BACK ISSUE PREVIEW
September | October 2018
Free Article Preview from this issue of Roast magazine
By Chris Hallien
The amount of information available to coffee professionals continues to grow at an accelerating rate. From farm-level activities to coffee consumption habits and all stages in between, data is measured, recorded, reported, analyzed and readily shared across the industry. The move from paper to digital formats has helped significantly in facilitating this. The digitization of archived material and old publications now delivers a steady stream of information that was once hard to find. Innovation and improved technology has delivered literally thousands of different sensors, and they are already in the many devices we use in our daily lives.
At times, it seems like all this information is outpacing our ability to benefit from it. Access to data provides tremendous benefits to a business as a key enabler for making informed decisions. On the other hand, data can create new challenges if it requires too much investment to capture and manage, if analysis is erroneous, or if the data itself is not sound. Identifying information relevant to a business and learning how to analyze it often comes with experience. It also requires constant reconciliation with dynamic business environments, evolving goals and changing products, as well as verification of information that originates from unchecked sources such as social media.
Over the course of 20 years in the coffee industry, I’ve been employed at various capacities across diverse companies. As a result, I’ve worked with products ranging from high-quality green coffee sourced from farmers performing experimentation at the agricultural level to low-grade greens requiring extensive cleaning and sorting on the receiving end; from manually operated roasters to fully automated roasting plants; from the full range of decaffeination methods and soluble processes to seemingly every grinding and packaging technology in existence. On multiple occasions, I believed I had encountered all possible combinations of variables within an equation, only to be introduced to an unexpected one. The commonality in all cases was the need to determine what data was important to capture and how to analyze it in order to make the best decisions about the product or process.
Hugh Morretta (front) of La Colombe Coffee Roasters and Chris Cross of Cafe Grumpy at the U.S. Roaster Championships in Seattle, 2018. | Photo by Taylor Wallace, courtesy of Cropster
Data Collection and Analysis Basics
Data integrity is an essential aspect of the data collection and analysis process. For coffee professionals, data integrity includes sample information, sampling protocols, sample preparation and analysis as well as verification of methods. Green coffee information is generally provided from the supplier and pertains to lot identification and grade—information such as International Coffee Organization (ICO) marks, certificates of origin and organic certification. Various other details often are available from origins requiring green coffee to be processed through a government agency as criteria for export. Information related to variety, altitude, processing type and other attributes have become increasingly commonplace—all of which reasonably can be assumed to be accurate.
Several organizations such as ICO, the Specialty Coffee Association (SCA), the National Coffee Association USA (NCA) and the Intercontinental Exchange (ICE) provide guidance in one or more areas around sampling, preparation and evaluation. Other organizations such as AOAC International and ISO serve as sources for analytical testing methods. Often, coffee companies will create their own standard operating procedures (SOPs) by combining information from various sources to fit their individual needs.
Data collection can help a roaster understand the variables available for manipulation during roasting and the influence each variable has on the roasted coffee. Pictured: Coava Coffee, Portland, Oregon. | Photo by Mark Shimahara
SOPs are extremely beneficial to data integrity as they outline the flow of a procedure, where to record output, and the unit of measure to use. Standardization of the units of measure is essential. Within any company, there are often ambiguities: assumptions based on context, for example, such as green coffee weight versus roasted coffee weight; lack of reference scales, such as Celsius versus Fahrenheit or Agtron Gourmet versus Agtron Commercial (or the absence of a scale altogether); or using subjective descriptions such as “medium grind” to refer to a grind size of 750 microns. Agreement on units of measure reduces ambiguities. The use of digital platforms for data management can assist by being formatted to a single standard that defines how to input data.
The majority of data a company collects is done so as part of the normal course of business. New data sources are introduced when a new process or device is implemented or when existing ones are modified. It’s important to note that changes made to the data collection process can influence the results; for this reason, many laboratories maintain calibration, service and replacement records.
Analyzing data collected during and after roasting can help a roaster determine the most appropriate profile for each green coffee. | Photo by Mark Shimahara
Project-based or initiative-based data collection can introduce new data points as well. These types of activities are usually short-term or periodic data collection efforts that serve to produce a targeted outcome. They can take the form of data-logging transportation and storage conditions of green coffee, or the rate of change in roasted coffee weight due to off-gassing. Parameters that begin as project-based can lead to ongoing data collection requirements if deemed critical to the quality of the coffee.
Data analysis is critical to a company’s performance. Data from all functions of a business can and should be collected for analysis and reporting. Ideally, this would occur in a common electronic platform to optimize reporting capabilities. Time-series analysis provides information on directional movement. That is, data collected on one metric over multiple time intervals can be analyzed to determine if a trend exists. This could be as simple as a quality control manager documenting the total moisture content or water activity of a green coffee lot in a warehouse, or the roasting department recording the average cup score of that same coffee, over a nine-month period. Correlation involves identifying the relationship(s) between multiple metrics. Correlations can provide significant insight for making business decisions. However, it’s important to remember that correlation does not necessarily mean causation.
Data collected on green coffee at different points in the coffee’s life cycle can be extrapolated and applied to other green coffee lots of a similar type. | Photo by Connie Blumhardt
Green Coffee Quality Data
A comprehensive green coffee quality program includes analysis and evaluation of all coffees purchased or considered for purchase; this includes green grading. For the purposes of this discussion, green grading pertains to the identification and quantification of physical defects in a representative green coffee sample according to an agreed-upon schedule of defects. Green grading also can include screen size and roast and cup quality. In addition to confirming the quality of a green coffee and its relative value, these activities serve other business purposes, including the following:
• Supplier Alignment: Alignment (sometimes referred to as calibration) between the buyer and supplier is a crucial step in establishing and maintaining supplier relationships, and serves as the foundation upon which other activities are anchored. Alignment in terms of green coffee grading and quality expectations, specifically flavor profile of green coffee, benefits both parties.
“We send team cupping results as a critical part of our ongoing supplier calibration process,” says Andi Trindle Mersch, director of coffee for California-based Philz Coffee. “We share detailed results for every sample we cup, whether [pre-shipment], arrival or offer/type. Through timely and regular data sharing, we ensure our suppliers understand exactly what we need and where our range of acceptability/approvability lies. We often secure their internal notes back as part of the conversation and calibration.”
• Producer Feedback: Observations about physical green coffee and/or cup quality can assist in farm-level practices related to harvesting and post-harvest processing. Some physical defects can be linked to specific processing steps. Feedback can serve as confirmation of improvements made at the farm level or work diagnostically to identify areas in need of improvement.
Philz encourages its import partners to send their cupping results upstream in the supply chain, says Trindle Mersch, to help producer and mill partners better understand what the company values and how the coffees are performing at the consuming end.
• Quality Trending: Grading and cup evaluation data recorded and tracked over time can provide important information to roasters and producers in terms of trending. Beyond the normal fluctuations that can occur in green coffee quality from year to year, annual variation in quality can provide insight into directional movement as indicated by increases or decreases in physical defects or cup quality.
“At the end of each year,” says Trindle Mersch, “Philz prepares supplier report cards, evaluating quality and consistency of lot delivery over the course of a year. Having data captured and at the ready for reporting is important for this process, which is also used to evaluate Philz internal processes and consistency at the cupping table.”
• Establishing Green Coffee Specifications: Correlations between green grading results and cup evaluation can be used to establish green coffee specifications. Once a roaster establishes the range of acceptability for green coffee, specifications can be set to guide purchasing decisions.
• Quality Dispute Resolution: Unfortunately, there are instances when green coffee is delivered that does not meet expectations. Having good supplier relationships, which includes alignment and feedback on expectations, can help minimize these events, and help resolve them effectively should they occur.
“Philz is lucky to have excellent suppliers that are very open to and interested in understanding and helping when defects are identified,” says Trindle Mersch. She credits the regular sharing of cupping data—which builds a common language and trust over time—for this generally positive resolution of issues.
All of these activities require diligently defined and consistent protocols for evaluation, data collection and analysis.
A comprehensive quality program includes analysis and evaluation of all green coffees purchased or considered for purchase. | Photo by Mark Shimahara
Data Over a Green Coffee’s Life Cycle
The quality of a green coffee will change over time. The rate of change differs from coffee to coffee and is dependent on several factors, including parameters at origin as well as transportation and storage conditions. A coffee company may be provided with several samples of the same green coffee over multiple stages of its life cycle. Offer samples, pre-shipment samples, port samples, arrival samples and inventory samples all provide opportunities to perform green coffee evaluations and capture relevant data to track changes in quality over time. Data collected on specific green coffee lots can be extrapolated and applied to other green coffee lots of similar quality type based on modeling to develop a better understanding of green coffee quality dynamics overall.
Purchasing, quality and roasting departments often develop an understanding of the changes that occur in coffee quality and flavor profile from pre-shipment to arrival samples, and between sample-roasted and production-roasted profiles. Monitoring green coffee quality data, such as total moisture and water activity, can be used in determining correlations between these values and cup quality. Correlations can be used to develop models for predicting or anticipating the shelf life of a green coffee. A rapid downward trend in water activity measurements may be correlated with loss of sweetness and acidity and a decline in overall cup score. Measurements can therefore be used to make inventory management decisions, such as reprioritizing the use of a green coffee lot following similar downward trends; or in blend management decisions, such as using a particular green coffee in a blend before it experiences significant loss of quality in storage.
Consistent roasted coffee quality results from consistent green coffee quality and control over the roasting process. | Photo by Mark Shimahara
Roast Profiling Data
A primary role of the coffee roaster is to determine the appropriate roast profile for a given green coffee. This requires an understanding of the green coffee quality, including all physical properties, intrinsic cup quality and the transformations that take place during the roasting process.
“There are certain data points that are essential to keep track of in regard to green coffee, and knowing those data points can give one insight on how to approach a new coffee,” says Steven Lee, director of coffee at SPC Group in Seoul, South Korea. “Moisture content, density and screen size are a must when wanting to take a more analytical approach to roasting coffee.”
It’s also crucial to understand the variables available for manipulation in the roast profile and the influence each variable has on the roasted coffee—all within the greater context of variables that influence roast profile but are outside a roaster’s control, such as relative humidity and atmospheric pressure. Developing this knowledge requires a lot of trial and error but, over time, a roaster will establish a toolkit of roast profiles based on experience and the information collected while building that experience.
“Initially, I collected basic data: time, temperature and color reading,” says Lee. “This was mainly to stay consistent to the profile I was following. It was important to have these data points as parameters to follow and evaluate whether or not your roast was going well. Using that basic data to graph roast curves, I could use those curves as control points and track how variations on those curves would manifest themselves in the cup. Over time, this process would develop into the regimen on which I would train roasters.”
The development of this toolkit follows a logical progression, by grouping green coffees of similar physical properties and intrinsic flavor qualities and applying similar roast parameters. Individual roasters may follow a unique approach to grouping greens and determining profile application. Unfortunately, this process of evaluation and grouping often is poorly documented in terms of the physical properties of the green coffee, profile manipulation, observations, changes to subsequent profiles and the rationale behind modifications to the roast profile. Data collection and analysis greatly assist in grouping green coffees into clusters, as well as determining successful roast profiles for the same or similar green coffee types. This data management creates efficiencies with new green coffees and facilitates the sharing of knowledge within a coffee company and with roasters throughout the industry.
Collecting and analyzing finished-product data can help maximize quality and consistency. | Photo by Mark Shimahara
Roasting and Production Analysis Data
Once a roast profile is established, the role of the roaster transitions to profile maintenance. Consistent roasted coffee quality is a result of consistent green coffee quality and control over the roasting process. Roasting equipment can introduce variables that often are not well understood. Two roasters of the same make and model can require different settings to achieve the same roast profile.
During startup, a roaster may require different gas settings and end-point temperature than it does in steady state to achieve the same minute-by-minute roast progression. Maintenance activities also can necessitate modification to roast profiles.
“Using roast data can help you track the performance of a coffee over time,” says Lee. “Logging the information and cupping results and referring back to them can help one identify drift, and can give one insight on how to handle similar coffees moving forward. ... Over time, a coffee can drift, and the same roast that worked in January may not deliver the same results in May.”
Robust data collection and analysis provides insight into roaster profile settings and idiosyncrasies that can be leveraged in scale-up efforts between roasters of different batch sizes and manufacturers. In addition to green coffee and roast profile data, this often requires finished-product quality evaluation. Results from finished-product evaluations are then available to assist in establishing roast targets and tolerance of a roast profile, ensuring a consistent product.
Additional benefits of data collection and analysis can occur when scheduling roasts. By reviewing production data, a roaster can determine if the performance of a specific roast profile in terms of consistency favors a specific piece of roasting equipment. A roaster also may review historical data to determine if roasting profiles in a certain sequence on a single roaster yields greater consistency. Significant changeovers such as transitioning from regular to decaffeinated coffees, or alternating from high-temperature to low-temperature roast profiles, can create inefficiencies and lead to inconsistent roasts. By characterizing the green types and roast profiles and reviewing finished-product data, a roaster can determine the most efficient schedule to deliver the most consistent roasts.
In addition to green coffee and roast profile data, finished-product quality evaluation is an important aspect of a robust data collection program. | Photo by Mark Shimahara
Business Benefits and Return on Investment
The collection, storage and maintenance of data is an empowering activity. It is available to businesses of all sizes and complexities, and the cost of entry is quite reasonable. To maximize the benefits, roasters should begin collecting as much data as possible as early as possible.
If collected and analyzed effectively, data can transition into information, information into knowledge, and knowledge into wisdom over time. The insight provided from initial efforts can be used to determine future data collection procedures, as well as whether to continue with specific data collection processes, scale back data collection in specific areas, and/or enhance data collection in others.
CHRIS HALLIEN is product manager for Cropster: Roast and Cropster: Coffee Lab. He is a long-standing member of and presenter for the Coffee Roasters Guild and the Specialty Coffee Association, where he serves on the Standards Committee. Hallien is a licensed C grader, a licensed Q grader, a certified sensory scientist and a food scientist.