Most of my consulting services are on analytical projects, where I help my clients, primarily utilities, analyze data and develop forecasts. Sometimes, I also take some management consulting projects, where I help some solution providers with their sales strategy, product management and marketing messages, with the hope that they can sell the right solutions to the right customers.
Energy forecasting is an interesting community. On the academic side, there are thousands of papers being published every year, of which most are nonsense and/or useless. On the business side, there are dozens of, if not hundreds of vendors and consultants trying to gain a market share. In such a noisy space, marketing becomes particularly important. A bad marketing practice may waste years of R&D efforts and make the sales job extremely difficult. Below I'm exposing three worst buzzwords for marketing an energy forecasting solution. Most of these thoughts are applicable to forecasting solutions in general.
Well, I guess most people would not walk to a marketing event telling customers the solution is inaccurate. Although ALL forecasts are wrong, most marketers love to tell customers the opposite, such as "our solution is very accurate." As I mentioned in Three More Must Know Basics about Forecasting, accuracy is never guaranteed. Telling the customers about an accurate solution is equivalent to telling them that the solution provider 1) has no idea about forecasting or 2) treats the customers as if they know nothing about forecasting.
If the statement "our solution is very accurate" is obviously incorrect, there is a less obvious one: "our solution is the most accurate one in the market". Since nobody can guarantee the absolute accuracy, shall we guarantee the superior accuracy over the others? My answer is NO. As discussed in Three Must Know Basics about Forecasting, forecasting is a stochastic problem. A solution may be better than others most of the time, but should not be always. There is a lucky factor in forecasting. Sometimes a poor solution may be well suitable for some cases and luckily result in less error than a good solution. Before doing a pilot, nobody can claim the superior accuracy. Even after the pilot, we cannot guarantee the consistent superiority in the future. Nevertheless, a well-designed pilot can offer customers a comprehensive view of the performance of competing solutions.
The most nontrivial way of claiming accuracy is through selected extreme events, such as "my forecast was right yesterday", "my model caught the annual peak last year", and "my MAPE was less than 1% ten days in a row". This marketing approach takes advantage of how human brain works: human beings tend to memorize and overestimate the likelihood of the extreme events. By emphasizing the extremely favorable events when a solution outperforms the others, the marketers bring customers an exaggerated message that the solution is super accurate. This approach seems effective, but is a double-edged sword. While there are extremely good days, there are very bad days too. When the solution's performance goes south, the customers can remember them as well. Being set at an unrealistic expectation by the exaggerated marketing messages, the customers may take an action stronger than necessary.
Many marketers like to tell customers that the solution is "award-winning". In other words, "not only us think we have a great solution, the market believes so too!" There are several issues with awards:
- If the award is really a prestigious one, that's fine. However, in today's market, there are too many marketing firms "selling" awards to make a living. It's hard to tell which awards really have heavy weights.
- Some vendors care less about the award, so they spend no time on the surveys from the marketing firms. Some vendors like to go after the awards as much as possible.
- There are niche players in a specialized area, where marketing firms don't even bother to make an award. For instance, I have not heard of an award for outstanding load forecasting vendors. If any of them participate and win the Global Energy Forecasting Competition, I will be happy to give them an award.
Somehow "proprietary" becomes a key word to indicate that there is some secret sauce making the solution unique, special and better than others. In reality, the word "proprietary" is quite often used to hide how crappy or naive the core technology is under the fancy cover. I do understand and agree that the code can be proprietary. On the other hand, the forecasting methodologies are being enhanced by the scientific community every day. A truly innovative solution provider should always look for ways to improve the forecasting methodology. Why bother making the methodology proprietary?
Another form of "proprietary" is patent. It's a legitimate practice to file patents to protect the company's intellectual property. However, having more patents than others does not mean that the solution is better. In a patent filing process, nobody would look into the idea and see if it is really valid. There is no direct connection between patents and prestige of the solution. For instance, I had an experience of sitting in an debate between a customer and a consultant. The customer said the solution is not doing what it is supposed to do, while the consultant said the methodology implemented in the solution was patented. It did not take me more than 3 minutes to figure out that the solution, or more specifically, the patented methodology, was not properly designed to solve the real-world problem.
Happy Chinese New Year of Horse!
Good advice. I'm retired now after spending my whole career in the load forecasting game, and it always drove me bananas to hear talk (mostly from those who didn't know any better) about the "accuracy of the forecast", not to mention the complete misinterpretation of the meaning of a confidence interval. Even worse, on one job, my performance was evaluated by my forecasting "accuracy", when my boss wouldn't let me make even the smallest decision related to methods or assumptions! Good riddance to that!ReplyDelete
And gung hay fat choy to all in the load forecasting world!