Under-Promise, Over-Deliver? A Smarter Way to Commit

The slogan is catchy, but incomplete

"Under-promise and over-deliver" is one of the most repeated lines in business. It sounds like a formula for happy clients: set low expectations, then pleasantly surprise them. In practice, it can cause problems. If you consistently quote a date that is far too generous, a client may not trust your estimates at all. Your team may learn to pad everything until nothing is useful. And a customer who expected a fast answer may still be disappointed by the time you finally reach the "pleasant surprise."

The better goal is not to promise less. It is to promise honestly. A realistic commitment, backed by a sensible buffer and an early warning system, earns more trust than a slogan ever will.

Why our estimates run short

The strongest evidence on this comes from psychology. In 1994, Roger Buehler, Dale Griffin and Michael Ross published a series of studies on what they called the planning fallacy. Their participants consistently predicted that tasks would be finished sooner than they were. University students expected to finish honours theses in an average of 33.9 days. The real average was 55.5 days, which is about 64 percent longer than predicted. Only around 30 percent of students finished within their own best estimate.

What makes the study especially useful is the fix it tested. When participants were asked to recall how similar past projects had gone and to link them to the current task, their optimism dropped away, and about 60 percent finished on time, compared with about 29 percent in the control condition. The problem was not a lack of ability. It was a habit of planning from the inside, focusing on the specific task and ignoring the record of similar work.

Bent Flyvbjerg, who has studied large infrastructure projects for many years, describes the same mechanism in a different setting. He argues that planners often take an "inside view" of their project, and that the remedy is to take an "outside view" by looking at how comparable projects actually turned out. His paper in the Project Management Journal describes this approach, known as reference class forecasting, as a way to bypass optimism bias.

Use your own history as the reference class

You do not need a major research team to apply this idea. Reference class forecasting, in a small-business form, has three steps.

First, identify a group of past jobs that are genuinely similar to the one you are estimating. Group by type of work, size and client, not by how you remember the job.

Second, look at the actual durations or costs of those jobs, including the awkward ones. Do not throw out the outliers without understanding them. The delay that came from a late client approval may happen again.

Third, place the new job within that range. If your last six website builds took between three and six weeks, a seven-week promise is an unlikely estimate, even if the new project feels simpler.

Keeping a simple log of each finished job makes this easy. Record the original estimate, the actual outcome and one note about what caused the difference. After a year, you will have a reference class that is far more reliable than your instincts.

Build buffers on purpose

A buffer is a deliberate margin between the promise and the expected outcome. It is not padding for the sake of it. It accounts for the normal uncertainty of work that depends on other people, weather, equipment or approvals.

One practical method is to think in three numbers: the best case, the likely case and the difficult case. Commit to the likely case plus a buffer that covers most of the gap to the difficult case. Buehler's data suggests that the likely case is often too short, so it is wise to begin with a realistic baseline before adding anything.

Buffers work best when they are visible. Put them in the plan as named tasks, such as "review and feedback" or "client content delay allowance," rather than hiding them inside the main estimate. That way, you can explain them to a client, and you can release them if they are not needed.

Say no in a useful way

Saying no is difficult because most people feel that every yes is income, reputation or career progress. But an overloaded schedule produces missed deadlines, and missed deadlines damage trust far more than a clear limit does.

Sheeran and Webb's 2016 review of the intention-behaviour gap helps explain why this matters. They found that goals which are over-optimistic, or which rest on social pressure rather than personal conviction, are less likely to be realised. A promise made only to avoid an awkward conversation is one of the least reliable promises you can make.

When you must decline all or part of a request, try a structured response. Acknowledge the request. State what you can do and by when. Explain what would need to change for the larger scope. For example: "I can deliver the first two modules by the 15th. The full package would need until the end of the month, or we could move the reporting section to next quarter. Which would work better for you?"

This kind of reply keeps the relationship open and gives the other person a real choice. It also avoids the silent over-commitment that leads to a missed date.

Renegotiate early, not at the deadline

The most damaging moment to change a commitment is the day it is due. By then, the other person has already planned around it, and any change feels like a failure. The best time to renegotiate is the moment you first see the risk.

Watch for early signals: a dependency that has not arrived, a task that is taking longer than expected in its first few days, or a change request that affects the schedule. When you see one, send a short update. Name the risk, say what you are doing about it, and propose a revised date or a reduced scope. Early notice gives the other person room to adjust, which is often the most valuable thing you can offer.

Gollwitzer and Sheeran's research on implementation intentions points to the value of specific triggers. You could write a rule such as: "If a task is more than two days behind its plan, then I will send an update to the client that day." Specific triggers make early action more likely, because the decision is already made.

A simple commitment formula

Putting these ideas together gives a practical sequence. Estimate the job from your history. Identify the likely case and add a buffer. Write the commitment with a date and a clear scope. Set one early-warning trigger. Keep a record of what happens, and review it after each job.

Used this way, realistic commitments become a reputation asset rather than a constant source of stress. Clients learn that your dates mean something, and your team learns that estimates are a tool for planning rather than a test of nerve.

A tool like Promikept can help keep this honest. Recording each commitment with its date, and tracking progress against it, gives you a clear view of your own reliability. Over time, the record shows whether your estimates are improving, which is the most useful feedback any planner can have.

Limits to keep in mind

Research on the planning fallacy describes a general tendency, not a fixed rule for every project. Some work is genuinely uncertain, and some estimates will always be wrong. The aim is not to become perfect, but to become more accurate, more transparent and quicker to adjust when the facts change.