Some experiences start badly and never quite recover — even when things genuinely improve. ACW is a framework for understanding why.
Most models of satisfaction assume a reasonably level playing field across time. If something goes wrong early but gets fixed later, the overall experience should average out. A bad start to a new job followed by five good months should feel, net, like a good first year. A difficult opening to a client relationship followed by consistent delivery should feel like a successful engagement. Early miscommunication followed by clear, responsive care should feel like a good doctor.
Except it often doesn’t. And the reason is that we don’t experience time equally.
The mental budget problem
When we enter a new experience — a university subject, a new job, a financial product, a course of medical treatment — we arrive with a mental budget. Not a financial one, but a cognitive one. We have a set of expectations about what this experience will deliver, and we allocate attention, effort, and goodwill accordingly. This is mental accounting in action: we don’t process experiences as a single running total, we open a ledger for them.
The problem is that debits and credits aren’t treated symmetrically. When early delivery falls short of expectation, the shortfall is processed as a loss — and under prospect theory, losses feel significantly larger than equivalent gains. An experience that lands 20% below expectation in the first week doesn’t get cancelled out by one that lands 20% above it in week three. The loss has already been logged. The gain goes into a different mental account.
This is the core of Asymmetrical Consequence Weighting — ACW. Early unmet expectations are overweighted. Later improvements are underweighted. The gap between them compounds quietly until dissatisfaction becomes the default frame through which everything else is interpreted.
Why timing is everything
ACW isn’t just about the size of the gap — it’s about when it occurs. An unmet expectation at the start of an experience carries more psychological weight than the same unmet expectation midway through. Early on, the ledger is almost empty. The first entry sets the tone for everything that follows.
This has implications well beyond any single context. Consider:
A new employee whose first week involves unclear role expectations and no structured introduction. By the time their manager gets around to a proper onboarding conversation in week four, the cognitive account is already in deficit. The improvement is real, but it’s fighting against an established loss frame.
A patient who receives confusing information about their treatment timeline at the first appointment. Later clarity from the same clinician is processed with scepticism rather than relief — because the mental model of “this system doesn’t communicate well” is already in place.
A government service that fails to acknowledge a new customer’s first contact. Subsequent responsiveness is filtered through the expectation of unreliability that the first interaction created.
In each case, the system eventually delivers. But the ACW distortion means that delivery lands differently depending on where in the sequence it occurs.
What ACW is not
ACW is not the same as first impressions mattering — that’s well established and not particularly novel. It’s specifically about the asymmetry in how we process shortfalls versus recoveries across time, and why the mathematics of averaging doesn’t apply to human experience the way we assume it does.
It’s also not a counsel of despair. Knowing that early misalignments carry outsized weight is precisely what makes them addressable. If you can surface the gap early enough — before the loss becomes entrenched — you can intervene at the point where recalibration is still low-cost and high-yield.
The question is whether systems are designed to surface those gaps in time. Most aren’t.
References: Thaler (1999); Kahneman & Tversky (1979); Gabaix & Laibson (2006); Chater & Loewenstein (2023)
The follow-up piece explores what an ACW-informed intervention looks like in practice, using higher education as the worked example: When the Feedback Arrives Too Late: Designing for ACW in Higher Education in Kynd Policy.
ThinKynd 