and CEO Alex Karp is starting to reveal more about his business, possibly as a precursor to an IPO in 2017.

If you are building a business offering analytics and data to the insurance market it may be worth taking a closer look at Palantir. Although well known in Silicon Valley, this is a company that not many others are aware of, and even fewer understand.

$100m+ contracts

Founded in 2004 by the current CEO, Alex Karp, Palantir was set up following the sale of PayPal to eBay. Peter Thiel was an early backer, and some of the first analytical concepts were inspired by approaches to fraud prevention developed at Paypal. Palantir likes to have large relationships. Karp claimed recently to have contracts worth in excess of $100m each with over 20 companies (over an undisclosed number of years). Axa and Zurich Insurance are clients. Buzzfeed recently gained access to internal memos and believes Palantir’s contract with BP is worth $1.2bn over 10 years. If this is all true, then the company is on track for significant growth. Revenues are still relatively modest: $420 million in 2015 according to Buzzfeed. A $20bn valuation is being contemplated by some.


Its two main products, Gotham and Metropolis, bring together disparate data sources and enable Palantir’s forward deployed engineers and its clients to rapidly find connections and patterns that aren’t obvious to the human eye. Applications include research on terrorism, cyber-crime, pandemics, natural disasters, health insurance and fraud.

Not good at the soft sell

Karp claimed his company was not very good at sales and marketing when he spoke to Fortune earlier this year. Palantir does appear to have positioned itself as a very exclusive club with demanding entrance requirements for potential clients.  Karp’s team want analytical challenges that must be compelling and Palantir expects to have long term relationships.  Client data must be made freely available to Palantir’s researchers with no constraints.

But whether Palantir is as phenomenally successful as a $20bn valuation would suggest, or ends up being just moderately successful isn’t really the point here. The company has benefited from experienced and well known backers, a record of success, and the mystique that comes from working with government intelligence agencies, fuelled by a heady confidence in its own ability. Yet much of what it is able to offer may not be that different from what is becoming possible with the tools and techniques that are coming out of many of the emerging businesses in the insurtech world. It’s how Palantir has deployed its tools that has made it successful, and this is where every technology and consulting company can learn.

Palantir's strategic asset, the "forward deployed engineer"

Look at presentations made by Palantir on YouTube from over five years ago and it’s clear they have maintained a consistent focus on how they add value. At the core of the company’s strategy is the deep-rooted integration between how they build technology and how their analysts support this at clients’ sites, or as Palantir describes them “forward deployed engineers”.

Saad Abdali, is one of Palantir's forward deployed engineers. In his presentation at GOVCON in Nov 2011, he explains why Palantir was so successful in its early years.

Abdali uses an analogy from chess. Ever since IBM’s Deep Blue computer beat world chess champion Garry Kasparov in 1997 it has been accepted that computers are better chess players than humans. Yet even the best computer can be beaten. Computers are excellent at handling large volumes of data and performing very large numbers of computations. They can win by sheer brute force. Humans play chess using insights and intuition, feeling out their opponents. We can take short cuts.  A human and computer working together have highly complementary skills. Relatively skilled human chess players working with a laptop computer running a chess playing programme have been able to regularly beat the most powerful chess computers playing without a human partner. The winning difference depends upon the efficiency of access to information. The easier it is for the human player to ask questions through a well-designed interface and tap into the computer for tactical computing power, the more powerful the combination of man and machine.

Get me to the answer on time

At Palantir, one of the core strategic goals has been to shorten the journey from raw data to business critical insight. This has always informed Palantir’s product design decisions. For many years Palantir’s forward deployed engineers were the primary users of Palantir technology, embedded deep within the client. Today clients want to able to use Palantir’s analytics to access their own data themselves, but the people on the ground remain a key part of Palantir’s value proposition.

Abdali described Palantir in 2011 as an “infrastructure for analysis”. Inter-operability with a client’s existing systems is critical. Palantir does not aim to replace all of the client’s existing databases. On the contrary, it's products are designed to work alongside them, leveraging the existing infrastructure.  Current descriptions of Gotham and Metropolis, remain consistent with how Abdali expressed the vision back in 2011.

Rapid growth can invoke maniacal passion. Some view Palantir’s forward deployed engineers as passionate evangelists, others complain of a touch of arrogance when the engineers come on site. Recent reports of high staff turnover and clients cancelling or scaling back contracts suggest that the company may be feeling the strain. The company is yet to make a profit but Karp still achieves a very high rating as CEO from his people on Glassdoor

When 1 + 1 = 3

Few other start ups or early growth companies in the insurtech world can expect to earn monthly fees in excess of $1m from individual clients, or receive the levels of funding of Palantir. The valuation seem a bit frothy and $100m contract values may be tricky to sustain. Some insurance companies will prefer to go it alone, building their own technology. AIG, for example, has over 100 people working in its data science teams, and builds its tools in-house. 

Yet there are lessons and opportunities for the fast growing breed of companies providing analytics to the insurance market, particularly those with products targeting the more complex commercial lines. Just as a skilled player armed with decent technology is the most likely to win at chess, so it is with complex business problems.

You want the best chance of ensuring full and rapid utilisation of your complex technology? Keep aiming to develop great products, but channel the messy but creative thinking of your people into an effective client support structure.

It's a tricky balance to get right. Historically, most successful organisations have focused only on one area, either consulting or product building. Those that have tried to do both have failed. Yet we are increasingly surrounded by ever more complex technology. The winners in the analytics wars will be those companies that are the most effective in deploying their messy, but smart, creative, intuitive and personable people alongside their technology.

This was published originally on LinkedIn.