Star Industries is a company that manufactures and markets windows in Western Australia. They are positioned at the high end of the market in an industry where most competitors bring in lower-priced windows from Asia, and Star trades on their premium branding and reputation. They have limited diversification, but sell throughout Australia, with manufacturing in WA and corporate headquarters in Sydney.
One of the biggest challenges for a small or medium-sized enterprise is to gather data. It is relatively easy to gather data from staff, because they are a captive audience. An app like Officevibe works fairly well in terms of asking staff a handful of questions per week about their views on the company and their jobs. Gathering data from clients is much harder. The response rate on client surveys is around 10-15% at best (Fryrear, 2015), which is usually not statistically significant. This means that the company is unlikely to work with anything other than anecdotal information. So a strategy should be in place to gather that -- ensure that all customer-facing personnel understand what information the organisation wants to get from its clients and that they should take advantage of contacts with clients to seek out that information.
2. Identifying business problems is tricky. You certainly want to have a clear, identifiable process for when a client comes to you with a problem. But ideally you also want a process whereby you can learn from clients about problems as well. A Client Action Committee or some other structure can provide a forum for the sort of open-ended client interactions that would yield information about business problems that may be less apparent. It is also worth checking up on where people talk about you online -- if your company has a Reddit you'll probably be able to read plenty of gripes.
3. Identifying information needed to solve a problem is a trickier matter. For this, the company needs to have a process of inquiries in place. Basically, the easiest problems to resolve are those with metrics, so if possible one should acquire those metrics. The best way to deal with this is by asking the customer. Use your CRM to find out what the customer is thinking -- in other words, how does the customer measure success because that is how you should measure success. You will never know if you do not ask, so use your tools to reach out, or talk to them in person, or have your reps talk to them when they contact you. Each customer interaction is an opportunity to learn how your customers define success.
4. Reliable information is tough, especially for smaller companies. You do the best with what you have at your disposal, recognizing that there is an ROI attached to finding out every piece of information.
5. Formal networks do not yield, by definition, yield information not held in formal systems. Informal networks by definition do. One of the better ideas for informal networks is actually to host events. These can be webinars, conferences, social gatherings -- it sort of depends on your customer base and their geographic dispersion. But you need to create those opportunities for interaction, as those are the informal networks on which you will rely for information.
6. It can be difficult to review knowledge not held in formal systems. The basic approach is probably just to get it into formal systems so that the knowledge becomes institutional, rather than leaving it in people's heads. There are basically two steps to this. First is documentation. This will be a theme here. If you have a documentation platform -- there are several, like IT Glue or something like that - you have a mean by which institutional knowledge can be recorded. But you also need a process for reviewing this knowledge. So within each division and unit you'll need to have a specific time frame for reviewing information that has been recorded. Recording information without setting aside time for review is almost pointless.
7. Testing information for reliability and validity is something that can...
Research design is critical -- you know ahead of time what a reliable sample looks like, and what it will take for a survey of information to achieve validity. You know the limitations of things like anecdotal evidence, such that if that is the best you've got you at least understand the ramifications thereof. Reliability implies multiple repetitions of the same test, while validity requires certain
sample sizes and other elements that the researcher can control (Phelan & Wren, 2006).
Part 2.
8. Ensuring that the objectives of the analysis are clear and consistent with the decision requires is a process before any information -- gathering. The organisation's starting point must be some sort of
strategic objective, and the information would therefore be structured around the idea of supporting that objective. Basically, you can only ask the right questions if you already have thought about what answers you need -- that part comes first.
9. Cause and effect are sometimes difficult to determine empirically. It is not hard to outline
correlation, but causation is a different matter entirely. The logical chain of events is less logical in real life. The biggest thing here is to understand the limitations to the correlation. If you want to prove causation, the data-gathering has to be specifically oriented towards that, to ask it, to measure it from the beginning. If not, the best thing to do is accept that this is a limitation in the research (Srivasta, 2015).
10.
Statistical analysis is critical for understanding data. First, you need someone conversant in
statistics to do this -- others can cobble it together but the interpretation of data will not be strong enough. Excel can run an ANOVA, but it is usually preferred to work with a more robust statistical package like SPSS. Ensure that the information you seek is conducive to statistical data -- again having someone on staff that can design queries, understand limitations and otherwise make a strong contribution to the statistical rigour of any research is important.
11. Sensitivity analysis is interesting because it requires guesswork -- estimates of future outcomes. So in that sense, it's GIGO work. If your
assumptions are trash, your research outcomes will be, too. But a sensitivity analysis just means running the test over multiple different conditions to see how the outcome of the test reacts to changes in a number of different variables (Investopedia, 2016).
12. Documentation should always reflect a logical approach to evidence and conclusions drawn. Ensuring this is a quality control function -- somebody needs to go into the documentation and look at it to make sure that it follows a logical approach and uses evidence. This is not necessarily going to be the case every time so it is the role of the supervisor to ensure that proper documentation is undertaken at all times, providing both carrot and stick motivators to ensure that this is the case.
13. A lot of
management information systems have some flexibility with
respect to their dashboards and with respect to how they gather information. So it is important to ensure that the
software you are using for
business intelligence is customizable to meet your needs. A system of periodic review of such application is necessary to ensure that they are meeting your needs, and if they are not if that is a matter of
training, or if you should look at competing systems.
14. It can be a challenge to ensure that decisions are made using data. So to…
