Data models provide insights, but humans make decisions, and those decisions are not always rational.
Behavioral economics research explores the nature of human decision-making, biases –like the “anchoring effect”, a common tendency to rely too heavily on the first piece of information when making decisions — and how seemingly unrelated things influence our decisions.
We’re here to report that the anchoring effect is alive and well in almost every transportation procurement decision. But why is this bias so important to shippers, their procurement efforts and transportation strategies and costs? How biased are decisions and what’s the best way how to overcome biases?
As a near-real-time case study, let’s look at how the procurement director at a large Asian consumer goods company chose to make a transportation procurement decision:
- While the director initially limited her expenditure approval to no more than a 5 percent increase in 3PL pricing (just before she started re-negotiation with the company’s existing 3PL), she increased her upper limit to 10 percent a week later, despite the fact that volumes were going up, market indices hadn’t changed and there were no new operations.
- While the director didn’t want to approve more than 5 percent, the company’s 3PL soon presented a whopping 50 percent increase. Subsequently “anchored” to the 3PLs higher price, she agreed to approve up to 10 percent increase. (Here’s where a little self-knowledge would have paid off for her company: If she had been cognizant of “anchoring”, she wouldn’t have increased her expenditure approval to 10 percent.)
Feeling like a winner or a loser? Risk-averse or ready to venture boldly?
Like every other human being, supply chain decision makers can be irrational and biased. For example, people tend to favor “90 percent on-time delivery” service levels while disliking, by a huge margin, “one fail out of every 10 orders”.
Over the last two decades, data use has increased in the supply chain as planning systems pioneered the effort to account for forecast biases. Dashboards, KPIs and S&OP have helped a lot in day-to-day operational decisions. However, strategic decision making still involves considerable biases:
- Data analytics creates a number of “network location scenarios” for supply chain network design. Inevitably decision makers will favor or dislike these scenarios, even before going through the numbers (perhaps because “folks in that plant are my friends”, “I had a bad vacation experience in that city” or simply because “it sounds more risky”). Nobody talks about these reasons, since they are unknown unknowns.
- Cognitive biases influence thinking and affect whether the decision maker supports or opposes the scenarios (e.g. “The model has not considered low attrition rate in that DC”, “I will be able to get good real estate deal here,” etc.). The decision makers’ comments then influence other team members, further complicating the decision process.
Given our penchant for fact-based decision making, we ask ourselves: Can we remove all biases and make sure data dominates? Probably not, as long as a human signature is required on the dotted line. But, we can at least identify biases using three classic negotiation and measurement techniques:
- Anonymizing data works great for most comparisons. Using generic names (like Supplier A, Supplier B, etc.) during an RFP response comparison is a simple way to eliminate cognitive bias, but this approach must be planned before sending out the RFP so all incoming data and key details are collected to rank branding, capabilities, service history and any other qualitative items
- Based on the prior step, we advise having all members in your project team write down their supplier ranking anonymously, in parallel. The average of all the rankings will offer the best solution and is a great tool during project planning, as well. Asking team members to write down their estimates for project length and averaging those estimates will give you the best timeline. This procedure is really to resolve the “anchoring effect” wherever it may pop up.
- Simulation brings all stakeholders together, to run through various scenarios and evaluate all stakeholders’ feedback. Start the discussion with which stakeholder stands to lose–or perceives loss–or other negative situations to address the “endowment effect” (wherein an evaluating team member “owns an item” and sees “losing it” as a big deal compared to gaining it when one doesn’t own it). For example, with a core network design team, see how “owners” react when every DC or plant is closed. The psychology of ownership, even in a simulated environment, is powerful; you will see some team members start to give every reason why it’s a bad idea to close down their DC. Detailed documentation and quantitative analysis of all the qualitative points raised is great input for data modelling and project planning. This step also minimizes additional factors creeping into later scenario discussions.
These techniques will nudge supply chain leaders to recognize not only team members’ irrationality, but also their own. Recognizing, by definition, takes out the first unknown.
Sivaram Murthy, senior consultant in Chainalytics’ Integrated Demand & Supply Planning (IDSP) competency, provides supply chain management, product management and technology implementation expertise in sectors including retail, manufacturing, financial services, education and technology. Chainalytics clients rely on him for cost savings and improved processes.
Vikas Argod is manager of the Supply Chain Operations competency at Chainalytics. Vikas specializes in warehouse operations, transformation program management and service delivery processes in project-based business environments.
[…] week I read the Logistics Viewpoints guest post by Chainalytics discussing the pitfalls of decision-marking biases. Although the Chainalytics article focuses on […]