Due diligence meets big data

My firm, Two Six Capital, was featured in Mergers & Acquisitions.

Due diligence meets big data: Two Six Capital, Oak Hill, Clarion Capital

Ian Picache:
“It is like it was in 1990 for public markets…. The use of big data techniques is nascent. It is only a matter of time before the private industry is disrupted. Just as the public markets saw the rise of massive quant hedge funds such as DE Shaw, Renaissance Technologies, and AQR, the same will happen in private equity. In the future, we expect big data to be heavily used in industry….

We see big data as a board room agenda. Two Six has developed a technology, process, and people-based playbook to drive operational improvements. The playbook has a broad range of applications across functions including marketing, sales, support, budgeting, R&D and operations. Just like ‘quants’ in the public markets, we can run large-scale iterative tests with management teams to make resource allocation decisions. Two Six can monitor and optimize portfolio companies using actionable dashboards that can be used to make board level strategic choices or tactical campaign-level management decisions.”

Sajjad Jaffer:
“Statistical, artificial intelligence, and machine learning techniques are becoming available and take advantage of the technology infrastructure and availability of data.”


My dissertation recently came out of embargo. You can find it here:
Essays on the Social Consumer: Peer Influence in the Adoption and Engagement of Digital Goods

In this dissertation, I study how consumers influence each other in the adoption and engagement of digital goods.

In the first essay, I study peer influence in mobile game adoption. Although peer effects are expected to influence consumer decisions, they are difficult to identify in observational studies due selection bias: Friends share common characteristics and behave in similar ways even without peer effects. I use a novel approach to estimate unobserved characteristics which endogenously drive tie formation and use the estimates to control for selection, without need for instruments. This is the first paper to use latent space to reduce bias in peer influence estimates. I find that peers account for 27% of mobile game adoptions, and that ignoring latent homophily would bias the estimates by 40%, in line with previous studies. In some samples, ignoring latent homophily can result in overestimation of social effects by over 100%.

In the second essay, I examine the effect of zero rating on consumer behavior in a social net- work. I use Facebook data on millions of users to quantify direct, peer, and long-term effects of zero rating, a campaign where consumers can access digital media over mobile networks for free, on social network activities. I find that zero rating does not have the same effect on all so- cial network activities. While the direct impact of zero rating is positive on all activities, users with more friends on zero rating create less, consume more, and give more feedback on content. In addition, zero rating does not have a uniform effect across consumers. Some consumers benefit more from zero rating than others, and I show that network characteristics can help identify those consumers whose network benefits the most from zero rating.

Ethics Post

Taking a cue about being open, this is my ethics page, just so you know where I come from.

I work at Two Six Capital, applying data science to private equity. My partner works at Omada Health. I will not comment on any product or policy by these companies.

I own stocks in a bunch of different companies, but most of my money is in index funds.

I have a small role helping with sales and marketing for Thai Orchids Lab, my father’s company in Thailand, which deals with agriculture, horticulture and forestry.

Typical disclaimer: The opinions expressed on this site are mine and do not necessarily represent those of my employer. You won’t find any confidential company information here, and while you’re welcome to get in touch with me, I’m afraid I can’t put you in contact with my employer.

Bricks and Clicks

First, there was a leak that Amazon might be planning to open 300+ physical stores. Then, the leak was retracted. Then, more news that Amazon might be planning for more than bookstores. Let’s just say that things are exciting!

But even before this phenomenon hit radio waves and newspaper headlines, four prescient researchers picked up the idea and investigated the effect of adding bricks to clicks, that is, adding physical stores to previously-purely online e-commerce sites.

Without digging deeper, it is not at all clear which way the research would go. Would physical stores hurt other channels (online + catalog)? Are they complementary?

It turns out in the cases that the researchers examined, physical stores had a negative short term impact on catalog sales (no short-term impact on internet sales), and positive complementary effects on both online and catalog sales in the long run.

So, is Amazon’s decision wise? Research says probably. We will see what happens when a store actually opens.

Here is the academic version of the paper.

Disclaimer: The researchers: Jill Avery, Mary Caravella, John Deighton, and Tom Steenburgh all have connections to HBS, and one of them (Tom) was my dissertation committee member.

Exciting paper on experiments in ads

This paper was recently accepted into Marketing Science, a top tier marketing journal:
When Less is More: Data and Power in Advertising Experiments.

There are three things that excite me about this paper:
1. This paper conducts research into experiments in advertising. Experiments in the ads industry are very hard for many reasons: identity, cross-platform, conversions capture, just to name a few! But the ads industry also provide amazing opportunities due to scale. Instead of simulations, lab experiments, or even the scale of panel scanner data or CRM type databases, we are talking about millions and millions of observations every day.

2. This paper pushes the frontier for practice by taking a seemingly simple idea (removing conversions prior to impressions reduces variance) and quantifying its impact which exceeds complex analysis involving big data. In essence, product wins data! Or, from another perspective, data quality wins data quantity!

3. Two of three co-authors are researchers from companies, and Garrett, at Rochester, was an intern at tech companies including Facebook! Industry researchers have a long history in management research; back all the way to Taylor’s scientific approach to management, or McKinsey and BCG’s contribution to various models. Indeed, science should be embedded in practice; that is the type of science which really excites me.

Election year impact on (non-political) advertising

With political campaigns reaching record spending in 2016, the flood of money can have spillover effects to marketers of non-politics-related companies.

At first glance, the influx of advertising presents a headache for many marketers. For TV, competition for inventory depends on geography and time to primary and general elections. As campaigns spend more on digital media, digital inventory could also be impacted. There are some basic advice for marketers here and here, but are such lists comprehensive?

Fortunately, there are a few technological advances which reduce the burden for marketers. First, if your marketing campaign has broad targeting, an optimization platform like that a Facebook will find efficient CPC/CPM automatically for you. So, if certain regions are more expensive (e.g., FL or OH), the optimization engine will take that into consideration and look to show your ad to people with lower CPC/CPM thresholds. Second, if your campaign overlaps with periods of high political spending, but not solely contained within that period, ad systems with pacing should help your campaign speed up / slow down spending relative to CPC/CPM volatility. These algorithms can help smooth out your spending for optimal impact, despite the influx of political money.

In conclusion, as presidential election campaigns spend more money on TV and digital outlets, marketers should consult with their measurement experts at agencies / ad platforms to understand whether there are systems in place to adjust and optimize for changes in pricing and inventory.

Best of 2015

HBS has a list of top stories for 2015. Here are some that piqued my interest:

Management / Marketing

Need to Solve a Problem? Take a Break From Collaborating
“The most-clustered groups gathered 5 percent more information than the least-clustered groups, … However, clustering also seemed to inhibit the breadth and number of answers that the players proposed…. ‘We realized that the network structure seemed to have opposite effects for searching for information and searching for solutions'”

Almost everything I do is in teams, in networks. Knowing that certain team structures are better than others (at specific tasks), this implies I should be more critical when engaging people within/across the company.

How Our Brain Determines if the Product is Worth the Price
“When the product came first, the decision question seemed to be one of ‘Do I like it?’ and when the price came first, the question seemed to be ‘Is it worth it?'”

When companies advertise, or when price and product are shown sequentially, the sequence can be very important to how consumers perceive the product. For example, should an advertiser focus on product features or specify price in the ad?

Tech / San Francisco:

HBS Cases: The Battle for San Francisco
San Francisco, the Bay Area, and other cities are getting more polarized as technology companies grow.
“At the root of the issue are a series of questions: What is the problem? Who is responsible for it? And who is responsible for fixing it? Coming up with answers to these questions is not easy.”

I care about SF because I grew up here (college, early work, now). One problem I have with this is that the case was written based on interviews with ex-HBS-MBA’s: “They invited 22 Bay Area-based Harvard MBAs who had graduated as recently as a year ago and as long as 30 years ago to a series of roundtables to discuss their perspectives on inequality and the tension between the community and technology firms and their employees. Far from the unfeeling interlopers depicted by the Google bus protesters, they found a group that cared deeply about preserving the culture of the city, and that wrestled with how they might understand and change the underlying conflicts.” I don’t quite understand how a single set of perspectives can come up with answers.

How to Predict if a New Business Idea is Any Good
“Expert interest was highly predictive of success in sectors that were R&D intensive, such as energy, hardware, medical devices, and pharmaceuticals. However, in non-R&D-intensive sectors, such as mobile apps and software, the ability to predict success was no better than random.”

When I was interviewing with startups, my main concern was about failure. One of the advice that resonated with me was by Andy Rachleff: “You get more credit than you deserve for being part of a successful company, and less credit than you deserve for being part of an unsuccessful company. Success will help propel your career. At a fast-growing company, chances are good you’ll have a higher position two years after you join. At a slow-growth company, no matter how good a job you do, you won’t have the same opportunities to advance.”

Tech, like its economic-boom predecessor, has a gender / race / diversity issue. These two articles deal with gender.

Kids Benefit From Having a Working Mom
“Women whose moms worked outside the home are more likely to have jobs themselves, are more likely to hold supervisory responsibility at those jobs, and earn higher wages than women whose mothers stayed home full time”

Men Want Powerful Jobs More Than Women Do
“While women and men believe they are equally able to attain high-level leadership positions, men want that power more than women do”

Both studies are based on surveys, so they are more descriptive about the current situation, not so much prescriptive. Companies need to look both inward and outward to understand gender issues at the workplace.

For those thinking about their careers, here is a fun bit of research:

‘Humblebragging’ is a Bad Strategy, Especially in a Job Interview
“The takeaway: By public perception, complainers are better than braggers. And humblebraggers are the worst.”