Measuring value from Salesforce: a KPI framework

Ask a leadership team what their investment in Salesforce has returned and the answers tend to fall into two groups. Some point to activity: users licensed, features live, agents deployed. Others point to a general sense that things are better. Neither is a measure of value.

This is rarely for want of data. The platform produces more reports than anyone can read. What is missing is structure: an agreed set of measures, linked to each other, with a person answerable for each. The framework below is the one we use with executive teams. It has four layers.

The four layers

1. Business outcomes. The results the board cares about: revenue growth, cost to serve, customer retention, margin. These are the reason for the investment. They move slowly and are influenced by many things besides technology.

2. Process measures. How well the work itself performs: sales cycle length, win rate, first contact resolution, time to resolve a case, campaign conversion. These move sooner and can be tied more directly to a change in how people or agents work.

3. Adoption measures. Whether the people and agents are doing the work in the intended way: active use, data completeness, share of enquiries handled by an agent, rate of handoff to a person. These are leading indicators. If adoption is poor, the process measures will not move.

4. Platform health. Whether the foundation is sound: data quality, system availability, integration reliability, release stability, the backlog of unresolved defects. Poor health quietly undermines every layer above it.

The value of the layers is in the links between them. Each business outcome should trace down to the process measures that drive it, the adoption measures that enable those, and the health measures that support them. When an outcome is off target, the chain shows where to look.

A worked example

Suppose the outcome is lower cost to serve in a service operation running on Agentforce Service, with an agent handling routine enquiries.

• Outcome: cost per resolved enquiry.
• Process: share of enquiries resolved without a handoff, average time to resolution, repeat contact rate.
• Adoption: share of eligible enquiries reaching the agent, rate at which advisers accept or correct the agent's suggestions.
• Health: accuracy and freshness of the knowledge content, completeness of customer records.

If cost per enquiry is not falling, the team reads down the chain. Perhaps the agent resolves well but too few enquiries reach it. Perhaps handoffs are high because the knowledge content is out of date. Each diagnosis leads to a different action. Without the chain, the conversation stays at "the agent is not delivering".

Baselines come first

A measure without a baseline is a number without meaning. Before any change goes live, record where each measure stands, over a period long enough to smooth out seasonal swings.

Three points of discipline matter here. Define each measure precisely, in writing, so it is calculated the same way before and after. Record the baseline before the change, because it cannot be reconstructed with confidence afterwards. And note what else is changing at the same time, such as a pricing change or a reorganisation, so that later improvements are not credited to the wrong cause.

Every benefit needs an owner

Benefits that belong to everyone are delivered by no one. Each outcome and process measure should have a named benefit owner: a business leader who has agreed the target, has the authority to act on it, and will report against it.

The owner is not the programme manager and not the technology lead. If the benefit is a shorter sales cycle, the owner runs sales. The programme delivers a capability. The owner turns it into a result.

A benefit without a named owner is a forecast. With one, it is a commitment.

A review cadence that leads to decisions

Measures only matter if someone looks at them and acts. We suggest three rhythms.

1. Monthly, operational. Adoption and health measures, reviewed by the teams running the work. The aim is quick correction.
2. Quarterly, executive. Process measures and progress towards outcomes, reviewed by benefit owners together. The aim is to decide where to invest further, where to adjust and where to stop.
3. Annually, strategic. Business outcomes against the original case. The aim is to test whether the priorities still hold.

Analytics tools such as Tableau can present all four layers in one view. The tool matters less than the habit. A simple scorecard reviewed every quarter by people with authority will outperform an elegant dashboard that nobody opens.

One caution: the most common mistake is measuring too much. Three or four measures per layer is plenty. A scorecard that fits on one page gets read.

Our approach is to help leadership teams agree that page early, with baselines and owners attached, so that the question of value has an answer ready whenever it is asked.