People analytics
Getting to workforce numbers that survive a board meeting.
Most workforce analytics projects fail before any analysis happens, because the underlying numbers cannot be agreed. If HR, Finance and Operations each produce a different headcount, no dashboard fixes that — it just renders the disagreement more attractively.
Latest articles
Articles on this topic are on the way. The guidance below is the short version in the meantime.
Start with agreement, not with a dashboard
Define the measures before you visualise them
Headcount is the obvious example: does it include casual workers, contractors, people on unpaid leave, people who left mid-month? Every function answers differently, and each answer is defensible. The number only becomes useful once the definition is written down.
Report from operational records, not a copy of them
The moment reporting runs on a separate extract, it has to be reconciled before anyone will stand behind it. Reporting from the records that payroll and attendance are approved against removes the reconciliation rather than scheduling it.
Prefer direction to precision
A turnover figure accurate to one decimal place is less useful than knowing it has moved four points in a year. Trends survive definitional argument in a way that point-in-time figures do not.
Attribute movement to a cause
“Workforce cost rose 4%” starts an argument. “Workforce cost rose 4%: 1.2 million from new starters, 0.8 million from overtime, offset by 0.6 million from leavers” ends it and starts a decision.
Be careful what you call prediction
Trend analysis on verified operational data is genuinely useful and widely available. Predictive attrition scoring is a different claim requiring different evidence, and it is worth asking any vendor — including us — to be precise about which one they are offering.
Ready to see People Analytics in action?
Tell us how your workforce operates and we will tailor the conversation around what you actually need to improve.