Ekta neatly summarises the essence of Jack Phillips' ROI approach. Thank you Ekta!
The Phillips approach has a number of useful characteristics but if you don't appreciate the subtleties of some of them you can create problems if you don't watch out for certain danger spots. Let me explain further...
1. Obtain data to demonstrate the changes in behavior, e.g., that
gathered through surveys, questionnaires, on-the-job observations,
post-program interviews, focus groups, performance monitoring.
The potential danger to watch out for with Step #1 is root cause - you need to KNOW what the root causes are for the changes in behaviour - just observing the changes, getting survey and interview data, esp. after the training, will only show CORRELATION, not CAUSATION. Get this bit right and you SHOULD have a number of measures of performance that are readily convertable in to financial values, which will be useful later on.
2. Isolate the effect of training, e.g., through the use of control groups,
trend lines, forecasting models.
There are a number of dangers to watch out for in Step #2 as well. First, it may not be appropriate to isolate the effects of training. If the training is part of a project to address performance, as is almost always the case, and if this training is a 'must have' part of the project, i.e. no training means no required improvement in performance, then it is meaningless to isolate the effects of training, you have to look at the project as a whole - and neither can you necessarily claim 100% of the benefits of the project simply because without the training the project will fail to deliver the required performance improvement. That said, if you do the root cause analysis properly you will get a solid measure of how much of the problem can be addressed by the training aspect - so there will be times when isolating the effect of training does make sense - just be sure you know what you are doing and why - there are plenty of CEOs and CFOs that will shoot your ROI claims down in flames because you have not been rigorous enough in this respect.
Secondly, trend lines again only show CORRELATION, not CAUSATION - unless you have done a thorough root cause analysis and your plots of trends are on the right data. Further more. trend line plots may not show easily problems in the processes that deliver the performance in the first place - they are not the best tools for showing if the process is 'in control'. If the process is not 'in control' then no amount of forecasting trends and extrapolating will get you credible data - where forecasts appear to be accurate will be out of sheer luck - nothing more.
Control groups are fine to use PROVIDED you account for all the major factors. It is almost always the case that comparing one group that has had the training with another that has not had the training will tell you NOTHING other than the fact that there may be a difference in performance between the two. It CANNOT tell you WHY the difference is there because you have not taken steps to consider and robustly discount other factors that could affect performance.
3. Convert the data to monetary value by focusing on a unit of
measure, determining a value for that unit, calculating the change
in performance data, determining the annual amount for the
change, and calculating the total value of the improvement.
If you recall my comments from Step #1, if you have a decent root cause analysis you SHOULD have some measures of performance that are readily convertable to financial values.
4. Tabulate the program costs: this is the value of the cost of taking
people away from their jobs for the training, including salary and
benefits.
Yes - and also the development and delivery costs of the training course, and any accommodation and travel costs that go with it.
5. Calculate the return on investment by dividing the net benefits
by the costs times 100 percent
Because these figures are based on a comprehensive root cause analysis, and robut use of control groups for example, and supported if you wish by estimates of contribution by operators and other experts impacted by the training, you can rest easy in the knowledge that there isn't a CEO or CFO alive that can shoot your claims down - except to fulfil a hidden agenda, in which case normal reasoning goes out the window and it's all down to politics!!
I speak from some experience - I followed the Phillips approach and used control groups and trend lines and observations of behaviour - presented them to the Board of Directors, who then promptly shot holes in the claims I was making - because I couldn't PROVE my claims. Along with 85 other people I lost my job during the next difficult trading period about 3 months later because I couldn't PROVE the contribution of the training function - just the fact that we were spending a lot of money on training.
Use the Phillips ROI approach, but like all methods (including my own 5 Boxes) use it with care and from a position of intimate knowledge.
Good luck!
Regards
Martin