What I see in most automation programs is that scaling stalls because the pilot runs on controlled conditions that do not exist in production. Once you move past the test workflow, you hit fragmented data sources, inconsistent document formats, missing metadata, and no defined ownership for exception handling. Without a governance layer, the automation cannot stabilize. In SMB environments, the constraints are usually data hygiene and limited architecture. You can automate fast, but the underlying systems are not standardized, so models and extraction pipelines drift quickly. Enterprises have the opposite problem. The data layer is cleaner and the infrastructure is mature, but expansion is slowed by compliance reviews, change management, and cross team integration. The organizations that scale well set clear data contracts, assign pipeline owners, and treat automation like a maintained service rather than a single workflow.
(1) The worst form of resistance to change is passive opposition. People show up to meetings and nod in agreement, but then keep using email chains and manual folders in their day-to-day work. Users often continue reentering data into spreadsheets because they feel it's the safest way to preserve the information. Employees who regularly work with documents tend to fear automation the most, worried that it might threaten their job security. (2) Those who resist automation usually see it as a punishment, believing it's meant to correct their inefficiencies. On the other hand, the strongest supporters are often overwhelmed with repetitive tasks and are desperate for any kind of relief. The employee spending their Fridays fixing others' mistakes can end up being your best internal advocate for automation. (3) After a successful pilot, we usually run into two big blockers: limited IT resources and fading executive interest. Often, automation systems get built in isolation, and connections between bots end up being patchwork rather than part of a sustainable platform. In one case, the legal department halted our document intake automation over concerns about how data was being managed. That delay stretched on for several months. (4) SMBs have to see ROI quickly because they don't have access to the kind of enterprise-level testing environments larger companies use. Operations leads at these companies often teach themselves tools like Zapier in the evenings because there's no centralized automation or IT team to support them. Ironically, that lack of resources can be a strength--decision-making is nimble, and politics don't get in the way. While enterprises focus on proof-of-concept evaluations, SMBs go straight to building real-world solutions that work.
In my experience leading digital automation projects, the biggest resistance comes from fear of job loss, unclear workflows, and teams who don't trust the data feeding the automation. What usually helps is mapping actual process time with tools like Power Automate Insights or UiPath Process Mining. It shows who is already optimizing their work and who is quietly bypassing systems. CoEs typically stall after pilots because documentation is outdated or ownership is unclear. When APIs change or volumes spike, no one wants to maintain the bot. SMBs move faster, but they lack governance. Enterprises move slower, but they need six layers of approval. A good benchmark is this: teams that invest 10-15% of automation budget into maintenance see projects sustain far beyond the pilot.
What I've seen is that resistance usually comes from two groups, the folks worried automation will expose messy processes, and the folks worried it'll replace their judgment. In document processing and data extraction, that shows up fast. If someone spends hours cleaning invoices or RFIs manually, they're either your strongest champion or your biggest blocker. You can usually spot the blockers early. They use phrases like 'that's just how we do it' and they avoid putting their workflow on paper. Pilots stall when companies automate a task but ignore the approvals, exceptions, and handoffs wrapped around it. That's where CoEs fall apart. SMBs feel this even more than enterprises. Smaller teams don't have room for complexity. Automation has to remove work immediately or people stop trusting it.