A comprehensive list of current and future predictive talent metrics
The use of predictive analytics is a hot issue and a developing trend in talent management. But unfortunately as a longtime thought leader in the area, most of the current prediction efforts are extremely shallow. And as a result, they will have a minimal impact because they only cover a few basic areas like predicting employee flight risk and identifying the selection factors that predict hiring success. What will eventually be needed is a broader array of second- and third-generation predictive metrics covering many more advanced talent management factors.
If you’re curious about what factors must be measured in the future, here is a comprehensive list of the predictive talent analytics/metrics that should eventually be developed by forward-looking talent leaders.
The First Generation of Predictive Analytics
The best firms like Google and the leading-edge vendors are already providing predictive metrics in these “first generation” areas. Consider the current limited round of predictive metrics as part of a first step toward the inevitability of a completely data-driven talent management function. The data-driven approach is inevitable because every other business function long ago shifted to that type of decision-making, which is one reason why HR is rated by executives as last among all strategic business functions (from Simon Mitchell of DDI). The most common current metrics in talent management include:
- Identifying which key employees will soon be a flight risk — perhaps the easiest of all talent factors to predict, this metric requires that you identify the factors or precursors that forecast that a key employee will soon consider quitting (Google is a benchmark firm). The resulting metric should assign a probability number to each targeted employee covering the odds of them leaving within the next few months.
- Identifying the selection factors that predict on-the-job performance of new hires —a hiring algorithm that can successfully predict the characteristics of applicants and interviewees that will result in successful and higher than average performance on the job. A similar algorithm needs to be developed for college hires (Xerox, Google, and GateGourmet are benchmark firms).
- Forecasting when a change in employee survey scores will begin to impact productivity — this predictive metric determines the correlation between productivity and employee survey scores. It then forecasts when significant change in the scores will positively or negatively impact a team’s productivity and retention.
The Second Generation of Predictive Analytics
Currently few firms (Google is the benchmark firm) are providing metrics in these advanced areas. Although these second-generation predictive analytics are more difficult to develop, it’s hard to argue against the potential business impacts that these predictive metrics will eventually provide.
The calculation and the delivery of metrics will also change during the second generation. Even though most talent metrics are now delivered to HR users, in the future they will be accessed via mobile phone and used directly by operating managers. Data will be gathered using scientific sampling rather than measuring every case or employee. These next generation metrics will be presented in both report and visual trend-line formats. There will also be an alert process to give managers sufficient time to prepare before an upcoming problem or opportunity occurs. And finally this new approach should report the correlation between a manager’s use of predictive talent metrics and their improved business results. The top 18 most-likely second generation predictive metrics are listed below:
- Projecting the dollars of business impact resulting from talent actions — this most important second-generation metric converts HR results into dollars, because that is the common denominator throughout the corporation for measuring business results. This metric results from a process of identifying and quantifying (in dollars) the revenue impact of each of a firm’s talent programs. Work with the CFO to ensure that the conversion process and the revenue impact dollar estimates are credible.
- A metric revealing the current and projected improvement in revenue per employee — the Holy Grail of talent metrics is a single number that serves as an indicator of both the current and future productivity of the total workforce. That single metric is likely to be projecting the yearly improvement in the “revenue per employee” calculation (which is corporate revenue divided by the number of FTEs). Or alternatively, projecting the improvement in the ratio between corporate revenue and total labor costs. This revenue per employee dollar number has great value because it can easily be compared both year to year and between different companies. (Note: the current revenue per employee for major firms can be found by typing in the stock symbol in the search box on MarketWatch.com).
- An easy-to-compare talent management performance index — outside of HR, indexes are widely used to combine multiple metrics into a single standardized number (e.g. the Dow Jones Average is an index). Having a single talent index will allow executives to easily and quickly compare the talent performance of different business units, teams and managers. The talent index would combine the performance of a unit in each of the key talent areas including recruiting, retention, development, innovation etc. into a single number. The talent index score would be a single number based around an average score of 100, where 80 would reveal talent performance problems and a 120 score would show excellence in that business unit. This single “talent performance index number” would allow talent leaders to easily identify and then focus on problem business units and managers.
- A “WAR” metric that places a numerical replacement cost on individual employees — perhaps the most influential recently developed metric in baseball is the WAR metric (Wins Above Replacement). It projects the likely value of an individual player if they were lost and were then replaced by a readily available minor league player. In talent management, although complicated, a single index called “output value above the average replacement” would make it crystal clear which employees you couldn’t afford to lose (because their replacements would likely underperform them). Calculating and placing a single numerical value on key individual employees who could quit/retire would help leaders also determine an employee’s relative actual value for compensation, retention, and succession planning purposes.
- Predicting upcoming productivity issues — employee productivity is rarely measured and reported in today’s corporations. In the future, an algorithm that statistically determines which factors positively impact employee productivity can add great value. A related algorithm that predicts where and when within the corporation that major individual and team performance/productivity problems will likely arise will also be valuable. Predicting upcoming productivity problems will give leaders sufficient time to develop approaches to mitigate those upcoming problems. Adding a visual trend line showing the trajectory of the productivity curve will make upcoming problems and opportunities easier for managers to spot. A similar algorithm and trendline revealing predicted decreases in innovation will also add value.
- A metric predicting upcoming employee behavioral issues — allowing employee behavioral problems to fester can be expensive. So providing a metric that predicts the specific areas where employee behavioral problems (i.e. excessive absenteeism, excessive sick leave usage, sexual-harassment, engagement, safety, low morale, a high error rate) will likely occur can allow leaders to act proactively. Using data to identify the most effective remedies to these behavioral issues would also add value.
- Plotting the career trajectory of new hires and employees — projecting how high and how fast a new hire is likely to progress upward is critical in a growing organization that needs employees who are capable of moving up fast. A similar career trajectory could also be plotted for current employees for career development and succession planning purposes. Plotting how long a new hire or an employee will likely stay in the organization would also be valuable.
- Identify the factors that impact manager success — having effective managers is essential for productivity, innovation, hiring, and reducing turnover. Rather than using the current unscientific approach, first use multiple regression to identify the factors that top-performing managers have in common. And then use an algorithm with those characteristics (e.g. project oxygen at Google) to guide your selection of new managers. Those predictive factors can also be used to identify and improve weak managers and to identify the ones that should be on the succession plan.
- Identify the factors that identify leaders or leadership capabilities — identifying high-potential leaders among new hires and current employees has frequently failed because the typical approach has not been data-driven (Research by the CEB revealed that “more than two-thirds of companies are misidentifying their high-potential employees”). An algorithm that successfully identifies the characteristics of potential and actual leaders can be used to more accurately screen applicants and interviewees for those with leadership capabilities. A similar algorithm applied to current employees would also be invaluable for accurate succession planning.
- Identify the selection factors that predict an innovator — innovation has a large and growing business impact. As a result, recruiting must become more effective in identifying innovators among those who apply. It must develop a hiring algorithm covering innovation characteristics that can successfully predict which candidates will be successful innovators in key jobs after they are hired.
- A metric predicting an upcoming large volume of position openings — going beyond predicting individual turnover, effective workforce planning requires that you can also predict which jobs will likely have a large number of upcoming openings. Knowing which jobs will soon have many openings because of turnover or growth will allow firms to proactively ramp up training, internal movement, or external hiring before those excessive vacancies have a significant negative business impact.
- A metric predicting upcoming external talent availability and talent opportunities — most firms hire exclusively when they have a current opening, while the best firms proactively “ramp up their hiring” whenever top talent is highly available. As a result, a metric that predicts time periods where there will be a talent surplus in key areas would allow a firm to “load up” on top talent. A similar metric revealing when there will be reduced competition for that talent (because of reduced hiring or a hiring freeze by your competitors) can also allow a firm to hire an exceptionally high volume of high-quality talent.
- Predicting areas where significant internal redeployment will be required — because business needs vary and change throughout a corporation, some business units will need to expand while others will need to shrink. A predictive metric can forecast when and where in the organization there will be a surplus of talent, so that it can be redeployed into areas where it would have a higher impact. Individuals who are soon to become “overdue for internal movement” can also be identified.
- Projecting the appropriate % of contingent workers — in a volatile VUCA world, the ability to rapidly increase talent capabilities or to rapidly reduce labor costs are both critical agility factors. As a result, a metric forecasting the appropriate future contingent labor percentage for that time period and business growth rate can add great value. A related metric predicting when outsourcing work is appropriate would also improve talent agility.
- Predicting the viability of technology substitutes — for years having employees was the only solution that HR had for every “I need work done” problem. But now that hardware, software, and the Internet have dramatically evolved, the time will come when HR must also routinely consider technology substitutes for labor. A metric-driven formula that can successfully determine when technology solutions are a superior substitute for labor will someday become common in HR.
- Forecasting upcoming retirements — unexpected large-scale baby boom retirements may create severe internal talent shortages. An effective predictive metric that can forecast retirement trends and accurately predict when and in which jobs those upcoming retirements will likely occur in will help with recruiting and succession planning.
- A metric predicting when an employee will become underpaid — feeling underpaid is a major cause of both turnover and lower productivity. Proving a correlation between an employee’s “underpaid status” and a decrease in their productivity and retention will be a valuable argument for convincing managers to proactively bump up pay. With such a relationship confirmed, a metric that forecasts into the future precisely when employees in each job family will likely reach “underpaid status” (compared to regional market rates) would allow leaders to act proactively before compensation became an issue. An algorithm that successfully predicts when key individual employees will (without a raise) reach an underpaid status would also add value.
- Identifying diversity roadblocks — as convincing evidence continues to emerge on the positive business impacts of diversity, it will become essential to increase diversity hiring, promotion, and retention. A process that can scientifically identify the barriers that restrict diversity hiring, promotion, and productivity will be needed. An algorithm will also need to be developed that determines the optimal percentage of diversity in a team and in which specific jobs does having the target diversity percentage have the largest business impact.
I am excited about all of the current attention being given to predictive metrics, but at the same time, I am concerned because most talent management leaders seem to be overly satisfied to be simply “working on” or “by having a few” predictive metrics. Most in the field who I encounter have simply not taken the time to identify the many additional areas where predictive metrics will add even more value. My goal is for this list of second-generation of predictive metrics to stir your thinking and to lead you to develop metrics in new and uncharted areas of talent management.
The Future of Predictive Analytics — the Next Generation of Talent Metrics to Consider (Part 2 of 2)
A comprehensive list of future predictive talent metrics
In last week’s part one of this article that was published on March 9, 2015, I highlighted the fact that the majority of current predictive metric efforts have focused on only a handful of basic metrics. I next provided a list of the top 18 metrics that should be developed during the second-generation of predictive metrics. This final part one covers the future predictive metrics that should be developed during the third generation.
The Third Generation Of Predictive Analytics
If predictive metrics in talent management follow the same path that occurred in supply chains and in professional baseball, we can expect a wide array of sophisticated public and proprietary predictive metrics to be developed over the next five years. Obviously it’s hard this far out to precisely predict the exact metrics that will be developed during this third generation. But I still have been able to compile a list of more than 25 possible advanced predictive analytics that would add tremendous value after being developed. Some of the metrics that are outlined in this section may even prove valuable enough to be developed earlier during the second-generation phase. The metrics in this final section are categorized into the different functional areas of talent management. During this third generation expect that modeling capabilities will become commonplace and if/then scenario planning will allow managers to pretest their talent decisions. You should also anticipate the extended use of heuristics to identify similar problem areas and to learn from data and improve.
Recruiting-related Third Generation Predictive Metrics
The recruiting function currently has the best metrics in talent management, but there’s always room for growth into these new areas.
- Identifying factors that predict if a recruiting prospect is about to quit — some vendors are already using public and social media information to identify which employees at their own firm are likely to quit. Eventually the same approach will be adapted so that recruiters can identify which recruiting targets may soon be receptive to a recruiting pitch.
- Projecting new effective recruiting sources — the effectiveness of recruiting sources is continually changing. So predicting where and when source effectiveness will shift will allow recruiters to use the best new sources, which will directly improve their quality of hire.
- Predicting which employees are most likely to make top referrals — because of the many advantages of employee referrals, a metric is needed that effectively identifies which employees are most likely to be able to make quality employee referrals for each job family.
- A metric that predicts the likely return of boomerang former employees — rehiring top performers who left your organization has a high ROI because they already fit your organization. As a result, a metric that predicts when the most desirable former employees are most likely to be receptive to an offer to return will result in many quality hires.
- Forecasting changing candidate expectations — in a volatile world, candidate expectations are continually changing. As a result, a metric that would predict when and how the expectations of our target candidates will shift would allow a firm to change their offerings to recruits to better match those changing needs.
- A metric that predicts new-hire failures —predicting which new hires are likely to be failures on the job within their first six months can allow recruiters to proactively begin looking for a replacement before the termination date of the employee.
- Forecasting future employer brand strength — this metric will predict when and why our employer brand strength will increase or decrease over time, compared to others. Projecting your future brand strength is important because employer brand strength is the No. 1 factor in attracting top talent.
- Identify future corporate acqui-hire targets — if your firm purchases firms for talent, develop a metric that identifies which talent-rich firms will be a good match for purchase or a merger.
Productivity-related Third-generation Predictive Analytics
Increasing workforce productivity is by far the most important talent area that is almost completely ignored by both talent leaders and metric vendors.
- Identify barriers that limit productivity — one of the most effective ways for increasing employee productivity is to identify and then reduce the easily fixable barriers that keep employees from increasing their productivity. As a result, a metric that identified current barriers and predicted upcoming barriers would have a large impact.
- A metric that predicts upcoming employee burnout or obsolescence — develop a metric for identifying individual employees who are approaching job burnout so that they can be moved or replaced. A related metric can also identify employees who will soon become obsolete but they can’t be retrained and those employees who have reached the top of their career trajectory.
- A metric that predicts which employees will likely need to be released for cause — this metric pre-identifies the individual employees who are likely to be fired or released for performance issues. A warning alert to the recruiting function can allow replacement recruiting to begin before the problem employee must be terminated.
- Develop a metric that projects possible future layoffs — predict the jobs or the business units where high labor costs, a surplus of employees, or a reduction in the work load will likely require layoffs. Use data to pre-identify the specific individuals who would likely qualify for an upcoming layoff.
Retention-related Third Generation Predictive Analytics
The issue in talent management that is about to explode and then continue to be an issue for several years is the rapidly rising turnover rate.
- Identify the employees who will soon become “overdue” — this metric would track and identify when an individual employee will likely become “overdue” on critical factors that may cause them to quit or be less motivated. Those overdue factors might include a raise, a promotion, more training, new equipment, recognition, a job rotation, etc.
- Develop a metric that predicts which employees will be targeted by external recruiters — in order to be proactive and to allow time to develop a blocking strategy, create a metric that predicts which specific employees are most likely to be targeted by recruiters from your competitor firms.
- A metric for forecasting the changing turnover causes and actions — even today I recommend using post exit interviews to identify why key individuals quit. Additional tracking and projections can reveal when turnover causes are shifting. A related metric can reveal what specific actions are effective for countering each major cause of turnover.
Leadership-related third-generation predictive analytics
Unfortunately the leadership function has been one of the weakest when it comes to using data and both regular and predictive metrics.
- Predict upcoming leadership needs and availability — develop a metric that can predict where and when there will be upcoming internal leadership shortages or surpluses. Also for succession purposes, be able to predict when your current key leaders will likely need to be replaced because of turnover, promotions, retirement, or because their performance has peaked.
- Identify who are currently ready for leadership positions — use data to accurately identify the current employees who are ready for either leadership development training or an actual leadership position.
- Identify those who will soon be ready for a promotion — delaying promotions can frustrate workers and increase turnover. Be proactive and develop a predictive metric for identifying who and when top individual employees will be ready for a promotion.
Training and Development Third-generation Predictive Analytics
The ability to learn rapidly has been identified by Google to be the No. 1 critical success factor across all jobs. Develop a predictive metric that can warn leaders long before employee learning issues become critical.
- A metric that projects the obsolescence of current skill sets — predictive metrics can forecast when currently valuable skills possessed by your employees will become obsolete.
- A metric that predicts future skill needs — in a rapidly-changing VUCA world, skills that are vital today will rapidly be replaced with new skill sets. Projecting those future skills will allow the training, recruiting, and retention functions to focus on those new skill sets. Forecasting the creation of new or redesigned jobs can also allow the training function time to create new skill development programs prior to when they are needed.
- A metric for identifying your firm’s learning leaders — because learning and innovation are so critical to organizational success, it’s important to use metrics to identify the employees in the organization that are creating and sharing new and original knowledge and ideas.
- Identify the factors that increase learning speed – use predictive metrics to identify where and when the overall organizational learning speed is likely to decline. Also use metrics to identify the factors that your best learners successfully use to increase what they learn and their learning speed.
Compensation and Benefits Third-generation Predictive Analytics
Because under-compensating is so damaging and over-compensating is so expensive, it’s important for this function to shift to a data-driven decision-making approach.
- Develop a metric to forecast upcoming excess overtime — predictive metrics can forecast when and where in the organization will excessive overtime usage likely occur. This metric can allow talent leaders to more accurately budget for overtime or to proactively reduce the use of overtime. Related metrics can identify the areas in the business where labor costs and staffing levels will soon likely become excessive.
- Project the obsolescence of benefits — predictive metrics can project when current employee benefits will cease to have a major impact on employee motivation, attraction, and retention.
Miscellaneous Third-generation Predictive Analytics
- Predict the talent actions of your competitors — in a competitive world, it’s important to anticipate and then counter the major talent actions of your competitors. Tracking software and predictive talent metrics can forecast future proactive competitor actions, as well as their likely reactions to your own firm’s talent management actions.
- Predict upcoming labor issues — statistics and predictive analytics can predict if, where, and when your organization may experience upcoming labor issues including union organizing, more grievances, and even strikes.
- Forecast the obsolescence of your organizational structure and building design — the most effective organizational structures eventually become obsolete as a corporation evolves. Talent management leaders should have metrics that enable them to forecast when and why the current organizational structure and org chart will become obsolete. A related metric should predict the likely date when the design of your physical facility becomes so obsolete that it reduces collaboration and productivity.
- Project upcoming organizational speed issues — in an organization where rapid product development is essential, the metric team must develop a process for forecasting when and in what areas will organizational speed need to be significantly increased.
- Forecast weak coordination between talent functions — whenever complicated interdependent work is handed-off between different internal functions or teams, the likelihood of an error or a slowdown increases dramatically. HR has a well-deserved reputation for having restrictive functional silos and poor coordination, so it’s important to track current and to project future talent handoff, coordination, and integration problems.
- Provide SWOT forecasts — external factors that are constantly monitored on the business side also need to be tracked and projected in talent management. So it is important to continually forecast SWOT factors (strengths, weaknesses, opportunities, and threats) that the firm will encounter in the talent management area.
The basic goal of predictive analytics is to get everyone in talent management to develop a forward-looking mindset, where everyone is focused on identifying and acting on upcoming problems. Consider the metrics that have been provided here — when acting in unison they could be a type of early warning system that scans the horizon for upcoming talent management problems and opportunities.
Even though the list of metrics that have been provided is long, realize that no individual firm is expected to implement more than a dozen of these forward-looking metrics. The final metric should be selected in conjunction with the COO and CFO, in order to ensure that they predict and forecast in the areas that are likely to have the largest business impact.