The executive who offers to build the analysis a decision will rest on chooses its questions, data, and comparisons, and the room starts its debate from that picture of the problem.
An example
A specialty apparel chain with 340 stores had to close some of them, and the executive committee could not agree on how many. The vice president of real estate wanted to close the weakest fifty by sales per square foot. The head of e-commerce believed stores drove online orders in their areas and wanted a shorter list. At the end of a tense meeting, she offered to have her analytics team build the model the committee would use. Nobody objected; it sounded like work. Her team matched store locations to online orders by postal code and found that shoppers near a store spent 30 percent more online than shoppers elsewhere. The model scored each store on combined sales. Real estate checked it and found no errors. The committee approved closing twenty-two stores, and the remaining debate was about which twenty-two.
Why it works
Whoever builds the analysis decides which questions get asked, which data count, and what is compared with what, and the room then reasons inside that frame. Measured on terms that capture its strengths, the proposal looks better, and agreeing is easier, since the committee is accepting a model it asked for. Alternatives can look worse when the chosen measures expose their weak points. The work has to be honest, with the assumptions in view, done by a team able to produce it, and open for rivals to check. A model built to reach a preset answer is found out quickly.
The thinking behind it
Jeffrey Pfeffer Pfeffer's Managing with Power: Politics and Influence in Organizations (1992) treats information and analysis as tools that are used selectively in political contests (chapter 13): whoever builds the analysis decides which facts and comparisons appear in it. The clean version of this tactic accepts that point and answers it with openness. Circulating the questions, including rivals' measures, and handing over the model keep authorship from turning into the selective use the book describes.
Herbert A. Simon Simon's Administrative Behavior (1947) holds that decisions are shaped by the premises an organization supplies, and that whoever controls those premises influences the decisions made from them. An analysis the committee commissions becomes its shared set of factual premises. The volunteer who builds it gets the factor they believed was missing, here online sales near each store, into those premises, and the committee then reasons from them.
When it fits
Decisions resting on analysis nobody has done yet are the setting, where the question can be framed more than one way and the deciders want someone to do the work. You need a team that can build a credible model, access to the data, and the standing to be trusted with a shared task. It suits a manager whose view depends on a factor others are missing.
How to use it well
Volunteer early, in the meeting where the need for analysis comes up, and frame the offer as service to the group. Then do the work in the open.
- Write down the questions the analysis will answer and circulate them for comment before you start.
- Include the measures your rivals care about alongside the ones you think are missing.
- State every assumption and show how the answer moves when it changes.
- Give the other side's analysts the model and the data, and fix what they find.
Present the results as the group's, and recommend only after the facts are on the table. The advantage you keep is a fair one: the factor you thought mattered is now in the picture.
What it costs and where it backfires
Analysis takes time and people from your own work, and you own its mistakes. If the result goes against your view, you are bound by it. Rivals will scrutinize every assumption, which is the point, but it makes the work slow. The line is the honesty of the figures. Once data are shaded, inconvenient measures dropped, or assumptions buried, this stops being framing and becomes falsification, and it discredits the analysis and the analyst for good.
When it is used on you
Accept when a rival volunteers to build the analysis, and ask to see the questions, data sources, and assumptions before the work starts. Ask for your own measures to be included, and assign someone from your side to check the model. If that is refused, commission your own figures in time for the meeting.