Measuring corporate responsibility by effect, not effort

Mistake: Counting activity as proof of benefit

Counted activity is the costliest substitute for evidence of benefit. Responsibility obligations concern effects on people outside a company; sessions, donations and policy coverage record what the company did.

Those quantities differ, and activity alone cannot establish an effect. Measures here concern payment, closures, communities, giving and governance; environmental accounting is separate.

Effort records are still useful as an account of spending. The error is presenting them as proof that spending worked, which is the standard shape of many responsibility reports. Pair each effort measure with the effect question beside it; answer in prose, including an admission when the answer is unknown. Giving may be measured as spend, but separately from effects a company causes.

Mistake: Leaving small-supplier payment timing out

Leaving payment timing unmeasured hides effects on smaller suppliers. Most large organizations pay materially later than their contracted terms, yet no single person owns the figure. A company can comply with its contracts and still cause a small supplier to fail, so an all-supplier average conceals exposure.

Fix: measure the gap between agreed payment terms and actual payment dates for the smallest suppliers, not across the whole ledger. This is the most neglected measure in most companies, affects real businesses now, and can be calculated from records already available in your ledger this week. Pair it with counting decisions against commercial recommendations.

Mistake: Reporting total harm without showing who bears it

Using only total harm conceals who carries it and whether anyone affected can seek a change.

Equal totals can mean an effect spread across a large population or the same burden concentrated on forty households near a site. Where recourse is absent, obligations may be met only if the company chooses to act.

Fix: for each effect, record what proportion falls on people least able to avoid it, then mark recourse yes or no per row, reviewed quarterly. State whether those bearing the effect have a route to change it, and name who would receive a complaint. Recourse is not a number; forcing it into one usually loses information.

Mistake: Treating commitments as completed when they are recorded

Treating a promise as completed because it was recorded leaves slippage invisible. Every internal or public commitment needs a date and current state. Commitments made in governance forums have a low completion rate precisely because nobody is measured on whether they are delivered.

Fix: record every commitment's date and status, plus how many slipped and who noticed; a register with no slipped items is not being maintained. One decisive test of whether the program exists is to count decisions against the commercial recommendation, with amounts and dates. A year with none means the function describes activity, and you can stop paying for the rest of the measurement.

Mistake: Choosing a denominator that flatters the result

Rates change with the denominator, and the base must match the obligation. Otherwise a company can make a result look favourable through outsourcing, a strong revenue year or differences in site size.

Per person affected is usually the honest basis for an effect on people, but requires knowing how many are affected.

Fix: state the denominator beside every figure. Per employee flatters outsourcing; per revenue unit flatters a good year; per site ignores site size. Use per person affected where the obligation concerns people. An unexplained base makes a rate less honest than a raw count because it presents apparent precision without its basis.

Mistake: Leaving the measurement boundary unstated

Unstated boundaries can decide the reported answer. A payment figure may include or exclude a subsidiary with its own ledger; a community measure may omit contractor traffic; a closure count may leave out workers employed by site suppliers. Each boundary choice is defensible, but each produces a different total.

Fix: write the boundary beside the measure and keep it stable. Changing a boundary and a comparison year in the same report, without showing both, is the most common way a responsibility figure misleads without anyone lying. State what is included so readers can understand what the number does and does not represent.

Mistake: Paying people for numbers that can be gamed

Once someone's pay depends on a measure, predictable gaming follows: the cheapest path to the number may avoid the outcome. A target on complaint volume can make complaining harder; meeting targets yield meetings, while supplier-coverage targets yield more contracts. The figure improves without proving the affected people are better off.

Fix: where targets are necessary, use a result you would still want if gaming were taken to its limit. Payment within terms is one example: the cheapest way to reach that target is to pay on time. Keep activity measures as records, not proof, and examine incentives before tying compensation to a responsibility figure.

Mistake: Publishing only the figures that look better

Reporting only figures that improve invites disbelief. A reader cannot judge a result if its basis is hidden, estimates lack uncertainty, or worsening rows disappear; responsibility reporting then appears selected rather than complete.

The quality of a claim depends on saying how the figure was produced and showing adverse movement, not just progress.

Fix: state whether each number is measured, estimated, modeled or reported by someone else, and show uncertainty for estimates. Report the rows that went backwards. A document with every line improving describes a selection, and experienced readers treat it accordingly.

Where claims touch supply chains or social performance, expectations for published statements are firming. The discipline expected of an annual modern slavery statement is a reasonable guide. Social and environmental performance claims are increasingly subject to rules on misleading statements that differ by market, so counsel reviews what goes out.

Mistake: Accepting shareholder duty as a reason not to measure

Treating shareholder duty as a bar to measuring or addressing effects accepts too much. What directors' duties require, and to whom, varies by jurisdiction and corporate form; the law is often more accommodating of long-term and non-shareholder considerations than the claim implies.

Fix: test the claim against law rather than accepting it. The Legal Information Institute's entry on fiduciary duty is a better starting point than the version repeated in meetings. Since directors' duties differ by jurisdiction and corporate form, a legal argument belongs with counsel, not simply whoever spoke last.

The wider account of what a company owes for the effects of ordinary operation is in corporate responsibility. The conduct beyond company walls that these measures reach is addressed by supplier ethics. Leadership ethics concerns who decides when a measure produces an inconvenient answer; sustainability ethics concerns honest claims about environmental performance.