How to Know Which Projects Are Actually Making You Money — A Portfolio NPV Framework

Spend Is Easy to Track. Returns Are Not.

Every project has a budget. Most projects track actuals against that budget. Very few projects track actual returns against what was promised when the budget was approved. This gap — between the rigour applied to spend tracking and the rigour applied to return tracking — is where value quietly disappears.

In our work with a financial services firm, we found that the project portfolio had 23 active initiatives consuming a combined $12M in annual investment. Eleven of those projects had original business cases. Of the eleven, four had been tracked against their business case targets after launch. Of those four, two were on track. The other seven were simply running — no one knew whether they were creating or destroying value.

The Portfolio View That Changes Decisions

Individual project reviews don't reveal portfolio problems. You need to see all projects simultaneously, plotted on two dimensions that actually matter: expected value (NPV, risk-adjusted) and risk level. When you do this, patterns emerge that are invisible in project-by-project review.

The typical portfolio has four quadrants: high NPV, low risk (fund fully and protect), high NPV, high risk (fund selectively, manage actively), low NPV, low risk (often the legacy projects nobody terminates), and low NPV, high risk (terminate immediately). Most organisations have too much capital in the third and fourth quadrants because the politics of killing projects is harder than the analytics of identifying them.

The scatter plot view changes the conversation. When a CFO can show the executive team that four projects have negative expected NPV and moderate-to-high risk, and that reallocating their combined budget to the top two high-NPV projects would change the portfolio return profile materially, the political conversation shifts. Data doesn't eliminate politics, but it changes the terms.

Risk-Adjusted Returns: The Missing Variable

Most project business cases use a single-point NPV estimate: here are the projected benefits, here are the costs, here is the net present value. This is analytically incomplete because it ignores the variance around the estimate.

A project with an NPV of $3M that has a 90% probability of achieving its targets is fundamentally different from a project with an NPV of $3M that has a 50% probability. Risk-adjusting the NPV — probability-weighting the scenarios — gives you a better comparator for capital allocation. In practice, this requires building three scenarios (base, upside, downside) with probability weights, and calculating the expected value across them.

For a fintech business finance function we supported, this risk-adjusted view revealed that two projects the organisation considered low-risk were actually highly sensitive to regulatory approval timelines. Adjusting for that uncertainty dropped their expected NPV by 40%. They were still worth funding, but at a lower priority than the original business case suggested.