Happy Tuesday, Transformation Friends. Another week, another opportunity to go Beyond the Status Quo.
This spring, the Auditor General tabled five performance audits: pay modernization, fiscal initiatives with First Nations, accessibility in the public service, RCMP recruitment, and reforms to the international student program. In looking at them together, there’s a common element: in each, the part that could be counted, funded, and pointed to was completed (the legible part); while the structural, slow, and hard-to-measure part was deferred (the illegible part). And in each report, the deferred part was the one that really tells us if the change or modernization actually worked (the outcomes we were trying to achieve).
Today we’ll unpack this. I want to start with pay modernization, since it is deeply familiar to all of us and clearly shows the pattern, especially in contrast to what was reported as the root cause of the whole Phoenix debacle. Then we’ll trace the same split, between what gets counted and what gets deferred, across the other four reports. And we’ll close on three things we can do differently when we scope the next reform, so the structural half stays on the table.
Grab your morning coffee, and let’s get started.
The pay file shows this pattern clearly
What did the Auditor General say in her report on modernizing the pay system? The audit (Office of the Auditor General of Canada, 2026a) found the project to replace Phoenix with Dayforce in its planning phase, with a first wave of three organizations, the Canadian Nuclear Safety Commission, Shared Services Canada, and Public Services and Procurement Canada, set to onboard in 2027, against a preliminary cost estimate above $4.2 billion.
Under the system's rules, the report found slow progress. After Phoenix, the government commissioned its own study, by Goss Gilroy, into what went wrong, and a central lesson was that pay rules should be simplified and standardized before a new system goes in, so the system does not need expensive custom shapes built around rules nobody streamlined. The AG found this lesson documented and understood. It also found the Treasury Board Secretariat had made slow progress in acting on it, because simplification depends on consensus with unions, and by the end of the audit period, the Secretariat had brought unions only one of the simplification areas it had identified, with no agreement reached.
The report then describes what the department did instead. Rather than wait for simplified rules, PSPC is having three custom applications, so-called “cloud extensions,” built so Dayforce can process the unsimplified rules as they stand, at an estimated cost of at least $4 million a year. The AG noted this is the kind of customization that Phoenix’s own lessons-learned had warned against.
The audit reported a similar pattern throughout the rest of the file. PSPC made progress clearing the pay backlog for first-wave departments, but the AG found it had not fully assessed how that prioritization affects departments onboarding later. The report also examined how timeliness is measured, and this one is worth pausing on, because anyone who has waited on a pay file will recognize it. The AG found PSPC uses a “stop-the-clock” approach, which counts only the time advisors actively work on a file and excludes the time the file spends waiting. Measured that way, reported timelines look considerably better than the wait an employee actually lives through. On top of that, PSPC compressed the schedule by three years after the audit period, and the AG found no indicators in place to confirm the new system would actually lower the cost of processing pay, one of the main reasons given for building it.
What gets counted gets done
Here’s the underlying theme that I found most interesting because I see it all of the time: the parts of the work that we can “see,” let’s call them the “legible” parts, are what get done. While the parts that are much harder to see, let’s call them the “illegible” parts, slip.
Two ideas explain why.
The first is about what large organizations are able to see. In Seeing Like a State, the political scientist James Scott describes how big bureaucracies act on the version of reality they can standardize, count, and map (Scott, 1998). The upshot is straightforward. What can be counted gets managed, and what resists counting, local knowledge, relationships, and informal practice, tends to be neglected, because the organization has no clean way to track or reward it.
Here is how that plays out in our large modernizations. A system going live, hardware installed, click counts, dollars spent, time spent, all of it is legible: it can sit in a status report under a named owner. Simplifying rules, co-developing a policy with partners, understanding why an indicator moved, none of it is. Under pressure to show progress, an organization does the legible work first because that is the work it can actually see itself doing.
The second idea is about what happens once the legible work is underway. The sociologist Robert Merton described the process now known as goal displacement, in which the measurable proxy for success slides into becoming the goal itself, so people end up serving the indicator rather than the objective it was meant to represent (Merton, 1940). In the pay file, the objective is to ensure accurate, timely pay for public servants; the proxy is a system delivered on schedule. The gap between those two is exactly where the cloud extensions and the unsimplified rules sit. Once the proxy becomes the target, hitting it can feel like success while the real objective drifts.
Put those two together: An organization does the part of the reform it can see, then treats that part as the whole. With that in hand, the other four audits start to look an awful lot the same.
The same pattern, four more times
The fiscal initiatives with First Nations
The audit (Office of the Auditor General of Canada, 2026b) found the 10-year grants delivered as designed, with more than $6.5 billion provided to recipients by the end of 2024-25. The other two commitments from 2017 were not met: the Default Prevention and Management Policy was not replaced with a capacity-building policy, with the AG noting the department cited insufficient funding, and a mutual accountability framework was not developed. The AG also found the department did not assess whether the grants were closing the socio-economic gaps they exist to address.
A set sum paid on a schedule is straightforward to confirm and report, and that is the commitment that was kept. Co-developing a replacement policy and building mutual accountability with partners are structural and slow, and those are the commitments that stalled. The money moved. Whether it advanced self-determination or closed any gap is the question the department left unanswered.
Accessibility in the public service
The accessibility audit (Office of the Auditor General of Canada, 2026c) found that all seven organizations examined reached the 7% disability representation goal. It also found their processes could not handle the volume of accommodation requests, that data on those requests was collected inconsistently, and that where organizations did not track resolution times, requests took, on average, hundreds of days to resolve.
Reaching 7% is real progress, and a representation percentage is also as legible as a number gets, easy to track and to show moving. The daily experience of an employee waiting on an accommodation is far harder to see, and harder still to put in a status report. The countable goal was met while the slower work of fixing accommodation processes was left undone.
RCMP recruitment
The RCMP recruitment audit (Office of the Auditor General of Canada, 2026d) found that the force’s Flexible Posting Plan, which let cadets choose their first posting, drew thousands more applications. It also found the RCMP had narrowed its workforce planning in 2023 to front-line policing, no longer knew how many officers it needed across the whole force, and saw chronic vacancies in hard-to-staff divisions get worse.
This is perhaps the clearest version of the pattern, because here the legible number did not just sit on top of a stalled prerequisite; it actively pulled against it. Application volume is a clean, easy-to-understand metric. The kind that signals momentum to anyone reading a dashboard. Knowing the true size of the force’s need is harder analytical work with no satisfying number at the end of it (a lose-lose most likely), and that is the work the RCMP set aside. The result is an organization that looked busier and better resourced on the easy-to-see metric, while the structural understanding that was supposed to guide all that recruiting thinned out underneath it. A transformation can be moving quickly and losing ground at the same time, and the legible number will not tell you which is happening.
International student reforms
The international student reforms audit (Office of the Auditor General of Canada, 2026e) found that IRCC reduced the number of new study permits well past its target. It also found that the department could not explain why its approval rates fell so far below projections, did not adequately monitor the heavier-than-expected impact on smaller provinces, and made limited progress on the more difficult integrity and diversity work the reforms also called for.
The permit count is the Globe and Mail number. Understanding why approvals collapsed, tracking where the cuts landed hardest, and following through on integrity enforcement are ongoing, investigative, and far less legible. The number came down on schedule, but the department’s grasp of what it was doing did not keep pace.
Naming the half that does not show up in a status report
If the pattern is this consistent, the response has to be structural, because nobody in these audits was failing to try. What needs to change is how a reform gets scoped in the first place.
Three changes are worth building in.
The first is to name the structural work out loud, at the very start. A system launch arrives with a champion, a budget line, and a project name. The prerequisite underneath it, the rule simplification, the policy co-development, the process redesign, usually arrives with none of those, and work that nobody names is work that gets deferred by default. When a reform is being scoped, one question belongs on the table before anything else: what outcomes are we actually trying to achieve, and how will these show up in a status report? Write it down next to the deliverable, so keeping it or dropping it becomes a deliberate decision rather than a hidden one.
The second, following the first, is to decide how you will measure the real outcome before the project starts. This echoes a recommendation the AG made directly in the pay file, that PSPC develop indicators to confirm whether processing costs actually fall under Dayforce. The discipline generalizes to any reform. If the only thing instrumented at the outset is the proxy, the proxy is what the organization will manage, and it can be years before anyone can say whether the actual goal was met. Building the measure of the real objective into the plan from the start keeps success something to be earned rather than declared.
The third is to sequence and fund the structural work ahead of the visible deliverable. The pay file is the cautionary case: rule simplification was the documented prerequisite, and it still lost to the system schedule, because the schedule had funding and a deadline, while the simplification had neither. The AG made a similar observation about the stalled replacement policy for First Nations default management, which the department attributed to insufficient funding. A prerequisite that runs on goodwill and spare capacity, while the deliverable runs on a budget and a deadline, loses every time. If the structural work genuinely has to come first, the funding and schedule have to reflect that.
Wrap up
Across five audits in five unrelated corners of government, the same thing happened. The visible and fundable half of each reform moved forward while the structural half, the half that decided whether it worked, lagged behind or stalled. The encouraging part is that this is a predictable result of how large organizations see, which means we own it as a design choice and can change it the next time we scope a piece of work.
Three questions to carry into the next big project you’re closest to:
What is the part of this change that will never appear in a status report, and what happens to the goal if it is skipped?
Is the prerequisite work funded and sequenced before the visible deliverable, or is it expected to keep pace with only goodwill?
Does anyone measure the actual outcome, or only the proxy that is easy to report?
Until next time, stay curious and I’ll see you Beyond the Status Quo.
References
Merton, R.K. (1940) ‘Bureaucratic structure and personality’, Social Forces, 18(4), pp. 560–568.
Office of the Auditor General of Canada (2026a) Modernizing the Pay System. Ottawa: Office of the Auditor General of Canada.
Office of the Auditor General of Canada (2026b) New Fiscal Initiatives With First Nations. Ottawa: Office of the Auditor General of Canada.
Office of the Auditor General of Canada (2026c) Accessibility in the Public Service. Ottawa: Office of the Auditor General of Canada.
Office of the Auditor General of Canada (2026d) Recruiting for the Royal Canadian Mounted Police. Ottawa: Office of the Auditor General of Canada.
Office of the Auditor General of Canada (2026e) International Student Program Reforms. Ottawa: Office of the Auditor General of Canada.
Scott, J.C. (1998) Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. New Haven: Yale University Press.