B2B social selling ROI with a counterfactual, contribution margin and honest uncertainty

B2B social selling return on investment is a ratio built from incremental benefit and incremental cost. It is not the revenue attached to contacts who saw a post. An English business should define the counterfactual, margin and cost boundary before calculating a percentage.

What to take away

Define the decision and outcome

State the programme, comparison and measurement window. Examples of outcomes include qualified progression, incremental contribution or a measurable reduction in staff time. Choose one primary decision outcome and preserve operational indicators separately.

Write the causal path from activity to outcome. A seller may research an account, publish useful material, begin a conversation, open an opportunity and close a sale. CRM tags can show that sequence, but several other contacts and market changes may have contributed.

Measure the complete incremental cost

Include resources that change because of the programme: supplier charges, implementation, media, loaded staff time, content, governance, analytics, risk response and exit. Keep recoverable VAT and cash timing visible, with treatment confirmed by finance. Exclude sunk costs from the forward choice, while recording opportunity cost where existing staff or systems are diverted.

Use the same period for benefits and costs. Annualising an early benefit while retaining only a short proof cost makes the ratio misleading.

Convert revenue to contribution

Use a finance-approved contribution margin that matches the products, customers and period in scope.

incremental contribution = incremental recognised revenue x contribution margin

Then calculate:

ROI percentage = (incremental contribution + other incremental monetised benefits - incremental cost) / incremental cost x 100

Do not mix gross margin, contribution margin and cash collection. Show the definition beside the input. Tax effects should be reviewed separately rather than embedded through an assumed generic rate.

Establish what would have happened anyway

A credible counterfactual may use phased rollout, random allocation or a carefully matched comparison group. The 2026 Magenta Book is government evaluation guidance, not a rule for companies, but it explains why comparison with an unaffected group or period matters for attributing change. The design must address differences between groups, contamination and other concurrent activity.

Where a strong design is not feasible, estimate a range for the incremental share and call the output a scenario. For example:

scenario incremental revenue = attributed revenue x assumed incremental share

The assumed share must be labelled, sourced if possible and varied. It must not quietly become an observed fact.

Treat time and control benefits carefully

Released time is a cash benefit only if expenditure falls. If staff redirect it, value the alternative work only where there is defensible evidence. Otherwise report hours released as an operational result beside, not inside, financial ROI.

Reduced risk or stronger approval can be valuable without a credible monetary amount. Avoid multiplying a possible incident cost by an invented probability. Record the control outcome and any evidence separately.

Present uncertainty, not one impressive figure

Run lower, working and upper cases for incremental share, contribution margin, adoption, cost and delay. State which inputs are observed, quoted, calculated or assumed. The 2026 Green Book warns public-sector appraisers about optimism in costs, duration and benefits; a commercial reviewer can use the same challenge without importing government appraisal rules.

Report payback only if cash timing is modelled, and show the first period cumulative net cash becomes non-negative. A return calculation is decision-ready when finance can reproduce it and a reader can see how much depends on causal evidence rather than attribution labels.

Before you act

Common questions

What costs should be included in incremental cost?

Include supplier charges, implementation, media, loaded staff time, content, governance, analytics, risk response and exit. Keep recoverable VAT and cash timing visible, with treatment confirmed by finance. Exclude sunk costs from the forward choice, but record opportunity cost where existing staff or systems are diverted.

How should released staff time be treated?

Released time is a cash benefit only if expenditure falls. If staff redirect it, value the alternative work only where there is defensible evidence. Otherwise report hours released as an operational result beside, not inside, financial ROI.

Why does the article recommend a counterfactual?

A credible counterfactual may use phased rollout, random allocation or a matched comparison group. It explains why comparison with an unaffected group or period matters for attributing change. The design must address differences between groups, contamination and other concurrent activity.