Negotiation Styles for Data People

Data people work with facts and data, so if we give the people the right facts, and the right data, they’ll act on it… right?

Getting people to do stuff is a central concern for data people. When you really boil it down, if you can’t get people to act on your analysis, then the analysis all feels a little pointless (and rightly so).

Most early career analysts do not put much thought into how to get people to act on their work, and it usually comes as a shock (and with a degree of frustration) when they first finally encounter the issue. More senior data professionals will recognise this issue as an old friend and know that it can never be taken for granted. Even in the ideal case where a stakeholder has promised they will act on the analysis if only they could please have it soon - we’ve all had cases where the numbers have been ignored.

Why

#

Before we look at how to combat this, we need to understand why the problem exists.

Too often, the rationale that data people reach for is that the person consuming the analytics doesn’t understand or isn’t listening; and while there is sometimes a grain of truth to this, to chalk the situation up as being only down to these causes is to miss a whole heap of other potential influences. Some of the more common reasons are:

1. They don’t trust the raw data. Without good trust in the source information, they will never trust anything you make from it.
2. They don’t trust you. More sceptical stakeholders may not yet have reason to believe that your analysis is reliable.
3. They can’t put your recommendations in context (“it’s too hard to digest”). Even if your recommendations are strictly true, if the person consuming it, can’t easily work out what to do with it then they will give up.
4. They have other sources of information which give different recommendations, and (perhaps without telling you) they’re taking a kind of consensus view.
5. They have other incentives to act in a different way, or to not act at all. These may be formal incentives (like bonuses, or promotions) or they may be implicit incentives like comfort, visibility or status.
6. They themselves are struggling to rally support from the people they need to, to actually effect change.

This is not an exhaustive list, but it should at least illustrate the breadth of reasons which might mean that someone doesn’t act on your analysis, even if they do understand it, and do agree with your methods.

In an ideal scenario, you’d understand some of these wider factors before you did the analysis up front so that you could take some of them into account. That isn’t always possible, and not all of our stakeholders are open enough, or conscious enough, to articulate these reasons to you openly.

So, knowing the minefield we’re walking into, what can we do about it?

Influencing Styles

#

Thankfully, the data profession is not the first profession to come up against this problem, and lots has been written about it from the perspective of sales, leadership and negotiation professionals. All fields where they live or die on their ability to influence people.

The lessons from those fields can be applied almost directly to the data field, and one of my personal favourites which I find useful in coaching scenarios is the Influencing Styles model by Gary Yukl and J. Bruce Tracey ¹.

The model describes 11 influencing styles, divided into a “positive” and a “negative” group. The positive set tend to build relationships, and the negative set tend to erode them. That isn’t to say however that it is never appropriate to use the negative styles, but it is to say to only use them sparingly, and to be conscious of their potential side effects.

Before we dive into the styles themselves, it’s worth thinking about which styles work best for which of the blockers we identified earlier. If they don’t trust the data (#1), or don’t trust you (#2), you’re going to need collaboration or consultation to build that trust by working together. If they can’t digest your recommendations (#3), try rational persuasion with better framing, or inspirational appeal to bypass the complexity. If they have competing information (#4) or other incentives (#5), apprising them of indirect benefits or exchange might help realign their interests. And if they’re struggling with their own stakeholders (#6), consider coalition building to help them make the case. The trick is matching the style to the underlying problem, not just defaulting to more data and better charts.

The Positive Styles

#

1. Rational persuasion. This is the favourite of the analyst, and typically our default mode. You know it, and so I’m not going to talk about it much.

2. Apprising or “indirect benefit”. This comes from explaining that the person gains an indirect benefit from taking the action, even if they might not agree with the facts. Often this works in data by saying that it will make them look good because they’re being “data driven”. Almost all of our stakeholders are under pressure themselves to look good, and look like they’re making the best use of data. This is something you can use.

3. Inspirational appeal. Do it because it’s the right thing to do. This works really well on topics where data is scarce, or where there is significant ambiguity (e.g. recruitment, marketing or sustainability). Often appealing to a foundational principle can be enough to sell an idea. Whether it’s symmetrical parental leave or flying less, you can do all the analysis you want, but it might be the values argument which actually gets things moving, rather than your comparative analysis.

4. Consultation. Let’s make a plan together, or if you’re feeling machiavellian, inception. The key here is that you’re involving them in your work, but asking for their input to shape the outcome. This can seem odd from a data lens, where typically someone is expecting you to do the work - but this comes up quite regularly in the “let me walk you through my analysis” approach where you gradually lead them through the thought process and involve them in the thinking - so that together, you come to a plan of action which you both understand and agree with. The analysis is still yours, but by consulting them throughout, they feel ownership of the conclusion. Think: “I’ve done some initial analysis on customer churn, but I’d love your input on whether these segments make sense before I go further.”

5. Exchange doesn’t come naturally to data professionals: you help me and I help you. But in side issues like budgeting or people conversations it can be invaluable. This works best when there is no obvious rational reason why what you want helps the other person, but where you can give them something that they want, and in return they can do the same for you. e.g. “if I can get this analysis to you by next week, do you think you can approve my OPEX plans for this quarter?”. Being in control of a scarce resource (like a data team) does give a certain amount of power, and sometimes this method of influencing can be a creative way to get things moving when other methods aren’t working.

6. Collaboration. Let’s do it together. Different from consultation, this is about actually working together on the task itself, rather than just involving them in your thinking. This works best when you are representing your team in a group, and you want others to act in the way you act. By also role modelling those same behaviours, you make it easier for them to do it too. Picture rolling up your sleeves together in a workshop to map out a new process, or jointly running an experiment where you’re both contributing equally. You’re not leading them through your analysis (consultation), you’re genuinely building something together as equals. Think: “Let’s spend tomorrow afternoon together prototyping this dashboard - you know the business context and I know the data, so we’ll need both.”

The Negative Styles

#

1. Legitimation or authority. This is the exercise (or attempted exercise) of hard power. It is very very rarely a tool which you can use as a data leader. You may find it occasionally used on you by C-suite leadership (and you may choose to resist), and very occasionally you may need to use it in a managerial context with your team, particularly on disciplinary matters. But for this conversation it’s basically off the table except to explain other influences on your stakeholders. It’s possible that they might agree with you, but someone is exercising authority over them to prevent them acting in the way you want. Recognising when this might be happening is critical to working around it.

2. Coalition is where you use the group to overpower an individual. Used without good evidence it’s just cruel, but can be used very positively if you need a group to reach consensus and know there are one or two who you’ll never persuade. By making sure you have sufficient support, you may be able to effectively outvote, or overpower the remainder.

   This is one style, which while in the negative section here, can be a tremendous force for good, if used with integrity. It also favours the general biases of data people to work one-to-one with individuals rather than in big group settings. If you have an introversion bias, you can often use this approach to make sure the decision is effectively made (by persuading a sufficient majority) before the decisions is “officially made”.

3. Pressure, which includes threats or repeated demands. This often leads to resentment and disengagement, but I see this shockingly often as a tactic used to influence data teams. Picture the stakeholder that keeps asking, even though you’ve said no, just to wear you down. While it’s not a pleasant tactic, it can be deployed sparingly as a reminder to use the work you’ve already delivered: “Act on this analysis, or we will not prioritise further work on this topic”.

4. Ingratiation, “buttering up”, schmoozing, call it what you want, and we’ve all seen it. It’s gross, and it works. This also includes appealing to favours, and the trading of favours. As an occasional option, for people with whom you already have a good relationship this can occasionally be a very fast way of short cutting a queue, but if overused causes bigger trust issues which are harder to solve.

5. Personal appeals or “because we’re friends”. As a data person, this can harm your perceived integrity quite quickly. Avoid if possible.

Choosing the right style

#

So you have 11 options. How do you pick?

The honest answer is that this comes with practice, and there’s no substitute for getting it wrong a few times to learn what works. That said, there are a few principles which can help you “read the room”:

Start with the person, not the problem. Different people respond to different styles. Some stakeholders are very analytical and respond well to rational persuasion. Others are more values-driven and will respond better to inspirational appeal. Some are collaborative by nature, others prefer to be presented with options. The more you know about how someone typically makes decisions, the better you can tailor your approach.

Consider the relationship. If you have high trust with someone, you have more options. You can use consultation or collaboration effectively because they’re willing to engage. If trust is low, you might need to start with rational persuasion or apprising to demonstrate value before you can move to more collaborative approaches. And if you have very strong relationships, you might occasionally get away with exchange or even ingratiation - but tread carefully.

Read the stakes. High-stakes decisions often require more formal approaches like rational persuasion or coalition building. Low-stakes decisions might be better suited to consultation or even personal appeals. If the decision is time-sensitive, exchange or apprising might help you move faster than building a perfect logical case.

Watch for resistance. If you’re getting pushback, it’s often a sign you’re using the wrong style. If someone keeps asking for more data, they might not trust you (#2) - try collaboration to build that trust. If they agree with your analysis but won’t act, look for hidden incentives (#5) - try apprising or exchange. If they seem overwhelmed, simplify with inspirational appeal. The style you choose should address the root cause of their hesitation, not just repeat your argument louder.

Be willing to switch. You don’t have to pick one style and stick with it. Often the most effective approach is to start with one style (say, rational persuasion) and then switch to another (say, consultation) if you sense it’s not landing. The key is to stay flexible and responsive to how the conversation is going.

None of this is covered explicitly in Yukl and Tracey’s original work - their focus was on identifying and categorising the styles themselves. The guidance on how to choose between them draws more broadly from negotiation theory and practical leadership experience. If you want to go deeper, Chris Voss’s “Never Split the Difference” and Robert Cialdini’s “Influence” both have excellent frameworks for reading situations and choosing tactics.

Summary

#

No one of these styles is “better” than another. What matters as data professionals is that we recognise the options we have, and the factors the others might be utilising around us. By being conscious of our context we can then choose the right tool for the job. That also includes the sparing use of some of the less analytical persuasion styles.

Fundamentally, you should choose the influencing style which you think works best for your target, not the style which works best for you.

If you’re interested in exploring more about how data work inherently involves navigating subjective judgements (not just objective facts), you might also enjoy:

Good luck!