The mid-tier AI adoption threshold
In mid-tier organisations, the daily pressure of business-as-usual sets a payoff threshold that typical AI gains do not clear, so adoption stalls even when tools and training are in place.
AI adoption stories at the top end of the market often read as success. The large firm has resources to invest in change, dedicated adoption teams, and enough slack in the operating model to absorb disruption while a new capability is stood up. Mid-tier organisations, where most of the economy actually sits, do not have any of those conditions. The pattern that emerges from work with them is consistent enough to name.
The threshold
In a mid-tier firm, operational capacity is usually consumed. Staff are firefighting daily workload, partners or managers are billable, and the discretionary attention available for anything new is small. Against that baseline, an AI tool that might save 15% on some class of task is not a compelling value proposition. The time cost of adopting it — learning it, integrating it, working out when to trust it — falls immediately. The payoff accrues slowly and unevenly. Until the payoff clears the cost threshold, the tool gets used when it is easy and abandoned when it is not.
This is not a story about staff being unwilling to change. It is a story about the payoff being too small, relative to the pressure on staff time, to justify the disruption. The threshold is set by business-as-usual, and business-as-usual is heavier in mid-tier firms than most AI-adoption frameworks assume.
Why the usual responses do not clear the threshold
Three conventional responses fail to shift the threshold.
Training lowers the per-user cost of adoption slightly but leaves the payoff calculus unchanged. A trained user who still sees small gains still stops using the tool.
Mandates force short-term usage but do not produce durable adoption. The work routes back around the tool as soon as supervision relaxes.
Better tools sometimes raise the payoff, but the ceiling on what any tool can deliver is set by the context it has access to (see Useful AI is a context problem). Without structural work on the underlying knowledge base, a better tool produces the same kind of plateau at a slightly higher level.
What does shift the threshold
What moves adoption in mid-tier firms is raising the payoff rather than lowering the cost. That typically means fewer, deeper use cases where AI delivers disproportionate value — usually because the context it is given is unusually good. A firm that can point to two or three high-payoff uses where AI reliably replaces significant work will build adoption habits that extend outwards. A firm that has twenty low-payoff uses will see the BAU threshold swallow all of them.
The pattern argues for sequencing: concentrate on the tasks where AI has the best available context and the biggest delta against status quo, and build momentum there, rather than rolling tools out firm-wide and hoping adoption emerges from exposure. The alternative is the plateau described in A tools-first AI rollout that plateaued.