First Steps Consulting

Will knowledge of a bias eliminate its impact?

Posted on April 12 2013 by brian

This is the second post in a series on decision errors, and in this installment we will take a look at “bias correction” and how to deal with it in a decision scenario.Ironically, once we are sensitive to our potential biases, we have a natural tendency to overcompensate for them. This is the decision error of bias correction.

I just recently finished going through a group simulation called “The Beer Game” as part of some training at MIT on System Dynamics. In this game a group of people operate independently at different stages of a supply chain that produces beer. As information is passed down the supply chain, participants try to adjust supply to fluctuating demand. In each iteration of the game, individuals go through a process of estimating what future supply will be, and the resulting chaos and fluctuation demonstrates our inability to operate in complex systems with limited knowledge. Lesson learned. If you want to try the Beer Game out yourself, here is an online version.

Here is the kicker. At one table was a group of individuals who had played the game before. The hypothesis was that this group understood that the game expected us to overreact. They would be more careful because they understood our natural bias to react to short term signals. They would do better we thought….

Actually, they were one of the worst performing groups. Their awareness of the “game” caused an over correction in their actions, which lead to an equally or more destructive set of decisions.

At this point we make throw our hands in the air and give up. It seems bias will get us either way. I highlight this as the second post in the series on bias (the first post was on the ambiguity effect) to quickly highlight that knowledge of bias alone is necessary but not sufficient to ensure good decisions.

To make good decisions, and to deal with bias correction, we need to understand what the biases are but also manage how much they affect our decision process and at what points in time they come into play. This is where “modeling” is useful.

When I mention the word modeling, I am sure the first thing that comes to mind are complex, data intensive models. These are important for big complex systems, particularly as “big data” becomes more accessible but modeling can also be appropriate for smaller decisions.

For this post I would like to highlight three modeling approaches along this axis of complexity.

System Dynamics specializes in the mapping of complex, dynamic behavior. Unlike traditional two dimensional models, System Dynamics allows us to model out multidimensional dynamic behavior in a visual format.

In addition, System Dynamics tools such as iThink or Vensim do an excellent job of allowing you to run simulations on your model. With System Dynamics you can understand and calibrate you actions to avoid an overreaction or bias correction.

Utilizing System Dynamics is not for the faint of heart or those with attention deficit disorder, but it is an extremely powerful tool for addressing particularly complex issues.

If you are looking for something a little more light weight, I might suggest using a hierarchical approach such as decision trees. Decision trees will allow you to map out possible futures and then model probable outcomes.

Unlike system dynamics they address issues in a linear fashion, but still allow the user to explore and experiment with the probable outcomes from different strategic actions. Here is a link to more information on decision trees.

http://www.mindtools.com/dectree.html

A Business Model Canvas is a recently popularized version of business planning that includes mapping out a variety of possible scenarios in a quick organized way. This rapid iteration and refinement of a plan can be useful in uncovering our biases and in avoiding any “over correction” of our biases. For more information on the Business Model Canvas, check out “Business Model Generation”, “Lean Canvas” or pick up Steve Blank's foundational read, “The Four Steps to the Epiphany”

So whether it is Bias or Bias correction, consider putting on a modelers hat. If you are interested in further help building your own model, please contact us.

To Think or To Act

Posted on March 24 2013 by brian

In this age of Lean Startups and Agile Consulting, it is often difficult to define when thinking or analytics is appropriate and when it is better to act. We just completed a draft paper exploring the question of when to “think more” or “act more” when faced with key decisions.

We expect this paper to evolve but hope that you find it interesting and thought provoking and welcome any feedback or thoughts you had on the paper or the concept. You can download the article for free on our resource page

To Think or To Act

Dynamics, Disruption and the Running Shoe

Posted on December 9 2011 by brian

This post is a self indulgent exploration of how the new minimalist running shoes represent an excellent example of a classic market disruption to a mature market that exhibits some interesting system dynamics.

The objective of this post is to showcase system dynamics at play in the context of a disruptive market. It is my hope that by the end of this post you will be interested in how system dynamics could provide insights into potential market disruptions.

The chart below is a simple System Dynamics Model developed for this post. We will explain the model a bit more later in the posting. (Sorry iOS crowd, simulation is a flash file, still looking for good HTML5 options for dashboards)

Could the Unknown be safer than the Known?

Posted on November 8 2011 by brian

This is the first installment in a series of posts on Decision Errors, or biases that tend to cloud our decision making process. The objective of this series is to highlight challenges in making clear decisions, and then point to tools or processes to help us overcome these biases.

The first bias we will address is the Ambiguity Effect. Stated formally in Wikipedia…

The ambiguity effect is a cognitive bias where decision making is affected by a lack of information, or “ambiguity”. The effect implies that people tend to select options for which the probability of a favorable outcome is known, over an option for which the probability of a favorable outcome is unknown. The effect was first described by Daniel Ellsberg in 1961.

In short, we tend to prefer the known to the unknown. This bias can lead to very unfavorable results. Let me give a simple example.

One major unknown in life is how long we are going to live. While we feel our probability of being around in the next year is high, uncertainty around whether we will be around 30 years from now biases us to live for the moment.

From a purely rational point of view, people would be wise to enjoy themselves to a moderate degree now, and to also prepare for the future. Regardless, in a recent EBRI research survey, “about 56% of workers report having less than $25,000 in savings and investments (not including the value of their primary home and benefit plans) and 29% of workers have less than $1,000 saved.”

“People are increasingly recognizing the level of savings realistically needed for a comfortable retirement,” said Jack VanDerhei, EBRI research director and co-author of the report. “We know from previous surveys that far too many people had false confidence in the past.”

So how do we overcome this bias?

Creative Tension Trumps Performance Management

Posted on October 19 2011 by brian

If you have been an entrepreneur and then enter an organizational work environment then one of the first new things you may encounter is “performance management”. I have recently been intrigued by the lack of performance management in startups, as well as the limited effectiveness of performance management in large organizations.

In considering the differences between the two, It occurred to me that the drivers in each case are fundamentally different. What inspires the staff of a startup is different than what inspires staff in a large organization driven by a performance management program.

The root difference is that startups are driven by the creative process, or simply put, what they are trying to create. In organizations using performance management, people are being driven by how well their existing stuff performs.

I hypothesize that people are driven more by the process of creating than refining.

Of course, refining is part of the success for any new offering, but I am wondering if there is a way to “creatively” refine.

To ground this idea, let’s visit the eating habits of my two year old. (A lot of my thinking these days revolves around my two year old). The act of eating for her now is largely a creative process. She is trying out different foods with different flavors, textures, colors, etc. If I put something interesting in front of her, it takes little prodding to get her to try it. Trying it is a creative experience that includes some eating, some spitting out, some rearranging, some mixing, throwing, etc. Entrepreneurial companies are like two year olds experiencing new food.

Now comes the refinement challenge.

When decisions matter, how do you decide?

Posted on September 10 2011 by brian

I have been intrigued recently by the question of when to use primarily intuition in decision making versus when to use primarily analysis. Clearly at the extremes there are obvious incentives to use one or the other. A decision to swerve when a child runs onto the road, or the decision to take a defensive posture when attacked are clearly instinctive. A decision about whether to undergo a major surgery is typically largely analytical.

Where it gets tricky is in the “middle”, and specifically what are the parameters in deciding how to decide? After reading through a variety of opinions on the subject. I surmised that some of the key influencers in decision approach comes down to “size of decision”, “amount of time to make a decision” and “depth of experience” in the topic of the decision. Using these three parameters, I came up with the following matrix of preferred decision types.

Part of what makes this topic interesting to me as an analyst is helping individuals or businesses make the right choice about when to use different decision types. Using the chart above, I believe there are several biases that lead to incorrect decision models.

Can Do Bias

First, for there is the miscalculation of the level of internal experience. Entrepreneurs fall into this trap because they develop a varied skill set in order to survive, and have a habit of acquiring knowledge that they lack. Ironically, while this is partly what helps an entrepreneur thrive early on, it also can be a significant barrier to growth. Entrepreneurs are trained to act instinctively, and have a difficult time switching to analysis. A typical first step into analytics for entrepreneurs is the establishment of decision support systems. These systems do not usually challenge the “ability” of the entrepreneur, but help provide targeted information quickly to support decision making.

Risk Aversion Bias

Second, there is the miscalculation of the significance of a decision. Corporations with long histories of custom analytics by consultants or in-house research departments typically fall into this trap. When the go-to decision becomes custom analytics, then the risk of over analysis goes way up. Suddenly, the company is spending $100k to research where they should have the holiday party, and what the appetizers should be. Companies with a culture of custom analytics need to complete a quick scenario analysis and determine if the marginal benefit of making the right decision justify the cost of identifying the right decision.

Procrastinator’s Bias

Finally, there is what I call the procrastinator’s bias. Some people simply wait until the last minute to address key decisions, and then are forced to rely on intuition. The earlier key decisions can be identified, the more likely they can be properly analyzed. If you are always relying on intuition to make decisions, ask yourself if that is because you always wait until the last minute to decide.

What other biases do you think impact how we approach decisions?

What we can learn from Hurricane Irene about trends

Posted on August 28 2011 by brian

As I wait for Hurricane Irene to hit Boston, I am reflecting on how Hurricane Irene is similar to a new trend or technology that makes landfall on a population of people.

It usually gathers momentum and strength out of the public eye. Only forecasters can attempt to project it size and trajection.
It has a vortex or core, and tend to sustain itself by drawing in anything within it’s reach to the core.
Even though people know it is coming, many people will be unprepared
If you are in it’s path, it will be all encompassing. If you are not in it’s path, it will be a side note.
While it’s impact on a population is somewhat predictable, it’s impact on an individual is not.
While one side of the storm tends to get more rain, the other side gets more wind..in other words, it is experienced relative to your perspective on it.
It passes…and is followed by another

Did I miss anything?

The most accurate and reliable financial projections have a secret ingredient

Posted on August 10 2011 by brian

Having worked with startups and corporate innovations for many years, and developed the financial projections for those efforts, I have discovered that the architecture used to developed top line revenue or unit growth can have a big impact on the reliability of said projections. Typically, if a projection is unrealistic it is because the projected growth is unrealistic or inconsistent.

For the architect of the financial projections, growth is typically estimated by calculating a unit growth percentage which is either applied consistently across all units of time (ie months) or in some chunked time block (ie quarters) or each month is given an individual input (ie 60 inputs for a 5 year projection)

Once this architecture is established, the architect will then try to massage the numbers to achieve a realistic growth curve. If the architecture is based on growth over chunked blocks, the growth curve is "chunky", which is not realistic. If the architecture is based on one consistent growth percentage, then the growth curve is linear, which is also not realistic. If the architecture is base on monthly inputs then the curve needs to be manually constructed using a large number of inputs, and what typically emerges is a curve similar to this.

The solution to creating consistent and realistic growth projections is to leverage logistic growth curves which have been developed to plot realistic growth over time. There are a number of functions that are used based on the nature of growth, but a common on is the Gompertz function. The Gompertz function assumes a point of maturity, a point of slow early growth and a point of rapid growth, which describes many startups or corporate ventures. With a Gompertz function, the rate of growth and the timing of growth can all be manipulated by tweaking several function inputs rather than dozens of inputs.

A typical Gompertz growth curve might look like the following;

To validate the Growth curve, it is best practice to calculate other benchmark metrics to insure the scale and pace of growth is realistic. For example, if you plot final membership using the Gompertz curve, you will also want to calculate the number of new members per month that this represents.

If you want to play around with what is possible with growth functions, check out this little flash tool for playing with the Gompertz and Lognormal function.

If you have other favorite growth functions you use for projections I would love to hear about them!

If you are interested in getting help with developing robust financial projections or models, contact us.

Are you measuring what matters?

Posted on July 1 2011 by brian

In a recent article by Healy Jones titled “VC Valuation Valley of Death”, Healy argues compellingly that startups go through a period of hyped valuation, followed by a period of depressed valuation and then an eventual return to the original or higher valuation if the startup survives.

One of the things I love about Healy’s article is that it considers one metric (valuation) over a period of time and shows the fluctuations in that metric over time. At different points in time, startups need to focus on different measures. Healy argues that while in the Valley, startups need to focus on cash.

Once out of the Valley, Startups may switch there focus back to valuation as they seek to access the further rounds of funding they need to scale. I began wonder what other metrics were critical to startups during their trip through the Valley.

The best paradigm I have come across to answer this question was put forth by Steve Blank in an article titled “No Accounting for Startups”. Steve argues that Startups start off searching for a business model, and then move into executing on the business model. At the execution stage, the standard financial metrics are relevant because the business is in the money. At the search phase, the metrics that matter are the underlying assumptions of the business model. The question is not are you making money, but is the business model sound, so if it was a scaled it could make money.

So in summary, if you are a startup that is in the valley what matters is cash flow and the measurement and management of the underlying business model assumptions. Don’t let other metrics that proceed or follow the valley distract you from what matters when you are in the Valley.

If you are interested in reading Steve or Healy’s articles, they are both referenced below.

Steve Blank
Healy Jones

If you are interested in getting some help in thinking through and measuring the right metrics for your startup, drop us a note.

Segment like a startup

Posted on June 27 2011 by brian

An interesting article by Alan Armstrong titled “Positioning Beyond the Product – Think Relationship” highlights the fact that many products or services start off focused on customer needs and deteriorate into a bundle of features. Essentially the product becomes the focus rather than the consumer.

Fundamentally positioning is about segmenting the market based on demonstrated or latent needs and then targeting the most promising segment with the appropriate feature set. Product features are typically the result of positioning rather than a driver in positioning.

Once a product or service is successfully launched, companies find it hard to operate outside of an established technology or service set. Inertia takes over….new startups emerge and replace existing offerings.

To stay competitive, companies need to repeatedly segment their market based on customer needs, just as a startup would, and make adjustments to positioning. The segmentation completed at inception has a shelf life. The consumer needs and drivers of yesterday will not be the same drivers of today.

Here are key indicators that it may be time to consider a fresh look at you market positioning.

1. You see a variety of startups entering your base market. If other entrepreneurs are seeing opportunity in your market, the landscape is likely shifting

2. Your latest product update generates no meaningful bump in sales or user satisfaction. Just because you can improve on a feature set doesn’t mean your customers value it. If the existing need is satisfied with the current feature set, then adding more features generates no value and is likely distracting you from new emerging needs.

3. Significant environmental changes to the demographic or geographic market you are serving. For example, if you are in the luxury goods market, a prolonged recession should trigger a fresh look at your positioning in the market. While luxury consumers may have wanted their wealth to be on display in prosperous times, they may prefer more modesty during down economies. The environment can impact preference. A review of the rise and fall of the Hummer might also be appropriate here.

4. Your youngest or newest employees are dissatisfied. Disruptions are often seen last by upper management. If your newest and youngest employees are dissatisfied, it may be that they know the direction of the market but have no power to influence change.

Please comment and add others you think I may have missed!

Recent Posts