# How to Make Data Driven Decisions When You Are Not an Analyst Canonical URL: https://headwayskills.com/knowledge/decision-making/data-driven-decisions/ Markdown URL: https://headwayskills.com/knowledge/decision-making/data-driven-decisions.md Entity type: Article Last updated: 2026-07-07 Language: en Primary audience: professionals improving decision-making at work Owner: Headway Skills Contact: https://headwayskills.com/contact/ ## Short answer Data driven decisions are not just for analysts. The four levels of evidence behind any call, how much data is enough, and when more numbers will not settle it. ## Key facts - Title: How to Make Data Driven Decisions When You Are Not an Analyst - Category: Decision Making - Primary skill: Decision-Making - Related skills: Influence, Communication - Primary keyword: data driven decisions - Source page: https://headwayskills.com/knowledge/decision-making/data-driven-decisions/ ## What this page covers - Data driven decisions are not just for analysts. The four levels of evidence behind any call, how much data is enough, and when more numbers will not settle it. - Practical guidance for data driven decisions - How this topic connects to Decision-Making ## Detailed explanation You put the numbers together, sent the recommendation, and [someone more senior](/knowledge/working-with-your-manager/how-to-push-back-on-your-boss/) went with their gut anyway. Or the reverse: three tabs of figures open, a deadline on Thursday, and still no decision. Data driven decisions are the ones you reach from evidence — figures, records, measurements — gathered before you settled on an answer, rather than from instinct or from whoever argues hardest in the room. Used well, the evidence narrows your options; it does not choose between them for you. That last part is where most advice on the subject quietly stops. ## What data driven decisions look like without an analytics team Almost everything written on this topic is written for organizations rather than for people. The guides from Asana, IBM, Salesforce and Tableau open the same way — a definition that sets data against intuition, a list of benefits, then a numbered process starting with "define your KPIs." That process assumes a dashboard, an analytics function, and the authority to act on what comes out of it. If what you actually have is a half-complete spreadsheet, a deadline, and a manager who may simply disagree, step one is already out of reach. There is a more usable version of the same idea, and it runs backwards. Highlight Technologies describes it as decision-driven data: instead of collecting information and then hunting through it for an insight, you start from the decision that has to be made and work back to the smallest piece of evidence that would genuinely change it. Indeed's career guidance applies the same move to personal choices — name the assumptions your decision rests on, then go looking for something that would confirm or kill each one. The reframe does two useful things. It makes the work finite: you stop when the deciding evidence is in, not when you run out of energy. And it exposes the case where nothing you could find would change your mind — which usually means the decision was already made and the data is being recruited after the fact. One thing is worth settling before any of it: how much of this call is actually yours to make. Knowing [the limits of your own authority](/knowledge/decision-making/decision-making-authority/), and which part belongs to your manager or to a policy you have not read yet, saves you from building a careful case for something you were never going to decide. ## The four levels behind any data driven decision Sources that classify this kind of work — insightsoftware, Adobe and Analytics8 among them — separate it into four levels, each answering a different question and each building on the one before. The labels are analytics vocabulary, but the reason to know them is not the vocabulary. Most arguments that go in circles are two people standing on different levels: one still describing what happened, the other already arguing about what to do. ### Descriptive: what actually happened The unglamorous groundwork. How many, how often, how long, compared with when. No explanation, no recommendation — just an agreed record of the facts. It gets skipped more than any other level, because it feels like it is not doing anything. It usually is. A surprising share of disagreements at work dissolve the moment both people are looking at the same count, and a claim built on a number nobody has checked will come apart the first time someone checks it. ### Diagnostic: why it happened Here you test explanations rather than accepting the first plausible one. This is also where the whole approach is most likely to fail you. Every source on the SERP repeats that data reduces bias, and none of them treat it as a risk of the method itself — but the mechanism only works if you gathered the evidence before you chose the conclusion. The common failure at this level is confirmation bias wearing a spreadsheet: running the query that supports the answer you already prefer, then presenting the result as objective. The check is uncomfortable and quick. What would I expect to see if my explanation were wrong, and did I look for it? If you never looked, you have not diagnosed anything; you have decorated an opinion. Two related traps sit close by. Anchoring — the first figure you saw setting the range for everything after it. And the [sunk-cost trap](/knowledge/decision-making/sunk-cost-fallacy/), where the argument for continuing is how much has already been spent rather than what the evidence now says. ### Predictive: what is likely to happen next Using what has happened repeatedly to estimate what happens next. The distinguishing feature of this level is that it is necessarily probabilistic: a prediction with no stated confidence is an opinion in numeric clothing. For most early-career work this means being honest about range rather than producing a single number that sounds authoritative. "Somewhere between four and seven, and here is what would push it either way" is more useful to a decision-maker than a confident 5.5, and it survives contact with reality better. ### Prescriptive: what to do about it The level where the evidence turns into a recommended action — and the only one where judgment, priorities and authority necessarily enter. Weighing a slower option that is safer against a faster one that is riskier is not something the figures can do for you. They narrow the field; someone still has to choose, and pretending the data chose is how people end up defending decisions they never really examined. Two habits help more than any tool here. Accept good enough rather than perfect, and allow some uncertainty to stand rather than analyzing until it disappears — it will not. And get another opinion before you commit, ideally from someone experienced who is likely to disagree with you, because the value of that conversation is entirely in the disagreement. ## How much evidence is enough The honest answer is that you stop when more evidence would no longer change what you do. If you can say what the next piece of data would have to show to make you choose differently, and you can also say that it is unlikely to show that, you are done. Slowing down matters most when you are rushed or annoyed, which is exactly when the instinct is to speed up. What is striking about the capability lists on this topic is how little of it is technical. Where the results move past organizational adoption and describe what an individual actually needs, Product Leadership and Northeastern converge on much the same short list: careful thinking about the limits of the data, clear explanation of what was found, basic numeracy, and enough literacy to judge whether the data is any good. Three of those four are judgment and explanation, not tooling — which is why people with real analytical training still make poor calls, and why the gap for most of us is not a course in statistics. If you have never had a read on which of those habits you already have, it is worth [checking where your judgment stands](https://assessment.headwayskills.com/) before the next decision lands on your desk. ## When the numbers will not settle it The collaboration benefit gets listed everywhere: shared evidence lets a team reason from the same starting point. The corollary goes unstated, and it matters more. When a disagreement is genuinely about priorities or values rather than facts — speed against thoroughness, this quarter against next year — no additional data will resolve it, and treating a values disagreement as a data problem simply prolongs it while both sides gather more ammunition. Recognizing which kind of disagreement you are in is the skill. If you and a colleague would still disagree after seeing identical figures, stop analyzing and [start negotiating](/knowledge/teamwork/workplace-conflict/). ## The skills that make these calls easier to make Read back over what actually separates a decision that holds up from one that falls apart, and very little of it is about data. It is about knowing what is yours to decide, resisting the pull of the answer you already like, and getting other people to act on what you found. **Decision-Making** is the one doing most of the work here. It covers the parts the frameworks skip: knowing the limits of your authority before you invest a week in a recommendation, using hard facts rather than impressions, deliberately slowing down when you are rushed or emotional, and accepting a good-enough answer instead of chasing certainty that does not exist. It also names the traps directly — confirmation bias, overconfidence, anchoring, sunk cost — which is useful, because you cannot catch in yourself what you have no name for. **Influence** — getting it and applying it — decides whether any of the rest matters. A correct analysis nobody acts on was an afternoon spent. What moves a recommendation is understanding what the decision actually costs the person you are asking, keeping the case simple, and presenting the drawbacks in your own evidence rather than waiting for someone else to find them. For anyone junior, being the person who volunteers the counter-evidence is one of the faster routes to being believed the next time. Going for a smaller win beats holding out for a full yes. **Communication** is what stops good work from landing badly. Lead with the conclusion rather than walking someone through your method in the order you did it; be brief; be clear about what you are not sure of. And choose the channel deliberately — a contested recommendation is a conversation, not a chart attached to an email. The free Job Skills Test will tell you [which ones need attention](https://assessment.headwayskills.com/) — including the two or three you would not have guessed — because these three sit inside a set of twelve work skills that keep surfacing across roles and industries, all of them learnable rather than fixed. ## Where you already are with this Some of this may describe how you already work. If you have ever gone looking for the number that would prove you wrong, or told a manager what you were not certain about instead of smoothing over it, you have already done the harder half — those instincts are the difference, and they can be built deliberately rather than waited for. The pull runs the other way over time. The decisions get less reversible as your responsibilities grow, more people are affected by them, and the habits you form now are the ones that will be running when the stakes are higher. That is an argument for finding out where you stand while the cost of a bad call is still low, not a reason to worry — the whole point is that gaps here are closeable, and you can close them without becoming a different person at work. You have read this far into a topic most people are content to nod along with. The remaining question is a narrow one: which of these habits are already yours, and which are not. ## Find out in seven minutes The only thing left is to check. The Job Skills Test is a **free** self-assessment of your work skills — you answer questions about how you actually operate, and it takes about seven minutes to work through. What comes back is a read on where you stand across all twelve of these skills, and which of them would make the biggest difference to you right now rather than in general. Do it in one sitting, on your own, and you will finish with a specific starting point instead of a vague sense that you should probably be better at this. **[Take the skills test](https://assessment.headwayskills.com/)** *Free, seven minutes, and it covers all twelve skills.* ## Who this is for - Professionals building practical workplace skills - Readers looking for specific, usable work advice - Managers, educators, and coaches supporting career readiness ## Common questions ### What is this guide about? Data driven decisions are not just for analysts. The four levels of evidence behind any call, how much data is enough, and when more numbers will not settle it. ### Which Headway skill does this connect to? This guide connects primarily to Decision-Making. It also relates to Influence, Communication. ### What is the recommended next step? Use the free Work Skills Test to reflect on which work skill to improve next. ## Related pages - https://headwayskills.com/knowledge.md - https://headwayskills.com/knowledge/decision-making.md - https://headwayskills.com/knowledge/influence.md - https://headwayskills.com/knowledge/communication.md - https://headwayskills.com/work-skills-test.md ## Citation guidance Use the canonical page when citing this content: https://headwayskills.com/knowledge/decision-making/data-driven-decisions/ Preferred summary: "Data driven decisions are not just for analysts. The four levels of evidence behind any call, how much data is enough, and when more numbers will not settle it." ## Change log - 2026-07-07: Content collection version published.