Chapter 06
Tracking outcomes after go-live
Why the business case should be revisited against actual outcomes, what to measure, and how the discipline of tracking changes future cases for the better.
01
Why outcomes are usually not tracked
Most business cases are written, approved, and forgotten. The team moves on. The actual outcomes are never measured against the predictions. This is a significant waste, and it is also why the next case tends to be easier to challenge — finance partners have no record of how the previous predictions held up, so every fresh case carries the discount applied to all unverified claims. The discipline is to track outcomes from go-live and report on them at agreed intervals. The point is not auditing the workflow; the point is improving the next case. A team that has measured one case against its predictions writes the next case with sharper inputs and earns a hearing on the basis of evidence rather than promise. The cost of tracking is small if the tracker is designed at the same time as the case. The cost of not tracking compounds across years. There is a structural reason tracking gets dropped. The team that built the case is rarely the team that owns the workflow on month thirteen. Sponsors rotate. Project directors finish the project. Finance partners move to other portfolios. Without explicit handover, the tracker dies quietly. The fix is procedural and small: at submission, the case names the tracker owner for years one and two, the refresh cadence, and the date by which the post-go-live review will be presented to the original approver or their successor. Naming the dates in the case itself converts the tracker from a good intention into a commitment the organisation can hold.
02
A short measurement plan, not a long one
Track the operational savings against the model. The recurring tasks named in the case, the time taken at six and twelve months, the loaded cost. Track the workflows that closed and the workflows that did not. Track the risk events that occurred and any in which the workflow contributed to the outcome — the time-aligned record that defused the dispute, the gate record that closed the delivery query, the safety loop that closed without a stand-down. The list is short on purpose. Tracking everything is the same as tracking nothing. Five well-chosen lines beat fifty observed irregularly. A simple shape that works: a one-page tracker with named owners, refreshed each quarter, that shows actual versus predicted on the five lines that mattered most in the original case. Refresh dates are calendared at go-live. The first refresh — typically at month six — is the one that catches early drift and lets the team intervene. A useful selection rule: the tracked lines are the same five that drove the headline total in the original case. Not the easiest five to measure. Not the five that look best. The five that mattered. If a line was 30 per cent of the headline total in the case, it is a tracked line in the post-go-live world. That alignment is what makes the tracker honest. It also forces the original case to be honest about which lines really mattered, which sharpens the case before it is even submitted, because the author knows in advance that those lines will be measured.
03
The post-go-live review at month twelve
Twelve months after go-live, run a structured review. The shape is small and predictable. Pull the original case. Pull the tracker. For each line, write three numbers: the predicted mid, the actual, and the percentage variance. For each variance above twenty per cent in either direction, write a one-line cause. For each line that fell short, write a one-line action. For each line that outperformed, write a one-line note on what to bring forward into the next case. The output is a one-page memo. It does not need to be circulated widely; it needs to exist and be referenced when the next case is built. A worked illustration of useful findings: a manual reporting line that landed at 80 per cent of predicted because the team kept a legacy report alive; a dispute avoidance line that landed at zero because no dispute occurred, which is the correct outcome and not a failure of the model. The discipline is to record both honestly. A second pass at month twenty-four is worth the small effort. By that point, the workflow has been through one full cycle of staff turnover and at least one budget cycle. The lines that look strong at month twelve sometimes erode by month twenty-four because someone re-introduced the legacy report or because the saving was reabsorbed into a new activity that nobody priced. The lines that look weak at month twelve sometimes recover by month twenty-four because the operational team has by then absorbed the new way of working. Both directions are useful inputs to the next case, and both are invisible without the second look.
04
How tracked outcomes feed the next case
A case that has been tracked produces inputs for the next case. Real proportions for replacement workflows. Real probabilities for risk events. Real attribution for capacity gains. Each tracked case improves the credibility of the next, both with finance and with the team that builds it. The compounding is real. After three tracked cases, the analyst no longer needs to reach for industry benchmarks for most of their inputs. They have their own evidence base, with their own bands, on their own projects. A finance partner reading the fourth case sees the citations refer to the organisation´s own prior cases and accepts the analyst´s judgement on the basis of demonstrated calibration. That is the durable benefit of tracking — not the audit of any single case, but the calibration of the people who build cases. Organisations that track for three years tend to write cases that look quite different from those written by organisations that have never tracked. The difference is visible in the bands. A small but high-leverage habit: maintain a single internal page that lists every prior case, the original predicted mid, and the tracked actual. The page is not glamorous. It is more useful than any benchmark database the organisation could buy externally, because it reflects the organisation´s own conditions, projects, and people. Authors building a new case start there. Reviewers verifying a new case start there. The page becomes the institutional memory of the practice, and it is the single artefact that distinguishes an organisation that builds credible cases from one that periodically writes them.
05
Failure modes in the tracking practice itself
The tracking practice has its own failure modes. First, drift to vanity metrics: the tracker that grows from five lines to thirty, half of them measuring usage rather than outcome. Resist the growth. The point is the predicted-versus-actual on the five lines that drove the original case. Second, late refreshes: the tracker that is updated whenever someone remembers, which is the same as never. Calendar the refreshes at go-live, with named owners and a fifteen-minute review at each. Third, defensive interpretation: the tracker that always concludes the case was right, regardless of variance. Variance in either direction is information. A line that came in at 130 per cent of predicted is as important to record as one that came in at 70 per cent, because the next case will be sharper for knowing both. The honest tracker is the one whose findings sometimes embarrass its authors. That is the one finance partners learn to trust, and that trust transfers to every case the team brings forward thereafter. A fourth pattern is the silent retirement of the tracker after a single negative finding. The team writes a case, the tracker shows the case fell short on two lines, and the practice quietly stops because the conversation became uncomfortable. The discipline is the opposite. The negative finding is the most valuable output the tracker has produced, because it is the input that will make the next case noticeably more accurate. A team that keeps tracking through one negative finding tends to keep tracking through all of them, and the resulting practice is the durable competitive advantage in how the organisation evaluates investments of this shape.
Practice
01. Design the five-line tracker you would attach to a current case at go-live. Name the lines, the owners, the refresh cadence, and the source for each actual.
Look for: Five lines that map directly to the largest items in the original case. Each line has a named owner, a quarterly refresh, and a single source for the actual figure. Anything more is too much.
02. Take an old case from your organisation. Estimate two of its predicted lines today against actuals. Note one finding you would carry forward into the next case.
Look for: Two predicted-versus-actual comparisons with rough percentage variances. One forward-looking finding stated in a single sentence — typically a sharper input range, a redefined risk event, or a more conservative attribution rate.
Checkpoint
For a case that went live last year, could you produce the actual versus predicted lines today, with named causes for any variance above twenty per cent?
Recommended reading