Current job level is ranked at 31st and 58th out of 92 features in terms of importance in predicting compensation. The highest rank refers to the breakdown shown in the second chart while the lowest ranking breakdown is not shown. The summary was used to mitigate the metric bias in favor of features with many values. Overall, even in the best case, current job level has relatively low importance.
Skill importance rankings for predicting compensation are distributed all across the list. The two highest in terms of importance are Scripted REST API Development and HTML which are ranked 4th and 5th respectively. Oddly, similar skills such as CSS, Front End Development, and Integration Development are a little further down the list. The importance of most skills appears to be highly circumstantial as opposed to a general trend across the population. Whereas most of the demographic and certification features that rank high in importance remain important regardless of decision tree construction, skills will vary greatly in importance depending on which early decision tree splits are chosen. Front end development skills and integration skills do seem to rank among the most important skills regardless of initial splits though, so there is some indication that their appearance is significant and not simply random chance. In the current models, what drives their importance is both the increased likelihood of being compensated above median value when frequently using those skills and the increased likelihood of being compensated below median value when the frequency of using those skills is lower. Both of these skill sets are considered niche skills in the industry, so there is also some common sense support for the results as well.
Most of the ServiceNow Product Usage related features were mid-ranked in terms of importance in predicting compensation. Custom Applications and CSM were ranked the highest among these at 12th and 23rd respectively. Both were associated with increased probabilities of being compensated above median value when using the products frequently and also being compensated below median value when using the products less frequently. Most product usage was not associated with a significant change in the probabilities in either direction.
The decision tree models also indicated that broadly using more of all products may be associated with higher compensation as well. Due to the low importance score of product usage, this was not examined further and the higher importance of skills and capabilities was accepted.
In further support of the findings regarding skills usage, the highest ranked ServiceNow capability was Service Portal / UX / Front End at 16th. Right behind it was the number of ServiceNow Capabilities that the respondent listed as Daily or Weekly used. Similar to the decision tree findings on products and skills, there is a strong indication that breadth of usage is often more important than depth. The survey did not capture depth versus breadth of skill or quality, however, so this should not be taken as an assertion that breadth is universally preferred to depth. Sufficient familiarity to frequently leverage a broad set of ServiceNow applications, skills, and capabilities does often have a stronger impact on compensation than high frequency use of specific ones with only a few examples already mentioned.
I do not contribute directly to maintaining or improving quality
50.00%
25.00%
25.00%
My org would not be negatively impacted if I left
34.62%
23.08%
42.31%
My org would not find it challenging to replace me
38.24%
20.59%
41.18%
Overall, one’s statements about their perceived contributions and importance held little value in terms of predicting compensation. Respondent’s response to the statement “My organization would be negatively impacted if I left” was ranked the highest at 29th. It is noteworthy that a respondents belief that they do not contribute to maintaining quality, improving quality, or reducing costs was associated with only a 25% probability of being compensated above median value. For those looking to maximize their value, a failure to contribute to an organization’s quality or cost reduction could be costly. Conversely, contribution to revenue had virtually no impact which is quite possibly the oddest finding of the analysis.
Satisfaction was strangely one of the least important feature sets in terms of predicting compensation with the most important feature being satisfaction with future career prospects at 54th in the list. Overall, satisfaction did not seem strongly correlated with any other particular feature either. Those with low compensation were as likely to be satisfied as not. The same can be said for those with low or high years of experience. A respondent’s satisfaction appeared to be largely independent of any other feature, including compensation.
Hours worked this year was ranked 41st in terms of importance in predicting compensation. There was very little evidence to indicate that there was any strong correlation between hours of work and total compensation. Whether examining this feature through correlated regressions or probability splits, this feature had little impact on compensation and appeared to be mostly uncorrelated to other features.
Let's Chat
Looking to ask a question or get some help? Feel free to reach out in whatever way is most convenient to you. Here are a few ways:
