From Business Questions to Actionable Evidence

Hello! Last lesson showed why a decision about extending a café’s hours needs shared input from marketing, operations, HR, and finance. Those functions can agree on the figures and still face a harder question: What do the figures actually tell us to do?

This lesson focuses on the data analyst’s role in answering that question. You’ll follow the proposed evening-hours trial from a broad concern to a measurable question, examine what evidence is available, and see how an analyst can present a useful conclusion without disguising uncertainty. Allow about 35–40 minutes, including the short video segments and readings.

Start with the decision, not the spreadsheet

Imagine the café owner asks, “Would staying open later be worth it?” An analyst could immediately request sales records, but those records alone cannot answer the question. First, the analyst needs to clarify what worth it means and what decision the owner is prepared to make.

Perhaps the immediate choice is whether to try two additional hours on two evenings, rather than whether to change the café’s schedule permanently. The owner wants the trial to cover its extra costs while keeping service manageable. A more useful analytical question is therefore:

How many additional orders per evening would cover the extra costs, and is there enough reason to expect that level of demand during a limited trial?

The word additional matters. An order placed later by a customer who otherwise would have bought the same thing before closing may not add as much revenue as an entirely new order. “Per evening” matters, too: demand and costs must be measured over the same period.

CareerFoundry’s A Step-by-Step Guide to the Data Analysis Process begins with this act of definition. Read its example for the distinction between the concern a manager first raises and the question an analyst can investigate.

The short Maven Analytics video makes a related point: identify the decision-maker, the desired business result, and what would count as success before exploring figures.

For the café, the decision belongs to the owner. The analyst’s contribution is to define the comparison the owner needs: expected additional sales against additional costs, subject to staffing and service capacity. That definition guides which evidence to gather—and helps prevent a polished analysis of the wrong question.

Decide what would count as evidence

Once the question is clear, the analyst can work backward to the necessary information. Last lesson’s four functions contribute different pieces: finance has prices and costs, operations knows supplies and capacity, HR can confirm coverage, and marketing may have evidence of customer interest. The analyst’s task is not simply to put those figures in one file. It is to check whether they describe the same trial and are suitable for the question.

The café might use its own order records to learn how busy the period just before closing usually is. It could review customer feedback about later hours and obtain estimates of extra wages, utilities, and cost per order. Each source has a different evidential role. Existing order records describe purchases that happened; customer responses describe reported interest; a forecast of 24 evening orders is an estimate, not an observation of hours the café has never offered.

Read the opening of CareerFoundry’s data-collection section for its distinction between the information a question requires and the sources that might provide it.

Before calculating, the analyst also checks definitions and quality. Do order counts include cancelled purchases? Are costs quoted per order or per evening? Are records from both proposed evenings complete? A duplicated sale could inflate expected demand; a wage estimate covering only one of the two evenings could understate costs. These are not minor technical concerns: they can change the answer presented to the owner.

There is also a gap no amount of cleaning can fix. Because the café has not yet opened during the proposed hours, it has no observed sales for those hours. An analyst should state that plainly. Past sales and customer feedback can inform a trial, but they cannot establish exactly how many additional customers will arrive.

Turn the evidence into an answer to the question

The next move is to analyze the gathered information in relation to the decision. In the video’s next segment, notice how the discussion of preparing data leads into looking for patterns that matter to the chosen measure.

The previous lesson used illustrative café estimates: an average sale of $12 per additional order, $5 in order-related costs, and $150 in extra wages and utilities per evening. Rather than repeating the total at the forecast of 24 orders, the analyst can now answer a more decision-oriented question: What is the break-even order count?

Each additional order contributes an estimated $\$12-\$5=\$7$ toward the evening’s extra wages and utilities. The number needed to cover $\$150$ is:

Orders are whole purchases, so 22 additional orders per evening would cover those specified extra costs under these assumptions. The forecast of 24 orders is only two orders above that threshold. At 21 orders, the café would fall short by $\$3$; at 22, it would exceed those costs by $\$4$.

That narrow margin is more informative than saying simply, “The forecast shows a positive result.” It tells the owner how sensitive the result is to demand. It does not prove that 24 orders will arrive, that all of them will be genuinely additional, or that the calculation includes every possible cost. Nor does breaking even necessarily meet the owner’s desired return. The analyst should check whether covering these extra costs is the right decision criterion before treating 22 as a sufficient target.

This is the difference between a calculation and usable evidence. A calculation produces a number. Usable evidence explains what the number measures, what assumptions support it, how close the proposed plan is to its threshold, and what remains unknown.

Give the decision-maker a conclusion they can act on

An analyst has not finished by handing over a table of orders and costs. The owner needs an answer framed around the available choices, with the important qualifications easy to see. CareerFoundry’s final section emphasizes both clear presentation and honest disclosure of gaps.

The Maven Analytics video adds a practical final step: connect a finding to a proposed action and a way to assess the result.

For the café, a concise decision brief might say:

Finding: Under the stated prices and extra-cost estimates, about 22 additional orders per evening would cover the specified costs. The forecast is 24, leaving little room for lower demand.
Limitation: Evening demand has not been observed, and some later orders might replace earlier purchases.
Possible action: Consider a two-evening trial if staffing and supplies are available. Record actual orders and costs, and review whether sales are additional before deciding on permanent hours.

The brief separates three things that are easy to blur: what the analysis finds, what it cannot yet establish, and what the owner could choose to do. A report, a simple chart, or a conversation could all convey this answer; the format is useful only if the intended decision-maker can understand the comparison and its uncertainty.

Key takeaways

A data analyst translates a business question into usable evidence by clarifying the decision and its measure of success, identifying suitable sources, checking their definitions and quality, analyzing them against the decision criterion, and communicating both the finding and its limits. In the café case, the break-even threshold gives the owner a clearer basis for considering a trial than a sales forecast alone—but the forecast remains uncertain until the hours are tested.

Next, you’ll use a simple business case to compare the forms and functions of two organizations, including how their different purposes affect the questions they ask of data.