Results and discussion

Example of methodology application

In this example, we describe an application of the methodology in a Canada-based retail business. The particular retail business studied here is a small enterprise (with revenues less than $50 million) with four local stores. The business has existed for over 15 years, establishing a strong foothold in one neighborhood, and is now planning to expand nationally. Three years prior to the study, the business was purchased by new owners who set national growth as a key strategic objective. As part of the expansion plans, market and competitive studies were conducted. In addition, the owners had created a scorecard that tracked key operational indicators and provided the ability to conduct an assessment of business results. Some data however, for example, the flow of customers through each of the locations, was not yet available in the scorecard.

At the time of the study, most revenues were earned through consignment sales. The business had started selling new items as well and was planning to invest in an online business. Revenue was driven by ensuring that stock was properly displayed, which in turn depended on assuring that enough staff were available to sort, tag, and lay out the products. The supply side of the business depended on the number of consigners available, the amount of product they brought to each store, and the speed at which these products could be displayed. The demand side depended on local advertising and word of mouth that stimulated traffic flow. All stores were situated in prominent locations with good visibility that stimulated walk-in traffic.

To begin our investigation, playing the analyst role in Step 1, we interviewed the CEO in order to identify the high-level goals as well as KPIs and drivers of organizational success. As depicted in Figure 17, the goals of the principals i.e., business owners, were related to market growth: they wanted to be the number one distributor within their geographical market. Store managers were aware of the growth objective, but on the short term, they focused on increasing revenues and the number of items sold in their stores.


Figure 17


First iteration model (aggregation formulas not shown here).


As discussed above, the first iteration of the model provides an initial alignment of higher-level goals and KPIs. We started with a minimal decision model and limited set of data just to illustrate the business goals and financial targets and to identify the indicators and driver KPIs required to monitor the business and to make informed business decisions. Figure 17 illustrates the first iteration of the model. At this stage, we also developed a rough dimensional model (Figure 18) in order to ensure that the data needed for the decision model would be available. The dimensional model helps the store to analyze the impact of the KPIs on goals based on their different store locations, in each period of time, by product type (e.g., clothing, electronics, etc.), and by product category (i.e., retail and consignment).


Figure 18


First iteration dimensions.


In Step 2 of the methodology (Table 2), more KPIs were added to the model as drivers and then linked to the high-level financial KPIs and organizational goals. In addition, we added the new KPIs as well as a new dimension called "marketing type" (e.g., outreach, online advertising, etc.) to the dimensional model. This new dimension allows decision makers to analyze which marketing initiative has a more significant impact on the goals. In this step we also created a decision options diagram (Figure 19) illustrating the specific actions managers can take to improve goal accomplishment. One of the decision options available to managers in this case, is to invest in an online business by increasing the advertising budget for the website and increasing the website maintenance budget. We consider this option as the decision made by managers and update the models while transitioning to Step 3.


Figure 19


Second iteration decision options diagram.


In Step 3 of the methodology (Table 2), we added situations, new KPIs and started the monitoring effort. Figure 20 depicts the third iteration model, which defines the expected impact of the actions identified in Figure 19 along with the KPIs and acceptable ranges for each of the relevant goals based on the situations associated with each goal. Table 6 shows the formulas defined between the KPIs in this model. There is one new situation element in the model reflecting the initial cost of investment in the online business. This situation is a threat at this point in time. Therefore, the situation increases the acceptable range of the profit values by moving the threshold value closer to the worst value, as explained in Table 4. Furthermore, there is also a new KPI used to monitor the investment made in the online business and its impact on the costs. The GRL strategy used for the evaluation here focuses on the use of the online business investment (other GRL strategies were defined to evaluate different sets of options and find the most suitable trade-off).


Figure 20


Third iteration model (evaluated).


Table 6 Third iteration model formulas

Formula Target KPI Source KPIs
2 × MC/10 Store traffic MC: Marketing cost
floor(MC/3.67) Number of consigners MC: Marketing cost
floor(1.76 × NC) Number of drop-offs NC: Number of consigners
365 × 8 × NSPD Staff total work hours NSPD: Number of staff per day
STWH/TNS Work hour per staff STWH: Staff total work hour
TNS: Total number of staff
STWH × 8 + STWH × 4 Staffing cost STWH: Staff total work hour
$8 per hour + $4 overhead
STWH × 1.5 Number of products available for customers STWH: Staff total work hour
StaffC + MarketingC + StoreC + R × 0.6 + OBI Costs StaffC: Staffing cost
MarketingC: Marketing cost
StoreC: Store cost
OBI: Online business investment
R: Revenue
R-C Profit R: Revenue
C: Cost
(R × 100)/MV Market share MV: Market value



Figure 21 depicts the third iteration dimensional model (including its new "Sales method" dimension) used to ensure that decision makers can compare and analyze the numbers for the online part of the business separately. In addition, this model now also supports the location dimension on more KPIs, allowing business owners to compare the stores. Note that, in jUCMNav, such figures can be split over many diagrams when they become too large or complex.


Figure 21


Third iteration dimensions.


Another enlightening aspect of the third iteration model, Figure 20, is the evaluation value of the defined KPIs as well as the relationship of the KPIs with various stakeholder goals. Note also in this figure that this performance monitoring view in jUCMNav was enhanced to show the KPI values themselves (in blue), with their units. As shown, the evaluation value of "Average turnover days", which has significant impact on the satisfaction of both Revenue and Consigners, is 55. In this case, the lower the number of turnover days, the better the outcome. Therefore, according to the model, business owners have to reduce average turnover time to increase both customer satisfaction and store revenues. According to Figure 19, the latest version of the decision model, the only option was to increase the number of staff. However, when the new location dimension added to Figure 21 is used, one could easily point out that the average turnover days were significantly worse in one of the stores even though the number of staff in all stores were the same. After some brainstorming, owners realized that the design of that specific store as well as how the drop-off is handled in that store is contributing to the higher number of turnover days. Therefore, their decision model was updated (Figure 22) to show process improvement as an alternative to increasing the number products available to customers, which according to the model directly impacts the average turnover days. This process improvement will be specified with Use Case Maps.


Figure 22


Fourth iteration decision model.


The change in the decision model also highlighted a potential enhancement in the third iteration monitoring model (Figure 20). Although the impact of "Number of products available to customers" on "Average turnover days" is illustrated in the model, the target value for the former is not accurate. To elaborate, the discussed KPI evaluation value is 100, which means it fully satisfies the expected value, whereas we know we have to improve this number to improve the average turnover days. This is another example that shows how not having enough information about the context and relationships at the beginning is not a showstopper for using this methodology. Analyzing the model at each iteration can help us both to understand the business more accurately and to improve the model itself and change the business targets accordingly.

In terms of adaptation, the owners decided to improve the drop-off process model. The as-is drop-off process was modelled using UCM, as shown in Figure 23. After some brainstorming, a to-be process was proposed as well (Figure 24). This to-be process was designed to minimize the time that staff spend on the drop-off process as well as reducing the wait time between item drop-off and item display.


Figure 23


Drop-off as-is process model.


Figure 24


Drop-off to-be process model.


As illustrated in Figure 23, in the as-is process, all the interactions with customers are done manually, including the registration of new customers and the entire process for dropping off items to be sold. This problem was addressed in the to-be process, Figure 24, by adding kiosks to the process to automate part of this work and reduce the staff workload.

In addition, in the as-is process, there are two sorting activities in the process, one in the presence of the customer and another one later on. If a customer does not want to donate the items deemed "unacceptable", then there is a follow-up activity to call the customer to pick up these rejected items. This sequence was streamlined in the to-be process by sorting all the items in the presence of the customer or mandatory donation as the two possible options for the customer. This change not only reduces the staff workload but also removes a backlog of items to be sorted from the entire process and significantly reduces the time lag between drop-off and display.

After the suggested changes to the process model are implemented, the methodology allows the owners to monitor the results and see the impact on the KPIs that were assumed to be improved using the new process model (i.e., average turnover days in this example).

Although the complete model in the third iteration, Figure 20, seems complex, this is not the view that everyone in the organization has to deal with. The tool allows organization to create the custom diagrams from the same model to adjust the complexity of the method for different level of users and only expose the users to the subset of the model and method that helps them to perform their job. For instance, Figure 25 illustrates a subset of the model that is customized to show only the elements that business principals would be interested in. This strategic view only has the very high level goals of the organization and the related high-level KPIs and does not include any operational goals that normally store managers would be interested in.


Figure 25


Principals/strategic view of the model.