0%
Forward Deployed Engineering
The Role
Discovery and Scoping
Customer Data and Systems
Building in the Field
Measuring Deployment Success
A deployment is not successful because it went live, because the demo impressed the sponsor or because the FDE team worked hard. It is successful when the customer gets the outcome they were promised, keeps using the system and would buy again. Measuring that is harder than it sounds, because outcomes take time to appear, many factors affect them and different people judge success differently. This lesson covers what success means, how to choose leading and lagging measures, how to measure adoption and value, how to report to customers and executives and how to measure the FDE practice itself.
Outcomes, not outputs
Outputs are what the team produced: a pipeline, a model, a dashboard, an integration. Outcomes are what changed for the customer: claims processed faster, fewer stockouts, analysts saving hours each week, fewer fraud losses. Outputs are necessary, but customers buy outcomes. Measure success in the outcomes agreed during scoping, using the success criteria the sponsor signed.
Success for whom
Different people judge success differently. The sponsor cares about the business outcome and the return on the investment. Users care whether the system makes their work easier. The customer IT team cares about reliability, security and support load. The vendor cares about renewal, expansion, references and cost to deliver. A deployment that satisfies only one group is fragile, so track measures that matter to each.
The success criteria as anchor
The success criteria agreed in scoping, as an earlier lesson described, are the anchor for measurement. They name the measures, baselines, targets, time frames and who checks them. Revisit them at kickoff, at go live and at each steering meeting. If the criteria need to change because the business changed, agree the change in writing rather than quietly measuring something else.
Baselines
Every outcome needs a baseline: how things were before the deployment. Without one, nobody can say whether a number improved. Capture baselines early, before the system changes anything, using the same definitions you will use later. If historical data is weak, measure for a few weeks before go live. A missing baseline is one of the most common reasons a successful deployment cannot prove its value.
Attribution
Outcomes are affected by many things besides the deployment: seasonality, other projects, staffing changes and market conditions. Be careful about claiming all of an improvement. Where possible, compare with a group that did not use the system, such as another region or team, or compare periods with similar conditions. Be honest about uncertainty. A modest, well supported claim is more credible than a large one that falls apart under questioning.
Time to value
Time to value is the time from contract signing to the customer seeing the first meaningful outcome. It is one of the most important measures for an FDE team, because customers judge the investment early and long delays put renewals at risk. Track it for every deployment and look for the steps that slow it down, such as access, data readiness or security reviews.
Leading signs of trouble
Some signs predict a struggling deployment long before outcomes are measured: users who stop logging in, a sponsor who misses steering meetings, data problems that keep returning, a long list of open requests or a customer team that has not taken ownership of anything. Watch for these and act early, as the stakeholder and monitoring lessons described.
Measuring cost to deliver
Success also has a cost side. Track FDE hours, infrastructure spend and support load for each deployment. A deployment that meets its outcomes but costs three times its contract value is not a business success for the vendor, however happy the customer is. Knowing the cost to deliver helps the team price future deals and decide where reuse would save most.
Defining measures precisely
Vague measures cause arguments later. Define each measure precisely: what is counted, from which system, over what period, which records are excluded and who calculates it. Faster claims processing could mean median time from submission to decision, time from assignment to decision or the share of claims decided within five days. Agree one definition with the customer, write it down and use it every time.
Measuring in the customer systems
Where possible, measure outcomes in the systems the customer already trusts, such as their operational database, finance reports or service desk tool, rather than only in the vendor system. Customers believe numbers from their own systems more readily, and those systems often capture what happens after the deployment hands work on. Agree access to these numbers early, as part of the success criteria.
Short and long horizons
Some outcomes appear quickly, such as hours saved by analysts in the first month. Others take quarters, such as lower fraud losses or better customer retention. Track both: short horizon measures show early progress and keep the sponsor confident, while long horizon measures prove the real business value. Make sure everyone understands which measures should move when.
When the goal changes
Businesses change during long deployments. A new executive may have different priorities, a merger may change processes or market conditions may make the original goal less important. When this happens, revisit the success criteria openly with the sponsor, agree new measures and baselines and record why they changed. Measuring against an outdated goal satisfies nobody.
Qualitative evidence
Not every success shows up in numbers. User quotes, stories of problems avoided, faster decisions in a crisis and the customer team taking ownership are real evidence too. Collect them alongside the numbers, with permission, and use them to explain what the numbers mean. Executives often remember one good story longer than a chart.
Failure is information
Some deployments will not meet their goals. Measure them honestly anyway. Knowing why a deployment failed, whether the problem was wrong, the data was not ready or adoption never happened, helps the team avoid the same failure in the next one. Hiding failures removes the chance to learn from them.
Who owns measurement
Measurement needs an owner on both sides. On the vendor side, the FDE lead usually owns collecting the numbers, keeping definitions consistent and preparing reports. On the customer side, someone in the business, often reporting to the sponsor, should own the outcome numbers and confirm them. Shared ownership prevents disputes later about whose numbers are right, and it keeps the customer invested in the result.
Measuring without slowing work
Measurement should not become a burden. Automate data collection where possible, reuse dashboards from the toolkit and avoid asking users to fill in long surveys. Pick a small number of measures that matter, rather than tracking everything. If measurement takes more than a few hours a month for an established deployment, simplify it.
Success at the end
At the end of a deployment phase, write a short success summary: the agreed outcomes, the baselines, the results, the evidence and what comes next. Have the sponsor confirm it. This document supports renewal, serves as the basis for a reference or case study with permission and gives the next FDE team a clear starting point.
Keep it to one or two pages, with the key numbers at the top.
Measure success in the outcomes agreed in the success criteria, not in outputs, and track what matters to the sponsor, users, IT and the vendor. Capture baselines before the system changes anything, be honest about attribution, track time to value for every deployment, watch for early signs of trouble and measure the cost to deliver.