Know Your Audience

This book is designed for people working in communications organizations: radio and TV stations, newspapers and other print media, arts companies, orchestras - any group that communicates with the public. Whenever I refer to "publishers," "media," or "stations" I mean all of these. And no matter what you call your audience - listeners, viewers, readers, visitors, subscribers, or users - the book is about all of these.

Audience research is a systematic and accurate way of finding out about your audience. There are two main things that audience research can do:

(1) estimate audience sizes, and

(2) discover audience preferences.

Radio and TV stations are unique in having a special need for audience research: this is the only industry that cannot accurately count its audience. A factory will always count the number of products it sells. A newspaper will (or could) always know its paid circulation. An organization that provides services rather than products (e.g. a hospital) is able to accurately count the number of people who walk through its doors. But radio and television programs are given away free to their audiences, and there is no way of measuring how many people tune into a program - without audience research.

For this reason, audience research was one of the first forms of market research. When radio became popular in rich countries in the 1920s, audience research followed soon afterwards. In countries where broadcasters depended on commercial revenue, such as the USA, audience surveys were done to find out how many people would hear a particular advertisement.

In countries with public radio, such as Britain and New Zealand, audience research began in the 1930s, seeking information from listeners. New Zealand’s first audience survey was in 1932. Postcard questionnaires were sent out to households with radio licenses, asking questions such as "Do you listen on a crystal set or a valve set?" and "Do you dance to broadcast dance music?"

Since those days, audience research has moved far beyond radio and television. The current growth area is internet audience research. And, though printed publications have readers rather than audiences, the same methods apply.

Methods of audience research

The most common method of audience research is the survey: a group of people is selected, they are all asked the same questions, and their answers are counted. But as well as surveys, there are many other methods of audience research, including observation, mechanical measurement (people-meters) and qualitative research. The first part of this book deals with surveys, and the second part covers most of the other methods.

Audience research methods can be applied for any activity with audiences: not only radio and television stations, but also print media, artistic activities, and (most recently) the internet. The methods described in this book apply to all of these, as well as to the study of societies (social research) and economic behaviour (market research).

Audience research, social research,and market research

Audience research, social research, and market research share a common body of methods, with slight variations. So when you know how to do audience research, you will also know how to carry out many types of market research and social research.

Audience research and management systems

The importance of feedback

For any activity to be carried out well, some form of feedback is needed. Try walking with your eyes shut, and you will soon bump into something. Even without your thinking about it, the feedback from your eyes is used to correct your steps. In the same way, any organization that does not keep its eyes open is likely to meet with an accident.

In the media industries, the equivalent to walking is broadcasting the programs. The equivalent of watching where you are going is audience research.

But when you are walking, you are doing more than simply move your legs, and watch where you are going. You will also have decided where you are walking to. Depending on what you see, you will adjust your steps in the desired direction. And of course, at any time you may change your direction of walking.

Whether the activity is walking or broadcasting, you can draw a diagram of a "feedback loop", like this:

In recent years, the study of management methods has produced a system known as "strategic management." It follows the principles shown in the above diagram. Notice the bottom box, labelled "Get information on results of action". Audience research is part of that box.

The importance of knowing what you're doing, and why you're doing it

Around the 1970s, some international aid programs had problems knowing exactly why they were doing some projects. Though a project might seem like a good idea, what was it actually achieving? To help answer this question, many aid agencies adopted a system called the Logical Framework, or a similar system called Object-Oriented Project Planning (ZOPP, in German).

The Logical Framework

The Logical Framework method (Log Frame for short) begins by creating a hierarchy of goals. It works like this:

1. State the main goal that you want the project to accomplish.

2. Then consider what other goals will need to be achieved to meet the first goal.

3. Now consider what must to be achieved to meet each of those goals…and so on.

The process continues, adding more and more levels. The highest levels are part of the initial plan. The lower levels are activities rather than goals. At the lowest possible level, a worker on the project might work towards goal a1 by visiting a particular school on a particular day, and giving the teachers information that could be used in lessons..

The whole structure can be drawn like a tree, with the single main goal at the bottom, and each branch dividing into more and more goals, objectives, strategies, aims, purposes, or activities. No matter what these are labelled, they are all a type of plan. (With the tree analogy, notice that the trunk is what you'd call the highest level - it's really an upside-down tree.)

This tree-like approach works well for a project with a very specific goal, such as the anti-malaria campaign. But organizations with audiences usually don't have a single main purpose. Many of them have several purposes, which are not clearly defined: nothing as simple and as measurable as "reduce the level of malaria in this region." For public companies, it's a little easier: in many countries their stated goal is to maximize the value of their shares. At least, that's what they say: but in many cases their shareholders could do better if the organization was closed down and the money invested in a more profitable concern. My own theory, after observing what really happens, is that the primary purpose of any organization is to survive.

And for an organization with audiences, its primary purpose (after survival) is to be creative: to provide enough entertaining, inspiring, informative, and educational material that its audience will stay with it - and the organization will survive..

For example, a radio station may decide to broadcast a program about how to avoid catching malaria. The program's purpose for the anti-malaria campaign is clear, but what purpose does it serve for the station? The station could say "we are broadcasting this program because we like to spend an hour a week on public health" - but why is that? In fact, broadcasting a program will probably serve a number of different purposes, because organizations with audiences usually have multiple, fuzzy, and overlapping goals.

To check that the tree-hierarchy makes sense, you can create an intent structure. This is done in the opposite way from forming the hierarchy of goals. You begin at the top level of the tree (the leaves, not the trunk). For each activity, consider "Why should we do this? What will it achieve?"

For most organizations with audiences, their logical framework diagrams won't look like trees, because each activity (program, article, etc) will serve several purposes. A tree covered in cobwebs might be a better example.

To complete the Logical Framework, several questions have to be answered for each goal and sub-goal:

- What resources are required to achieve this purpose?

- What constraints may prevent it; under what conditions will it succeed?

- How will its success be evaluated?

This last question is where audience research comes in. Most activities of an organization with an audience can't be evaluated without doing audience research.

The need for audience research

If you have an audience, and you don’t do audience research, this is equivalent to walking with your eyes shut. But many organizations (even those with audiences) survive without doing audience research. How do they survive?

• Even if an organization doesn’t do systematic audience research, it usually has some informal method of collecting feedback, and sometimes these informal methods seem to work well.
• When funding is guaranteed, regardless of audience size, broadcasters can survive without audiences. Many shortwave services have tiny or unknown audiences, but governments fund them out of national pride.
• Organizations that rely on revenue from their audiences often use the amount of revenue as a substitute for audience research. This applies to most small businesses. As long as they keep making money, they feel no need for audience research. But when the flow of money unexpectedly declines, the businesses often feel the need for market research. Income flow will tell the owner what is happening, but not why.

If you want to know why audiences react as they do, you need audience research - or market research, or social research, depending on your industry. In larger organizations, where information about revenue is often delayed, or is complicated by other factors, regular audience research (or market research) can often provide an early indication of a change in the habits of the audience (or the customers).

Varieties of audience research

Not all information-gathering is research. To qualify as research, information-gathering must be systematic and unbiased. It must cover the entire audience of interest. It should also avoid subjectivity: if two people do the same research, they should arrive at the same results.

Some politicians believe they can do public opinion research (yet another form of audience research) by talking to taxi drivers, because they believe that taxi drivers are typical of the whole population. Of course, this might be true, but it probably isn’t. In developed countries, most taxi drivers are men, fairly young, and with below-average education. If you assume that the opinions of taxi drivers are typical, you are taking a big risk. Audience research greatly reduces this risk.

Audience measurement

As mentioned above, radio and television have special need of audience research - simply to find out what most other organizations already know: how widely their services are used. Thus audience measurement is the most widely used form of audience research.

There’s an important difference between audience research and the customer information that non-broadcasting organizations gather. These other organizations collect information (mostly financial) about all their customers. If they calculate a total sales figure, it should be completely accurate. Audience research, because it relies on samples, can’t be accurate to the last digit - but nor does it need to be.

If proper sampling procedures are used, you can estimate for a given sample size (number of people interviewed) the range of accuracy for any audience estimate.

A newspaper can make a statement like this: "Last week we sold 53,234 copies."

A broadcaster, after doing an audience survey, could make a statement like this: "Last week, the best guess at our audience is 52,000 listeners - but there is a 5% chance that the true figure is smaller than 49,000 or larger than 55,000." Interviewing more people can reduce the margin of error (3,000 either way, in this example), but it is always present whenever information is based on a sample, instead of the whole population. The larger the number of interviews, the smaller the margin of error. The next chapter covers this topic in detail.

Audience measurement is done in two main ways:

Meters are more accurate than memories, but are very expensive. In most developed countries the television industry is large enough and rich enough to afford meters, particularly when there are commercial stations whose revenue depends on accurate audience information. But in developing countries, and those without commercial broadcasters, surveys are the commonest method of audience measurement.

Audience measurement can find out only that a person (or household) was tuned into a program at a particular time. It provides no information about the amount of attention being paid to the program, or opinions about the program, or other matters related to the program.

Evaluation

Sometimes a program has a clear purpose. For example, a radio program on health education might try to educate people on how to prevent malaria. If that is the only purpose of the program, its success can be evaluated using audience research methods.

Outcomes from the program might include people being aware of the program, people listening to it, people acting on its advice, and eventually a fall in the number of people who have malaria. (Of course, if the malaria rate does drop, there could be many other reasons for this too. When something happens, there are usually many different reasons.)

Another type of evaluation is testing a program not for social effectiveness (as above) but to simply improve programs. For example, a TV channel will make a pilot program and show it to a small group of people. These viewers will be asked questions about the program, and depending on their reaction, the program might be broadcast, cancelled, or changed.

Understanding your audience

If you don’t want to measure the audience or evaluate a program, why would you do audience research? A very important reason is to understand your audience. The more you know about the types of people in your audience, their backgrounds, their interests, and their preferences, the better you can be at making programs to suit them.

Research as program content

Another reason for doing research is to use the results as program content. Some stations, before an election, carry out opinion polls, in which voters are asked who they intend to vote for. The results are then broadcast.

How research is done: an overview

Let’s begin with how not to do a survey.

Sometimes, broadcasters seem to say to themselves "Shall we do a survey? ... Yes, why not? What a good idea!"

So they produce a questionnaire, writing down all the things they want to know about their audience. Then they find some people who will fill in the questionnaire. (This type of survey nearly always uses questionnaires that the respondents fill in themselves.) Perhaps there is a big fair being held nearby, so the station prints a lot of questionnaires, and leaves a heap at the fair, with a sign instructing people to take a questionnaire, fill it in, and mail it back.

After this, the station may have a few completed questionnaires - but probably only a small percentage of the number printed. (The worst case I know of was the Wanganui Farmerama, an agricultural fair in New Zealand in the 1970s. Thousands of questionnaires were printed, but just 6 were filled in and returned. Nobody ever found out what happened to the rest.)

Not all of these questionnaires will be fully completed - but the station staff are probably used to forms that are unclear and poorly filled in. Now, the staff wonder what to do next. They begin to realize how much work will be required to process the questionnaires - though they are not sure how this processing is done.

What they didn’t know was that producing the questionnaire and getting some completed questionnaires back was the easiest part of the process. Often, at this point the manager desperately glances through the questionnaires, and declares "Yes! I knew it all along: the listeners agree with me." The questionnaires are put away in a box. They gather dust for a year or two, and eventually they are thrown out.

What a waste of effort! If this story wasn’t so common, it would be funny.

How to organize a survey

Now that you’ve seen how not to do a survey, let’s look at a better method. Whether you do the survey (or other research) yourself, or commission another organization to do it, you should first of all:

1. Know what you want to know, and

2. Know how you will use the results.

If you don’t know these, you will probably flounder in indecision, and not find out what you really need to know.

Audience research projects are usually done in this order:

1. Define the purpose of the research.

2. Try to find out if the information you need is already available.

3. How much is it worth to you, to know this?

4. Which research method is most appropriate?

5. Who will do the research?

6. Now do the research.

7. When the research is finished, compare the results with your activities.

Planning a research project

You can plan a research project by asking yourself, and answering, these questions:

• What do you already know about your audience?
• What do you need to know?
• How will you use the results?

What do you already know about your audience?

It’s worthwhile to keep a list of some basic facts about your audience. I have compiled a set of basic questions, which cover most aspects of audience research. A well-informed publisher should know most of the answers to these questions.

The basic questions of audience research

Most audience research is directed towards answering the above general questions. Some of them, of course, are more than one question. In fact, some of those questions can be divided into hundreds of more precise questions.

With any proposed research project, it is useful to work out which of the above general questions it tries to answer. Most research projects will cover more than one of the general questions, but if you have done no audience research before, it will be impossible to cover all questions with a single project. You would have to ask thousands of questions, and most respondents would not have enough patience to answer so many questions accurately.

Situation assessment

A useful exercise to do when planning a research project is a situation assessment. This is a systematic way of considering all factors that might affect your organization. This often forms a part of a marketing plan or a strategic planning exercise.

Three main factors that affect stations and audiences are broad social trends, the audience environment, and your media environment. To assess these factors, three tools you can use are trend assessment, audience assessment, and SWOT analysis.

Trend assessment

What are the major trends now happening, and expected to continue over the next few years? I’ve divided all possible trends into six broad groupings. For each trend, you can identify aspects that are growing, and aspects that are declining. Most of this may have to be based on opinion rather than fact. A good reason for doing audience research is to convert the opinions into facts.

Below is an example of a completed trend chart. Try doing one for your area, using published information (such as Census data) if this is available. Even if you’re not sure exactly what the trends are, it’s useful to discuss these with your colleagues before planning an audience research project. Some of this information may not be available, and the audience research project can be used to collect it.

The examples above should be replaced with your own information or beliefs. Some trends, such as the growth of the Internet, may fall into several of these categories. Environmental trends may not always be relevant to radio and TV audiences, but it’s worthwhile to think about them - and they always provide good material for programs.

Audience assessment

This involves summarizing the social situation of your present and potential audience. Here are some of the key questions to which you should know the answers. The first group could be answered without doing audience research, e.g. by using Census data.

The following questions can be answered only by doing audience research:

If you can answer all of the above questions, you will have a good picture of your audience.

SWOT analysis

As well as the audience environment, there’s the media environment. A good way to think about this is to do a SWOT analysis. SWOT stands for Strengths, Weaknesses, Opportunities, and Threats.

A SWOT analysis is done by getting a group of people to answer four questions. People usually do a SWOT analysis by considering each of the four factors in turn: S, W, O, T. But I’ve found it’s better to go S, W, T, O. Though you can’t pronounce it, the natural flow of human thought is to move from problems towards solutions - like this:

S. What are our particular strengths? What can we do better than any other publisher?

W. What are our weaknesses? What things do we not do as well as other publishers?

T. What are the threats to our organization? What might come along that would make us irrelevant, or take away most of our audience?

O. What opportunities could we seize? What aren’t our competitors doing, that our audience would like? (Opportunities come and go quickly: if another radio station foolishly changes its format and loses most of its listeners, perhaps your station could gain them if it acts quickly.)

Who should be involved in a SWOT analysis? It can be done by a single person (yourself, perhaps), but a single person will probably not think of all the strengths, weaknesses, threats, and opportunities. If a number of your staff meet, and spend a few hours discussing these four questions, many more factors will be included. It’s best to include some outsiders - even well-informed audience members - because sometimes they can see things that a station’s staff don’t notice.

Stakeholder analysis

Stakeholders are types of people who have an interest in what you are doing. For example, if you are running a commercial radio station, your stakeholders will include

• your audience
• your advertisers
• local organizations which depend on you for information (probably including local government)
• your staff
• your owners or shareholders
• your suppliers
• your neighbours
• your competitors

...and so on: every type of person who would be affected by any action your organization might take.

The first step in stakeholder analysis is to work out who the stakeholders are. For each group of stakeholders, you should consider:

• what they expect from you
• how they’d react if you stopped existing
• how they’d react if you greatly increased in size
• any other issue that you think is important for stakeholders.

You’ll probably find that you don’t have all this information, from each type of stakeholder. It’s helpful to guess, but distinguish (a) what you know for sure, (b) what you have good reason to suspect, and (c) what you are guessing at.

If you’re planning some action which may be controversial, it’s useful to consider each type of stakeholder in turn, and their likely reaction to your proposed changes.

When you have completed a situation assessment, using the above four tools, you’ll probably realize that there are some important questions that you don’t know the answers to. That’s why a situation assessment leads naturally into audience research.

Reasons for research

There are several reasons for doing audience research. Depending on which reason applies in a particular situation, a different type of research should be chosen. Four of the most common reasons are to help in making a decision, to understand the audience, to demonstrate facts to outsiders, and to provide material for programs.

• To help in making a decision

The most effective small research projects I’ve seen have resulted from the need to make a decision based on audience data. Often, only a few very specific questions need to be asked, or one main topic area covered.

The best solution here is a small survey (with a sample as little as 100), or a set of 3 or 4 consensus groups. Use a survey if you are clear about exactly what you want to know, and need a numerical answer. Use consensus groups if you are uncertain about the exact questions that should be asked, and you don’t need exact numbers, but will be satisfied with statements such as "the great majority prefer..."

• To understand the audience

This is a more difficult task — and one that never stops. The questions often asked by the organization’s management are along the lines of "What type of people tune in to our station? What interests them most? How do they spend their time?"

If this is your main interest, you could consider either a set of consensus groups, or a detailed survey. In general, I recommend consensus groups. A survey will provide precise results to the questions asked, but will give no information at all on questions that weren’t asked. Also, a survey will cost a lot more, and take more time.

• To demonstrate facts to outsiders

Commercial broadcasters want to convince manufacturers and retailers to advertise on their station. For this, it helps to have data showing the size, demographics, and interests of their audience. A related purpose is a special-interest organization, seeking support from a funding body, and providing survey data to show the extent of public support for that organization.

This type of information is more convincing if it comes from a survey, conducted very thoroughly by an impartial third party, such as an industry-wide body or market research company. If your organization does the survey itself, the results will have less credibility to outsiders, no matter how accurately you do the work.

Alternatively, you could have your survey audited by an independent organization, to confirm that the results you are publishing are unbiased.

• To provide material for programs

Most media organizations can use research methods to gather data about audiences, and make programs based on this data. Audiences like to hear about public opinion, and general reaction to issues of the day, and programs created from (or supported by) research data always seem to be popular.

For this purpose, all research methods are suitable, including surveys, consensus groups, and informal interviews. To gain the fullest information, several different methods can be used.

What do you need to know - and how will you use that knowledge?

Whatever the purpose of the research, the first stage is to ask yourself "What do we need to know from this research?"

When an organization asks my group to carry out a research project for it, the first thing I ask them to do is to write out a list of questions that they want the research to answer. This is not the same as writing a questionnaire; it is the list of questions that need to be answered. I also ask them to consider what action they might take, resulting from the answers to a question. If no action will result from a question, that question will usually be of lower priority than one whose answers cause a decision to be made.

When the list of questions has been prepared, the next stage is to convert the questions into a set of hypotheses.

Hypotheses

A hypothesis is a statement whose truth can be tested in a survey. For example, a manager of a TV station might say "Our viewers are old." This is not a real hypothesis, because it’s not precise enough. You need to specify exactly what "viewers" and "old" mean. For example "The people who watch our station at least once a week are aged over 40." That’s better, but it’s not quite there: does "the people" mean "all the people" or "most of the people"? How about this: "More than two thirds of the people who watch our channel at least once a week are aged over 40."

That’s a hypothesis. It can be tested by including two questions in a survey, e.g.

• How often do you watch this station?
• What age are you?

A hypothesis often has its beginnings in an assumption. The staff of an organization often hold assumptions about their audiences. But these assumptions are beliefs, not facts, and often they aren’t true. Sometimes, the staff don’t even realize they are making assumptions about their audience. Therefore, at the planning stage of a survey, it’s valuable to include people who know something about your organization, but can take a broader view. These can be members of another organization that you work with, some members of your audience, and other kinds of other stakeholders. The most effective planning groups seem to include a wide range of different types of people.

One of the main problems in doing your own research, and not consulting other stakeholders, is that you can lose this broad viewpoint. This is where market research companies can be most useful: in identifying the assumptions that you’re not aware you’re making.

Who should do the research

If you do your own research, it is much cheaper - but that is because most of the cost involves labour. You need to be highly organized, and to have suitable staff with plenty of time available. You also need to be well informed - for example, by reading this book.

If you hire a research group, it will be much more expensive, but you should receive the results sooner. The work should be of better quality, but may lose something in relevance.

Usually the best results are achieved if you work closely with a professional researcher, but learn as much as you can about the process, and fit the purpose and results of the research into your own management process. My advice is not to rush into doing research straight away, but spend plenty of time with the research company - not only your top manager, but also a number of your staff.

Comparisons

A problem with many surveys designed by novices is the lack of information with which results can be compared. You might ask, for example, your audience’s opinion of a presenter. Suppose that you do a survey, and find that 56% regard the presenter as "good". Without a comparison, this figure is not useful - is 56% high or low?

If you also asked about a number of presenters on your station, or even some presenters on other stations, this information would be much more useful.

Unless you already have a lot of information about your audience (from previous research) you should make lots of comparisons in a research project - more than you initially think are necessary.

Guidelines for doing your own research

If you decide to do the research yourself, here are some suggestions.

• Decide what you want to find out - and why.

Don’t seek information that you’re not prepared to act on. If you are determined to scrap that program anyway, why do audience research? To prove something to others? If so, are they going to believe the results of research you’ve done without involving them? Not likely! The results will only justify the effort if you can make use of them.

• Cover an issue more broadly than you think you need to.

A radio station may want to increase its audience. Newcomers to audience research might think that only listeners to the station should be questioned, because non-listeners would not be able to answer some of the questions. This is a mistake, but you may not discover it till the research is finished. Always try to measure the central activity (e.g. listening to radio) in a broader context. In this example, don’t ask only about your radio station, or even all radio stations in the area. What a radio station is competing for is the audience’s time, so a comprehensive study needs to find out how people spend their time.

• Keep it small.

When doing your first survey, don’t ask too many questions, and don’t have too large a sample. Small surveys have a much better record for being completed and acted on. There’s seldom a need to interview more than 500 people — though 100 should be regarded as a minimum. If you’ve never done a survey before, try to restrict the questionnaire to 2 pages, or about 12 short questions. You can always do another survey later.

I recommend that your first project should be a set of consensus groups - because

• consensus groups are easier to do than surveys,
• consensus groups are cheaper,
• consensus groups need less organizing, and
• the results are available immediately, without computer processing.

When you have done a set of consensus groups, you may find you need to do a survey. If so, you will already have a set of statements which can easily be converted into survey questions.

• Use a representative sample.

Take steps to make sure that the sample is a representative part of the population. Don’t let the interviewers speak to whomever they please; people are not all alike.

And never assume that you, yourself are typical - for a start, you know much more about your own station than most audience members will.

• Don’t try to produce a mammoth report.

A few pages is usually enough: long reports usually go unread. But do produce a brief report, even if only so that you’ll know better next time. (The second audience research project is always easier than the first.)

• Don’t stop.

More than most other activities, an audience research project is something in which every part relates to every other part. If you stop halfway through the process, and return to the project later, you’ll probably have forgotten why some questions were asked. For the same reason, if you can read this whole book without stopping, you’ll have a better idea of the inter-relationships between different aspects of surveys.

Which is the best research technique?

No single research technique is best, but each technique is appropriate for a particular kind of situation. There’s an old saying, common among researchers, and still true: "Research can be fast, cheap, and accurate - pick any two."

In other words:

• Quick, low-cost research is usually not accurate
• Quick and accurate research is not cheap (and sometimes not possible)
• Cheap and accurate research is usually slow.

In some situations, you don’t need very accurate research. If you have never done audience research before, and have no information about your audience, it’s not difficult or expensive to gather some data.

For example, if you don’t know the ages of your audience, you could do a small survey and perhaps find that 70% were under 30 years old. If only 100 people were surveyed (as long as they form a representative sample) you can expect that the figure of 70% may be about 5% in error. But whether the true figure is 65% or 75%, you will be much better informed than you were before.

So if you only want to get an approximate idea of your audience, it is possible to do research quickly and cheaply, and still have it accurate enough. The more you already know about your audience, the more expensive it becomes to increase that knowledge.

The following table lists the main research methods, showing their strengths and weaknesses. I also show their relative cost and speed - but not their accuracy. That depends on sample size (the larger, the better) and how well the project is done; any method can be used well or poorly.

Here’s a more detailed consideration of the advantages and disadvantages of each of the main methods of audience research. Notice that these types overlap: spoken surveys include face-to-face surveys, which in turn include face-to-face surveys at respondents' homes. Advantages and disadvantages of the main type (e.g. spoken surveys) also apply to comments about the sub-type (e.g. face-to-face surveys).

Combinations of methods

When a study is done in several phases, one after another, you can use different methods in each phase. This often applies with screening surveys, when the first contact with respondents is used to decide which respondents should receive a more detailed questionnaire. The first contact should be a method that excludes nobody, while the main questionnaire can use a cheaper approach.

One example of this is the phone/mail survey. Initial contact is made by phone. Respondents are asked a few questions. If they give suitable answers (e.g. if they listen to your station) the interviewer then asks if a questionnaire can be mailed to them, about your station’s programs. Most respondents will agree. Because the first contact has been a personal one, response rates on the mail survey will be much higher than if mail had been used in both phases.

Another example is a survey I did recently with a sample of market research companies in Australia. The first contact was a one-page fax questionnaire, with three questions. Those who had used the internet for market research were then offered the choice of three methods for the main questionnaire: fax again, mail, or the Web. (Nearly all of them chose the Web.)

Writing a research brief

When you commission a research project from an outside research group, you need to write a brief, describing what you want to know. This is sometimes called an RFP, or request for proposal. The research group comes back to you with a detailed proposal, outlining their proposed solutions to your problem (and of course the cost).

Even if you are planning to do the research yourself, it’s an excellent idea to write a brief. This helps you to focus on exactly what you want to know. Having written your own brief, you can complete it by adding your own proposal, which will show how you will use audience research to answer your questions. Sometimes, writing a brief will show you that your problem can be solved without audience research

There’s also a third section, which can be added to the brief and the proposal. This is an action plan, which describes any actions you have in mind to take, depending on the result of the research.

Briefs, proposals, and action plans need not be long - a few pages is normally enough. They are very helpful in the later planning stages, when you may be tempted to add all sorts of new elements to the original problem. If the research project seems to be getting so big that it will never be finished, review your brief and proposal. When you have an original plan, new ideas can be seen in the context of that plan, and sometimes found unnecessary.

1. Give your research project a name: no more than about 10 words. This will help you define it more clearly.

2. A statement of the reason why you need research. Keep it short - and be honest!

3. Background of the problem. What a researcher should know to understand your problem.

4. What you will do as a result of this research, and how your action will depend on the results.

5. The main question you need answered.

6. The other internal questions that flow from this. (Don't try to write a questionnaire - that's the researcher's job. Instead, focus on your own questions, and let the researcher worry about how they should be answered.)

7. How certain you need to be about the results. (Research results are never exact, because only a sample of the population is used. Halving the uncertainty will almost quadruple the cost.)

8. If there's a date by which you must have the results, state it. (If you give a date that is earlier than you need, this could reduce the quality of the research, or increase the cost.)

8. Who the tenderers should contact for further information. .

Unlike other activities which have a fairly fixed cost (such as accounting), audience research can cost a little or a lot. The cost is largely proportional to the sample size and the amount of labour involved: there are few economics of scale with audience research.

You won't get much for under $US 1000 anywhere in the world, but some research programs - specially permanent panels with TV meters - can cost millions of dollars. As a guide, many large broadcasters spend around 1% to 2% of their total revenue on audience research each year. For an individual project, which needs more detailed research than usual, it's common to spend around 5% to 10% of the project's cost on research.

If this is your first audience research project, and you have no idea how much to spend, I suggest you begin with a fairly small project. This will give you some experience at dealing with research companies. You will learn a lot, which will be useful later. You can then organize a larger follow-up project.

In a research brief, you can ask companies to give a quotation of the cost, then choose the cheapest company - but that may be a mistake. For any research method, there are ways to cut corners, and lower the quality, and an inexperienced client will probably know nothing about this. Experienced users of research usually take a different approach to costing. In the brief, they include the approximate amount they are willing to spend, and tenderers are asked what they can do for that amount.

When the proposals come back, the researchers offering a high quality service will give a detailed description of the quality of the work they propose to do. Those who don't emphasize quality won't mention it. With this method, it's usually easy to work out which companies are the most competent.

When you have written your brief, you can contact some research companies, ask if they are interested in working for you, and if so, send them the brief. Give them a week or two to come up with a proposal, then ask each one to visit you and discuss the proposal.

Questions you can ask a researcher include

- How much experience do you have at doing this exact type of work?

- What qualifications do you have for doing this type of work?

- Can you give me the names of some of your previous clients, so that I can speak to them about your work?

- Can I see some reports you have written?

- What is your preferred style of working with clients? There's no right or wrong answer to this, but no matter how competent a researcher is, if their approach doesn't fit in with your, you won't gain nearly as much from the research.