Once you have the results from your market research, what are you going to do with them? More importantly, what decisions are those results meant to improve?
They sound like obvious questions to ask at the start of any research project. But they are often missed. Instead, many projects begin with a list of questions, a preferred methodology and a vague hope that useful insight will emerge at the end.
That is where market research often goes wrong. The failure is baked in early. Once the wrong questions are asked, the damage is difficult to undo. The result is familiar: patchy data, unclear analysis and findings that are interesting, but not especially actionable.
Backwards research avoids this problem by starting where the process usually ends. Rather than moving from questions, to answers, to analysis, it begins with the decision the research needs to support. From there, you work backwards to identify the analysis, evidence and questions required.
That is the nature of backwards research: a more disciplined way to produce actionable insight that improves decision-making, rather than filling presentations with demographic charts that never influence strategy. For marketers, this is not just a research technique. It is a disciplined approach that makes research commercially useful.
A brief history
The backward research method was first set out by marketing professor Alan R. Andreasen in his 1985 Harvard Business Review (HBR) article, “Backward” Market Research. Its central principle is clear: start where the research process usually ends, then work backwards. Begin with the decision, the action and the final output. Only then decide what data you need, which questions to ask and which method to use. (Harvard Business Review)
The problem with traditional market research
The following research request will sound familiar. Like clockwork, it appears every year and usually produces the same predictable outcome: broad questions, vague findings and very little strategic change.
The annual research merry-go-round starts with a broad request:
“We need to understand our customers better.”
“We need insight into the market.”
“We need to know what people think of the brand.”
These questions are not inherently wrong. Most marketers would want the answers. The result is usually the same: a long survey, a large sample, a dense report, a detailed debrief and very little change in the final marketing decision.
Mark Ritson’s summary of backwards market research, from the Mini MBA in Marketing, is especially useful here. He argues that marketers should first determine how the data will be used, then envision the final report or presentation that makes the right course of action clear, and finally design the analysis and samples needed to support it. Only after that should the research commence.
Traditional research asks: “What do we want to know?”
Backwards research asks: “What decision are we trying to make?”
How backwards research works
A backwards research process follows a simple sequence.
First, define the decision.
This is the most important step. Are you deciding whether to enter a new market? Which segment to target? Whether to increase price? Which campaign route to back? Whether to kill or scale a new product concept? If there is no decision, there is no research problem. There is only curiosity.
Second, define the possible actions.
Research should help you determine the appropriate response to each possible outcome. For example: launch or do not launch; target segment A or segment B; hold price or increase it; retain a current positioning or change it. Agreeing the possible actions before conducting the research creates alignment early. It shifts the conversation away from debating inconvenient findings later and towards deciding how the business will respond once the evidence arrives.
Third, define the action standards.
These are the decision rules set before the research begins. For example:
“If awareness is below X, we invest in reach before consideration activity.”
or
“If the new proposition performs significantly better among high-value non-buyers, we progress to product development.” Action standards prevent teams from bending the results afterwards to support the answer they already wanted.
Fourth, sketch the final output.
This is one of the most powerful parts of the method. Before fieldwork begins, create the empty charts, tables or slides you expect to need. These are often called dummy tables. They do not contain fake results. They simply show the outputs required for decision-making. The World Bank’s sanitation marketing guidance, for example, references Andreasen’s backwards research and uses dummy tables to visualise relationships between data sets and guide analysis before reporting.
Fifth, work back to the analysis.
Once you know the final output, it becomes much clearer what analysis is required. Do you need a segment-level comparison? A price elasticity read? A cross-tab by current behaviour? A ranking of barriers? A claimed purchase intent measure? This is where the research starts to become precise.
Sixth, work back to the data and method.
Only now should you decide whether you need qualitative research, quantitative research, secondary data, social listening, customer interviews, ethnography, concept testing, pricing research or some combination of them. The method should serve the decision. It should not drive the project.
Why backwards research is so useful for marketers
Backwards research forces marketers to stop commissioning “general insight” and start commissioning evidence for a specific decision. That makes the research sharper, shorter and more likely to be used. It also improves stakeholder alignment in key decision-making moments. When senior decision-makers agree on the possible actions and final outputs before the research begins, they are less likely to challenge the findings simply because the results are inconvenient. The debate moves from “Do we like this result?” to “What did we agree we would do if this happened?”
Many research projects become bloated because teams try to cover every possible eventuality in a single study. What begins as an attempt to improve efficiency quickly becomes the opposite: an expensive exercise collecting data that will never influence a decision. A team may ask 60 questions when 20 would be enough to support the agreed decisions. Backwards research cuts out unnecessary data collection and helps marketers choose properly, because it starts with the decision that the evidence must support.
The choice between qualitative and quantitative research is now much clearer. Qualitative research is excellent for understanding language, motivations, barriers and meaning. Quantitative research is needed when you must measure scale, magnitude, incidence or difference. If you do not know what decision the research needs to support, you cannot choose the right methodology in the first place.
Finally, it makes research more commercially credible. Applied research should be judged by actionability - whether the findings help a decision-maker choose what to do. Behavioural scientist and marketing professor Utpal Dholakia makes this point directly, arguing that backwards research is effective because it forces the decision-maker and researcher to agree at the start what results are needed and how they will be used.
A real example: the bank with slowing sales
Dholakia gives a useful example of a bank facing slowing new account openings. A traditional approach might involve broad customer and employee research, producing a general report on satisfaction and attitudes. The problem is that such a report may confirm what the bank already knows without telling it what to do.
A backwards approach starts by clarifying which actions are actually on the table. The manager may be willing to test a referral incentive or a better interest rate for new accounts. They may not be willing to change employee pay, alter work schedules or advertise on local radio. That immediately changes the research design. The study should focus on customer receptiveness to referral offers and rate incentives, not on irrelevant options the business will never implement.
That is the value of backwards research in practice: it removes the illusion that all information is equally useful.
Another example: segmentation
Backwards research is especially valuable in segmentation. Too many segmentation projects begin with the ambition to “understand the market” and end with colourful personas nobody uses.
The backwards version begins with the strategic decision. Are you deciding which segment to target? Whether to prioritise penetration or loyalty? Whether different segments require different propositions, channels or messages? Should the company stop serving a low-value customer group?
Once that decision is clear, the research can be designed around usable segmentation outputs: segment size, value, accessibility, needs, current brand penetration, category behaviour and the practical implications for targeting and positioning.
Even Andreasen’s original HBR article from over 40 years ago offers the same learnings, opening with an entertainment company executive who wanted a more sophisticated view of the consumer segments the company served, after relying on broad groups such as younger audiences, older audiences, families and singles. The backwards lesson is that segmentation research should not merely describe groups. It should help management decide what to do differently because those groups exist.
What backwards research is not
Backwards research is not about deciding the answer in advance. That is just poor research thinking and design. It is not about manipulating the methodology to get the conclusion you want either. One of the first lessons from The MiniMBA in Marketing is that you are not the consumer. Preconceived assumptions must therefore be challenged, not protected. Backwards research means knowing what the evidence must help you decide before you start collecting it, not about the answers themselves.
The commercial value of backwards research
Backwards research is not about creating better questionnaires. It is about improving marketing judgement and avoiding vague insight projects that never lead to action. It improves research design and reduces wasted spend. It makes agencies more effective because they are executing against a tighter brief and a sharper problem. It gives stakeholders confidence because the decision rules are agreed early in the process. And it turns research from a passive reporting exercise into an active part of strategy.
If the research will not change what you do, do not commission it. But if the decision matters, work backwards. Define the action. Design the output. Build the analysis. Then collect the data.
Map the route before setting off. Do not pretend you already know the destination.