Most of us, at one time or another in our lives, have found ourselves in a conversation that started off with someone else saying to us “Hey, I don’t understand something, can you explain this to me?” which quickly morphed into (or what they really meant but were being polite/indirect) was something more like “I have a mental model of how things are supposed to be. As far as I can tell, the way they currently are does not match that mental model, which confuses me, and I would like you to justify to me why things are not the way I expect. I also want you to either change the way things are, or at least advocate and spend your political capital within the organization, to advocate for the change I want.”

I call the former kind of conversation an “informational” conversation – that is, a conversation where one person is explaining to a second person the way things are right now, with no implication of needing to justify or change them. I call the second kind of conversation a “negotiatory” conversation, which is where there is an expectation from at least one of the parties that something will be changed, and the other party in the conversation is going to do the changing or at least advocate for it.

Confusion between these kinds of conversations often occurs in large, matrix organizations where responsibility for decision-making versus execution is not clear, and also where some members of the organization have an expectation of a more democratic or distributed decision-making model (“Everyone gets to have a say in things”), versus other members having an expectation of a more hierarchical or authoritarian model (where decisions are made more centrally and the results of those are propagated through the organization.)

This has major implications for analytics projects because analytics projects themselves, by necessity, have a number of implicit decisions within them regarding the meaning of data, how the business systems should operate, and the design of metrics which in turn will directly influence the work, incentives, and relationships of everyone in the organization. One important aspect of this is centralizing the definition of the calculation of metrics (such as allocation of sold revenue credit to a particular units, departments, or individuals in the organization) and then pushing the automated calculation of that metric out to members of the organization. Then, further, using the metric to drive changes to behavior and accountability.

A well-run analytics project will have good up-front conversations about these metrics, and well-run companies and leadership teams will think about the broader implications of how they want to use the metrics to incentivize their teams. However, regardless of how much in the way of effort and planning is done for this, there is no way to get everyone to agree 100% on how these metrics should be calculated, and the details of them often quickly become important. Commonly, when members of the larger organization encounter the metrics in practice, they question and push back on them. Various different groups all have a particular viewpoint on how the metric should be calculated, and all the steps backwards from the metric to where the data is being entered, stored, and maintained. These users then come back to the analytics team who coded the metric and the calculations underneath it – but who were not themselves responsible for deciding the processes and procedures around gathering the data nor were they responsible for the requirements around calculating the metric and also are not responsible for the changes in behavior that the metric is intended to produce – and expect the analytics team to be responsible for a) changing the metric to match their expectations and b) advocating on their behalf to the leadership of the organization for their change. In other words, they want to have a negotiatory conversation with the analytics team.

This approach does not work because the analytics team a) does not usually have the authority or political capital to advocate for change, b) does not have the organizational status, ability, or skills to adjudicate between the competing visions of all the different entities who want to see a change to the metric, and most importantly c) it is probably the case that the greater leadership team actually understood that members of the larger organization would disagree, and explicitly decided to give the requirements that they did with the intention of causing change in the organization anyway.

This does raise the question of why there often isn’t more direct and explicit communication regarding these changes beforehand to the larger organization. One reasonable and common explanation is just that these things are complicated and often it’s hard to coordinate all of this. Another explanation is that this kind of communication and consensus-building is time-consuming, tiring, and causes friction, and people in large organizations tend to be motivated to avoid conflict because it uses political capital. So sometimes it’s easier to just make the change, push it out, and then deal with the objections on a case-by-case basis. Also, if you’re a smaller company or a startup and iterating quickly, sometimes it’s easier to just throw things at the all and see what sticks.

This is where we come full-circle, back to confusion between informational versus negotiatory conversations. How do you handle it when you suddenly realize that you are in a meeting or conversation where someone else wants to negotiate about something, and believes that you can change things for them, but you don’t have the power and/or aren’t in a position to do so? On an analytics project, the first thing that I like to do when I can tell that I’ve been pulled into one of these conversations is to, as quickly as possible, stop the conversation and say “Look, I really want to help you here. Unfortunately, I think there may be some confusion here about who needs to be part of this conversation to make the changes you want. I understand that you would like changes to the metrics and reports that are being published. This is really a negotiation about differing opinions on how the whole organization should be run. I am happy to explain to you what the report/metric you are looking at is intended to calculate, and to some extent the thinking behind that insofar as I understand it. However, if you disagree with the way the metrics are designed, and by extension, the management decisions for the organization, that would need to be a conversation with the greater leadership team. The current design is intentional.” Which is sort of a negotiatory conversation itself, but it’s a much reduced version of it versus actually discussing change to the metrics and reports.

The larger solution to this is for the greater leadership team of the organization to be more proactive and explicit around expectations for metrics, performance, and behavior, and to communicate in advance that they understand that there may be discomfort and disagreement around change. They also will likely need to be available for more time to discuss feedback and potentially even be open to further changes to processes and the metric itself. In other words, this is a change management exercise spanning the whole organization and not just a technology or data conversation.

I know that some of you may be reading this and saying “Josh, this is abrasive and way too direct, regardless of whether you are talking about the application of this principle in analytics projects or more broadly in life.” First, I believe that it is more productive to be direct about what you do and do not have the power to do, and also to avoid a lot of wasted time and frustration with merry-go-round conversations that bounce around because nobody can decide who is “the decider.” In other words, identify the misunderstanding, then go right to the heart of the disagreement as soon as possible. In the longer term everyone will be better off for it. Second, it is possible to phrase this in a way that isn’t overly abrasive. Try to have the conversation in a gentle way and express sympathy. I believe that confronting the confusion directly is the more authentic and thoughtful response because to avoid conflict when the drivers of it exist but are submerged is inauthentic and fosters all sorts of corporate dysfunction.