Remember your last budget meeting? If you’re like many CFOs, it was a long, exhausting process that was not particularly effective. As the presenters showed you their plans, you probably challenged every number and explored every assumption. In the end, you may have raised their targets a little, but if you’re honest, you have to admit that each unit’s final budget for next year looked a lot like the one its managers proposed at the beginning of the budget process — which in turn wasn’t much different from the latest forecast for this year or actual performance for the previous year.
We hear variations of this story time and again at companies across industries and geographies, and executives wonder why the process unfolds like this and what they could change about it. The short answer, in many cases, is that you’ve been anchored. Anchoring is a well-known psychological bias whereby one piece of information sticks in your mind and influences your interpretation of subsequent information, even if you’re unaware of it. In the case of budgeting, getting stuck in the same numbers from year to year is almost unavoidable. But there are ways to orient the process to challenge the status quo or default allocation — and they work with other target-setting or resource-allocation processes as well.
Many studies have shown that even obviously irrelevant numbers influence estimates. In one, for example, respondents were asked to estimate the age of Mahatma Gandhi at death. Before they had to answer, half were asked if Gandhi was younger or older than 9 when he died; the other half were asked if he was younger or older than 140. Both questions are absurd and their answers obvious, so you’d think the respondents would have disregarded them entirely. Yet the first group, on average, gave estimates of Gandhi’s age at death that were 17 years younger than the second group. If people can be anchored by such obviously irrelevant inputs, imagine the gravitational attraction of highly relevant numbers in the budgeting process, such as this year’s outcomes when discussing next year’s targets.
Many management techniques attempt to overcome this challenge. Zero-based budgeting is one, but the process is time-consuming and unrealistic on an annual basis. Another is the “what would it take” exercise, where the CFO quizzes managers on, for example, what it would take for them to double their assumed rate of growth or to achieve the same results with half the resources. When used sparingly, these are useful challenges that can get teams to rethink their assumptions.
But they can also lead to ineffective budget conversations if they become too familiar, as presenters learn to expect the challenge and sharpen their arguments for a given budget number or performance target. Most such techniques are still susceptible to the influence of past performance. In fact, we’ve found anchors to be so powerful that only another anchor can dislodge them. Reanchoring combats the anchor of history and convention with another anchor, grounded in a different set of facts. For example, consider just one aspect of a budget discussion in which you agree on sales targets for a number of regions. To reanchor the discussion, you would need to take three steps.
Identify what will determine performance. Set fact-based criteria that define what’s possible with respect to sales, such as market size, current market share and sales-force head count relative to competitors. Don’t try to include every factor that affects sales; the criteria need to be plausible, not exhaustive. It’s important to make sure that objective data can be found for those you do choose. This year’s targets (or results) should not be a factor — they already have enough weight as an anchor. Alternatively, data from competitor benchmarking can be quite useful in this context. We have found that three to seven objective criteria usually suffice to obtain a plausible, “good enough” model. For instance, one consumer-products company built such a model using just four criteria: current market size, projected market growth (in absolute dollar value, not percentage terms), current competitive position and a composite metric for competitive intensity.
Estimate sales potential. This should not require a massive effort. The aim here is to set next year’s sales targets as if you didn’t know this year’s, relying only on the criteria you defined. Many modeling techniques do this, but the simplest — a statistical regression based on your criteria — is usually sufficient because this isn’t intended to be a predictive model.
To calibrate your model, you need to check that it is directionally consistent with sales of the past few years, yielding plausible outputs in a majority of cases. It’s probably precise enough if the model output is within 10 percent of the historical numbers in two-thirds of sales territories. If that’s not the case and model results are extremely different from your existing plans in the vast majority of cases, it could be because the criteria or the data used in the model are wrong. But if you confirm that the criteria and data are right, your historical numbers may be entirely arbitrary. In such circumstances, you’ll need a more fundamental rethink than reanchoring can provide.
Position the model as a second anchor. With a well-built model in hand, you can use its output to challenge the status quo and change the dynamics of the discussion. For instance, instead of starting a budget meeting by saying, “You’re on track to sell 100 units this year, and you’re aiming for 103 next year. I think you can do better,” you can change the conversation into something more concrete: “You’re aiming for 103 units next year, but the model tells me you have the potential to aim for 120.”
Of course, this will require a longer discussion. Fortunately, in our experience, such conversations are seldom necessary for more than a third of a company’s sales territories. For the rest, the proposed targets will be close to the model’s estimate, so discussions can be moved through more quickly.
Companies that apply this kind of approach find that it focuses debate where it’s needed and reduces the inertia that anchoring induces. It is not a panacea: at the end of the day, you will still have to make tough decisions. But reanchoring will help make difficult conversations considerably more productive.
Dan Lovallo is a professor at the University of Sydney Business School, a senior research fellow at the Institute for Business Innovation at the University of California, Berkeley, and an adviser to McKinsey; Olivier Sibony is a director in McKinsey’s Paris office.
This article was originally published by McKinsey Quartlery. Copyright © McKinsey & Company. All rights reserved. Reprinted by permission.