As a little treat for the developers and customers at Dot Net Solutions we’re buying a new coffee machine.
We’re a pretty diverse bunch: several people have coffee machines at home and have emotional attachments to different brands, and there are tea drinkers who may be served adequately by some of the choices, or not at all by others. Not dissimilar from the situation some of our customers are in when agreeing features for a new software system.
When evaluating competing architectures for a software solution, one common technique is to evaluate the different options quantitatively and we’ve applied this to selection of the coffee machine.
We considered the core requirement “I want to be able to make great coffee in the office” through several rounds of analysis.
In the first round we consulted all the developers (including the tea drinkers) on their preferences for a coffee machine. We came up with a simple list of quality attributes that we would measure against. This gave us a list of around 20 different attributes to judge the coffee machine on. 20 different criteria is a lot, so we filtered them through voting. Giving each developer 4 votes to select the attributes they believed were the most important, and merging similar attributes brought the number quickly to 7 with a basic weighting.
| Attribute | Weight |
| Espresso quality | 11 |
| Maintenance effort | 6 |
| Frothy milk drink quality | 4 |
| Usability | 4 |
| Speed per cup | 4 |
| Price per cup | 2 |
| Capacity to make tea | 1 |
Forced to make a selection, I found it interesting to watch people choosing attributes other than their own proposals. This was the major reason attributes dropped off the table and was a productive way to achieve consensus.
A range of coffee ‘machines’ was put on the board, from a pump machine taking ground coffee to various capsule machines aimed at different market segments. One interesting outlier was proposed at this point: CaaS. Coffee as a Service would be negotiated at a local coffee shop and presented as an option.
Each machine was evaluated in isolation giving scores out of 5 for each of the quality attributes based on on-line reviews and subjective opinion.
There were a few high points from the feedback. By presenting a regular work stream to a local coffee shop we were able to gain agreement for reduced cost, as well as on site delivery. Negotiation, even in its simplest form, can pay great dividends.
Feedback was also used to normalize scores – as this was a group activity we were able to question each other and ensure scoring was consistent so that machines with similar maintenance activities were scored the same for example.
| | Espresso Quality | Maintenance Effort | Frothy Coffee | Usability | Speed | Price | Tea | Weighted Total |
| Weight | 11 | 6 | 4 | 4 | 4 | 2 | 1 | |
| Tassimo | 4 | 4 | 2 | 5 | 3 | 4 | 5 | 121 |
| Dolce Gusto | 4 | 3 | 1 | 4 | 4 | 3 | 0 | 116 |
| Nespresso | 4 | 3 | 3 | 4 | 4 | 5 | 1 | 117 |
| Pump machine | 5 | 2 | 5 | 1 | 2 | 5 | 5 | 114 |
| Cafe | 5 | 5 | 5 | 2 | 1 | 1 | 5 | 124 |
The first surprise was how close the scores were. When there is a clear leader things are simpler: that should be your selection. Given the margin for error in our evaluations these scores are remarkably close. Perhaps a marketing guru will give us an explanation of why this occurs, perhaps it is that individual manufacturers are well aware of variations in the market and target these accordingly so that for individuals there is a clearer lead, however for a small team the values tend to average out.
We eliminated some of the extremes. e.g. the semi-automatic pump machine clearly makes great coffee, but there will be a limited number of people in the office prepared to use it.
The Cafe fails a several criteria in the original requirement “I want to be able to make great coffee in the office”. There are several advantages of on site production compared to outsourcing, it tends to make a great coffee smell through the office which several of us enjoy, and the office community is built up by people serving each other through making drinks. The ability to make customers a coffee is similarly valuable and sending out for coffee as people arrive would be disruptive. Finally despite the price per cup being weighted quite low everyone was still shocked at how much we would spend on coffee even given the negotiated discount.
In the end we came down to two options the Tassimo and Nesresso machines and after discussion the Tassimo won out on reduced maintenance effort. We were still left with a problem. There were some reports that the Tassimo milk tastes quite nasty and so we parked discussion to look for ‘architecural variants’.
We found an architectural variation – milk frothers are available as separate devices and after a little research settled on an Aeroccino Nespresso milk frother.
We cut through all the office politics on coffee machines in an enjoyable half a day and have a solution that serves a broad range of tastes. Come see what we can do for your business – and have a coffee while you’re here.