Skip to main content

Performance

Performance is last for a good reason: nothing in payroll or attendance depends on it, and it needs the people and the structure already in place. It is also the stage with the most moving parts, so build it in order and do not start from the cycle.

The order below is not negotiable. Each screen picks from the one before it, and a cycle built before the template has nothing to cut sheets from.

The order​

  1. KPI. Where a measurable number is read from.
  2. Evaluation Metric. The thing a person is measured on.
  3. Evaluation Template. The appraisal form itself.
  4. Performance Target Entity. Who is measured, and by whom.
  5. Evaluation Cycle. The period, and the run that cuts the sheets.
  6. Employee Goal. One person's sheet.
  7. Evaluation Tracker. Where the scoring happens.
  8. Calibration. Where the committee settles the numbers.
  9. Evaluation Increment Matrix. What a score band leads to.

KPI​

KPI is where a number the system already holds is made available as a measure. Sales won, leads converted, production output, tickets closed: each is a KPI source, and the screen says which records count, how a record is tied to a person, which figure on the record is the number, and the date it is counted by.

It also carries a base filter, so that only the records that qualify are counted, and there is a limit worth knowing before you design one: the filter is a plain condition and cannot itself go and look something up elsewhere. A measure that needs a lookup has to be expressed as a condition on the records themselves.

A client whose appraisal is entirely subjective does not need this screen at all. A client who wants a sales target measured automatically rather than typed in by the manager does.

Evaluation Metric​

An Evaluation Metric is one thing a person can be measured on. It has a code, a name and a category, and the fields that decide how it is scored:

  • Scoring Basis, which is the important one. Based on Target means the person has a number to hit and the score comes from how close he got. Score Band means there is no number and the evaluator picks from a scale.
  • Default Meet and Stretch, which are the expected level and the beyond-expected level. Which side the stretch sits on is what tells the system whether higher is better or lower is better, so a metric where a lower number is the good one is expressed by putting the stretch below the meet rather than by a separate setting.
  • Default Weight, the weight this metric usually carries on a form.
  • Auto Scoring and Max Value, for a metric bound to a KPI source rather than typed in.
  • Visibility, public or private, for a metric only the evaluator should see.
  • Key Result, for marking the metric as one of the handful that really matter.

Build the client's whole metric library here, once. A metric is reused across templates and across years, and the library is the thing that makes next year's appraisal cheap.

Performance Unit names the units metrics are expressed in. Organisation Goal is where the company's own goals are recorded and cascaded, with a weightage and a share, for a client running a balanced scorecard.

Evaluation Template​

An Evaluation Template is the appraisal form. It is built as sections, each holding items, so that a form can be thirty percent competencies and seventy percent objectives and the weight sits on the section rather than being repeated on every line.

The header settings are what make one template different from another:

  • Company Weightage and Self-Set Weightage, which divide the form between what the company sets and what the person sets for himself.
  • Min Objectives and Max Objectives, which stop a form with two objectives and a form with forty.
  • Allow Metrics From Employee and Allow Target From Employee, which decide whether the person may add his own measures and set his own numbers.
  • Allow Tracking-Only Metrics, for a measure that is watched but not scored.
  • Max Achievement Percent, which caps over-achievement. Without it a person who did three times his target can arithmetically dominate a whole form.

Most clients need more than one template, and the split is usually by level rather than by department. A management form weighted towards objectives and a non-management form weighted towards competencies is the normal shape, and the two weightings are not the same. Get them from the client in writing, because a form built on the wrong split has to be rebuilt rather than adjusted.

Performance Target Entity​

A Performance Target Entity is a group of people measured together: a sales team, a plant, a function, a grade band. It is what a cycle is run against.

Membership comes from the people, not from a rule on this screen. A person is in a target entity because his Employee Organization Assignment names it. That is worth being clear about with a client who expects to build the population here with conditions, and it is the right design, because it means a transfer moves a person's appraisal population on the same record that moves his department.

The screen also holds the business multiplier, which is how a whole group's target is scaled by a business factor, and the run coverage view, which is where you check that the population resolved to the people you expected.

The evaluator chain is named here too, and it is named again on the cycle. The cycle's copy is the one that is frozen when the cycle is activated, which means a later period can change the chain without disturbing a period already under way.

Evaluation Cycle​

An Evaluation Cycle is one performance period: which template, which target entities, and the dates.

Its halves are the periods inside the cycle, with their windows, and those dates are typed rather than worked out. A cycle does not divide its window by the number of steps in the chain; each stage's dates are stated. There is also a calibration window with its own start date and number of days.

A cycle moves through Draft, Sent to Users, Activated and Closed, and two actions on it are worth understanding separately:

  • Activate resolves the population and freezes the chain. It works out who the cycle covers. It does not cut any sheets.
  • Generate Sheets is what cuts the sheets, one per person per line, and it only creates the ones that are missing, so it can be run again after people have joined without disturbing the sheets already in use.

If a window has to move after sheets exist, applying the new dates pushes them onto the generated sheets, but a sheet that was dated by hand is left alone. And copying a chain from another cycle copies its structure, not its dates.

A client's first cycle is often loaded rather than generated, because the targets were agreed on paper before the system existed. That is a legitimate thing to do, and the cycles after it should be generated properly.

Employee Goal​

An Employee Goal is one person's sheet for one period. Its lines are the metrics he is measured on, each with its weight, its target, its meet and stretch levels, its achievement and its score, and a section it belongs to.

Three things on the sheet that confuse people:

  • A line may be tracking only, which means it is watched and reported and carries no weight.
  • A sheet can hold the same measure more than once: the scored line, plus a breakdown of it by quarter or by another dimension. Those breakdown lines overlap the scored line deliberately, so a sheet should never be read by adding every line together.
  • The Net Score on a line is the settled shared score. Each evaluator's own score is held separately, which is what lets a chain of evaluators disagree before a number is settled.

A sheet moves through Draft, Submitted, Approved, Returned, In Review, Rejected and Closed, and the goal status is what a manager chasing an appraisal should be looking at.

Evaluation Tracker​

Evaluation Tracker is where the scoring and the conversation happen inside the window. It is the screen a line manager actually opens, and the one to spend training time on, because everything above it is a consultant's work and this is the company's.

Rating Appeal is how a person contests a score, with the original score, the score he is asking for, his grounds and the reviewer's decision. Performance Improvement Plan is the other direction: a plan with objectives, how each is measured, the support required and an outcome.

Calibration​

Calibration is the committee's screen. It shows the actual score, the calibrated score and the movement between them, per manager and per level, so that two managers who score to different standards can be brought onto one.

Calibration is a settling of the numbers for the record. It is not itself a payout instruction.

Evaluation Increment Matrix​

An Evaluation Increment Matrix is a table of score bands, dated from an effective date and tied to a target entity. Each line is a band: a score range, a band name, a bonus percentage range, an increment percentage range, bonus and promotion eligibility, what the band leads to, and a maximum salary.

Two warnings, and both have cost clients time.

A matrix line is a band, not a person. The population comes from the target entity. A consultant who tries to list people here is on the wrong screen.

Not every column on the matrix drives something. The bonus bands are what the system reads and acts on. The increment percentage, the promotion eligibility, what the band leads to and the salary cap are recorded for the committee to act on, and a populated column is not by itself evidence that anything downstream uses it. Where a client is told a number will be applied automatically, check which one before promising it.

Performance Payout is where the bonus figures per person are produced from the matrix and, where the client wants it, carried into payroll as a named wage type. Salary Revision records the revision itself: the increase, the revised salary, and a new position or grade where the person has moved.

If something looks wrong​

  • "The cycle is activated and nobody has a sheet" - Activate resolves the population only. Generate the sheets.
  • "The population is empty" - nobody's Employee Organization Assignment names that target entity. Membership is on the assignment, not on a rule.
  • "The run coverage view is blank" - the target entity resolved to nobody, for the same reason.
  • "A sheet's total does not match the sum of its lines" - the sheet holds breakdown lines that overlap the scored line. Do not add every line together.
  • "A line scored zero and the achievement looks fine" - the line is tracking only, so it carries no weight by design.
  • "Moving the half's dates did not move one person's sheet" - that sheet was dated by hand, and a manual date is respected.
  • "The increment percentage on the matrix has not been applied anywhere" - check which columns the client is relying on before promising automatic application.