Change agent networks (CANs) are a popular mechanism for scaling adoption across complex organizations. In theory, they extend the reach of a small transformation team by deputizing credible insiders to translate strategy into local behavior change. In practice, however, many networks struggle to show measurable impact. They are busy, visible, and well-intended – but not very effective. Here are my thoughts on why CANs frequently falter.

The Ambiguous Charter

Most networks launch with a fuzzy mandate: “be a champion,” “spread the word,” “collect feedback.” That language invites activity, not outcomes. Without 3-5 concrete, observable behaviors per role (who does what, where, how often) tied to a small set of operational metrics, CANs drift into communications clubs and morale events. People feel busy and leaders see activity, but there’s no explicit line-of-sight from effort to effect.

Antidote: Charter the network for behaviors, not “vibes.” Replace fuzzy mandates with three to five priority behaviors per role, each explicitly tied to a single operational KPI. Publish baselines and targets so there’s a visible line from effort to effect. When the purpose is concrete and measurable, activity naturally organizes around outcomes rather than exposure.

Selection Bias

Volunteer sign-ups and manager nominations skew toward extroverts, enthusiasts, suck-ups, and the already-convinced. You end up with dense clusters inside functions (ops with ops, sales with sales) and thin or non-existent bridges between them. The very people who carry influence (the skeptics, brokers, and credible doers at friction points) are underrepresented. The network looks big on a slide but lacks the “wide bridges” where adoption actually stalls.

Antidote: Select for influence and coverage, not enthusiasm alone. Use lightweight ONA and peer nominations to surface brokers, credible doers, and even constructive skeptics. The people who sit on real handoffs can bridge Sales-Ops, Branch-Central, and Finance-Operations. Shoot for wide bridges across the workflow rather than dense clusters inside one function.

Capacity and Authority

Typically, change agent work is layered on top of the day job with no time release, budget, or decision rights. Agents can surface issues but can’t remove blockers. Escalations float upward into “air traffic control,” where they vanish into the ether. After a few cycles, agents stop raising impediments; the network becomes a broadcast channel instead of a learning and feedback system.

Antidote: Provide capacity, authority, and an unblock lane. Change agent work must come with 10–20% time release, a modest discretionary budget, and a clear decision path to remove impediments. Stand up a visible impediment board with owners and ETAs, and enforce a 72-hour “unblock” SLA so issues don’t disappear into air traffic control.

Gravity of the Status Quo

Local leaders control priorities, schedule, and psychological safety. If their incentives point to short-term throughput, practice time dies on the calendar. Agents push; managers pull the other way. In the tug-of-war between “hit today’s numbers” and “learn tomorrow’s way,” today always wins.

Antidote: Build an operating rhythm that converts learning into change. Hold a focused 30-minute weekly huddle to surface the top three blockers and top three learnings, and commit to the next two tests: 1) Every two weeks, showcase a concrete before/after with evidence, and 2) Each month, bring sponsors into the room to remove systemic blockers in real time.

Measurement Theater

Many CAN scorecards track outputs (sessions held, emails sent, attendees) rather than outcome-relevant behaviors and operational effects. Baselines are absent, leading indicators are undefined, and attribution by site or cohort is missing. Without a counterfactual (“what would have happened otherwise?”), leaders can neither claim success nor learn from failure.

Antidote: Instrument outcomes so you can separate signal from noise. Track leading behaviors, like percent of eligible transactions using the new workflow, required-field completeness, rework prevented; and pair them with lagging KPIs such as cycle time, error rate, and cost per transaction. Attribute results by site or cohort and publish a simple red/amber/green scoreboard.

A successful change agent network isn’t a poster, a kickoff, or a feel-good club—it’s a working system that costs real time and energy. It needs dedicated resources so agents can do the work instead of squeezing it between tasks: time release, a modest budget, and a clear path to remove blockers. It also requires the right participants: credible doers, brokers at the messy handoffs, and constructive skeptics, not just the most enthusiastic volunteers or (worse) the most available.

And more than anything, it demands hard work: the discipline to surface and fix impediments, steady practice to turn old into new, and the humility to measure what matters and adjust. When you combine resourced capacity, the right network topology, and persistent effort, CANs stop being theater. They become the mechanism that turns strategy into everyday behavior — one resolved blocker, one practiced workflow, one measurable improvement at a time. If you’re not prepared to fund it, properly staff it, and sweat the details, don’t launch a CAN. If you are, you can build a lean, sharp network that reliably moves the needle.