Using data to make and justify business decisions

If two hundred thousand people said they preferred your new recipe, would you switch? In nineteen eighty-five, Coca-Cola did. It had run blind taste tests with around two hundred thousand people, and most of them preferred a new, sweeter taste. So it changed the recipe of its famous drink. Within three months, after a storm of complaints, the original recipe was back on the shelves. How could that much data point the wrong way? Those tests counted something real, just not the thing that mattered.

Every piece of data can tell a business some things and not others, so a good decision starts with knowing which is which. Quantitative data, numbers that can be counted, is precise: it is easy to compare over time and between options, and you can calculate with it. Its limit is that it shows what and how many, but never why. It may be out of date, or come from a small group that does not match the business's customers. Past figures may not predict the future, and what people tell a survey can differ from what they go on to do. Now back to nineteen eighty-five. Coca-Cola's tests were numbers about which sip people preferred. What could numbers like that not tell the company? They could not tell it why people bought Coke. The tests measured taste in one sip, not what the original drink meant to people who had drunk it for years. Qualitative data, words that give opinions, feelings and reasons, fills that gap. It explains why people behave as they do, and gives detail a number cannot. Interviews with loyal drinkers could have shown how attached they were to the original. The numbers said which sip won, and the words could have said what people would lose. Qualitative data has limits too. It is opinion-based. It is hard to measure or to compare. It often comes from a small sample, meaning only a few people are asked, and it takes time to collect. So numbers are not always more reliable than words. Coca-Cola had two hundred thousand results, and they still misled it, because they answered a different question from the one that mattered. That is why a good decision uses both kinds: numbers for how many, and words for why. Now a different question, about a restaurant. It reads one customer's long, angry review online. Which limit of qualitative data should stop it changing its menu because of that review? The review is one person's opinion, a sample of one, so it cannot show how many customers feel the same.

Turning data into a decision takes five steps, and each one asks its own question. Select picks the data that matters to this decision. Interpret says what it shows. Analyse says what it means for this business. Evaluate asks how far the data can be trusted, and what sits on the other side. Decide answers the question and justifies the answer. A justified decision names the fact from the data that tips it. Then it says why that fact outweighs the other side, and what would change the answer. If an answer says "it depends", its judgement has only started, and the answer still has to finish it. The exam-skills chapter practises this routine on each kind of long question, and the routine itself stays the same. Greenway Grocers is deciding whether to stock an organic fruit and vegetable range, at two pounds more per bag than its normal range. It has two pieces of data. In a survey, seventy-five per cent of two hundred customers said they would pay two pounds more for organic. In a two-week trial on one shelf, organic bags sold thirty a week, against a hundred a week for the normal bags. Using both pieces of data, should Greenway stock the organic range? Pick one, then hold on to it before I reveal it. Select comes first, and both sources matter, because a question that gives two sources expects you to use both. Interpret the survey: seventy-five per cent of two hundred is a hundred and fifty customers who said they would pay more. Interpret the trial: organic bags sold at less than a third of the rate of normal bags. Analyse for Greenway. The survey suggests demand that rival shops may not be meeting, but the trial shows what customers did when the bags were on the shelf. Now the evaluate step, as a question for you. Which should Greenway trust more, the survey or the trial, and why? Greenway should trust the trial more, because it records real purchases. A limit of the survey is that people often say they will pay more than they really will. That gives the judgement. Greenway should stock a small organic range rather than a full one, because thirty real sales a week matter more than what customers said. It should review after three months, and expand only if organic sales climb towards the normal range's hundred a week. That last condition is what would change the answer. One examiner's report on a question that gave more than one source says, "Too many candidates did not use the data, or only used one source which meant they were limited to a maximum of 5 marks." The fix is to use every source the question gives, the way Greenway's answer used both the survey and the trial. Now try the pattern on a new case. A bakery's survey says many customers want gluten-free bread, but a month-long trial sold two gluten-free loaves a day. Which fact should tip the bakery's decision? The trial's two loaves a day should tip it, because those are real sales, while the survey only records what people said they wanted.

A justified answer uses the case's own data, while a generic answer could be written about any business at all. Two students answer this question: should Brightwater Swim School add Saturday afternoon lessons? The data says its Saturday morning classes have a waiting list of forty-five children, and its pool hire costs sixty pounds an hour. Answer A says, "Yes, because more lessons would bring in more money and the swim school would grow, which is good for any business." Answer B says, "Yes. Forty-five children are already waiting for Saturday places, so three new afternoon classes of fifteen would fill at once. At eight pounds a child, each one-hour class brings in a hundred and twenty pounds against sixty pounds of pool hire." Answer B goes on, "The risk is that waiting families may not want an afternoon slot, so Brightwater should ask them before booking the pool." Which answer is justified, and why? Take your pick, and hold on to your reason before I show you. Answer B is justified. It uses the case's own figures to show why the decision fits this swim school, and it says what could go wrong and what to check first. Answer A could be written about any business. It uses none of the data, and its reason does not come from the case, so it does not justify the decision for Brightwater. Another examiner's report records that "some candidates made no attempt at justifying their decision", ending instead on a bare line saying what the business should do. The fix is to end on the reason, never on the bare decision. And answer the decision the question asks. Brightwater's question is about Saturday afternoons, so the answer says yes or no to Saturday afternoons, not to some other idea. A third report describes answers that used the data well, like this: "Candidates who used quantitative data well selected data from the table which showed financial data and used this to support their advice." That is what Answer B does with the waiting list and the pool cost.

Now the questions turn to you, one at a time, starting with the limits.

What is one limit of quantitative data? It shows how many but not why. It can also be out of date, come from an unrepresentative group, or record what people say rather than what they do.

Now the other kind. What is one limit of qualitative data? It is opinion-based and hard to measure or compare. It often comes from a small sample, and it is slow to collect.

Now a cinema to decide for. Its ticket sales fell by a fifth last year, and many of its customer comments say the seats are uncomfortable. What decision could it justify, and with which facts? It could justify replacing the seats, starting with its busiest room. The fall of a fifth shows how much sales dropped, and the comments show one reason why.

Now the final question. What turns a decision into a justified decision? It names the fact from the data that tips it, says why that fact outweighs the other side, and says what would change the answer.

Coca-Cola's mistake in nineteen eighty-five comes down to this. Its numbers said which sip people preferred, but never why they bought the drink, and the decision turned on the why.

If that's in the bag, give it a thumbs-up and move on. If not, save it. This one comes with the questions, and every decision you practise makes the pattern more familiar.

Next in the chapter is Business size, scale and context: how they shape decisions. Hopefully that's next in your feed, but if not, search for ScholaFly B U oh two, oh five.

Every other video is on scholafly.com too, easy to search by video code or by keyword. Whatever you're revising, go there and have a look, or, watch the next video.