Postal surveys have a very bad image, so much so that by far the majority of industrial marketers dismiss them as being inappropriate as a research technique.
There are two reasons for this view. First, most postal questionnaires end up in the waste bin and this implies that whatever response is generated comes from an atypical minority. Secondly, postal questionnaires do not allow a subject to be discussed in the detail that is possible in a face-to-face interview or even in one held over the telephone.
Though there is some truth in both these assertions, they can be countered. The representativeness of postal response can be assessed by a telephone check on a sample of non replying companies, while the physical absence of an interviewer can on occasions be beneficial as it allows the respondent to consider answers with no bias caused by the researcher’s presence or questioning technique. Postal surveys are a specialised research tool which have an important role to play in industrial marketing research as long as they are used in the right circumstances and with the necessary controls.
When deciding whether to use personal interviews, telephone interviews or a postal survey, two questions must be asked:
- What type of information is required?
- What is the structure of the industry being researched ?
Where the information required is qualitative, unstructured or detailed, there can be no substitute for an administered interview. Open ended questions in a self completion questionnaire can produce vague and useless answers. For example, the question “What made you choose your current supplier?” may well elicit the response “Because they were best.” The same question asked face-to-face could generate a wealth of information as the interviewer probes why the supplier is considered best. Postal surveys are most appropriate where questions are simple, factual and can be asked in a closed form (i.e. questions where the respondent merely has to tick a box which indicates his choice of answer).
The structure of the industry that is being researched also has a bearing on whether administered interviews or a postal questionnaire should be used. Wherever the industry contains a few dominating companies it would be dangerous to use a postal questionnaire as the failure of just one company to respond could affect the findings. Imagine how representative a survey of main frame computer manufacturers would be if no response was obtained from Fujitsu-ICL and IBM who together dominate the British market. Furthermore, if the buying decision is split between two, three or even more departments, as frequently is the case with industrial products, a postal questionnaire would fail to pick up the opinions of all the decision makers.
The divided nature of the buying decision need not detract from using a postal questionnaire if the objective is to collect hard facts (the works manager can say how many tons of steel he uses just as easily as the buyer), but it does present problems when exploring factors which influence the buying decision (the works manager may believe quality to be the most important factor while the buyer may say price).
The over-riding factor determining whether or not to use self-completion questionnaires is the interest of the respondent in the subject. A high level of interest, and a strong relationship between respondent and questioner, will result in a high response rate. They are, therefore, ideal for customer satisfaction studies when the respondent feels obliged to help in the survey because of their partnership supplier/buyer status.
Although we have argued that postal surveys have a special role to play in industrial market research, they need not be used in isolation. The 80/20 rule applies to nearly all industrial samples and enables a researcher to divide the list of companies to be contacted into three groups according to their size. All the largest companies can be picked off for the personal interview programme. A sample of the medium sized companies can be visited and telephoned while data from the small companies is collected by postal questionnaire and telephone. Such a balanced method of sampling ensures that key companies are covered, all levels of the industry are researched and at the same time the costs are much lower than a survey involving only personal visits.
The variable costs in organising a postal survey of 1000 questionnaires is £600 to £800. A typical example is shown below. This excludes the cost of a mailing list which could be around £100/1,000 for a computer listing.
| Cost | £ |
|---|---|
| Outward post | 200 |
| Return post (30% response) | 60 |
| Printed questionnaire (A3 as A4 booklet) | 150 |
| Reply paid envelopes | 50 |
| Outward envelopes | 50 |
| Typing & folding | 100 |
| Total | 610 |
Table 1: Cost Of Mailing 1,000 Questionnaires
What matters is not so much the total cost of a mailing, rather the cost for each returned and completed questionnaire that is valid for analysis; which is dependent on the response rate. Response rates to postal surveys can vary from 5 per cent to over 50 per cent from one mailing. As the objective of every researcher is to achieve the maximum possible response rate it is worthwhile considering the factors which affect it. As has already been stated the key factor is the strength of the relationship between the research sponsor and the respondent. Other factors are:
The accuracy of the list of companies in the sample has a marked effect. A sample frame which contains many companies who have moved, or do not buy the product that is being studied, will generate a poor response. Time spent on weeding the sample frame to make it as accurate as possible pays great dividends in raising the response.
Addressing the questionnaire to the right person is similarly important. It is better to address the envelope and letter to a respondent named personally, but as this is frequently not possible at least the correct title should be used. The title should be specific (e.g. the Chief Buyer – Computer Products) rather than vague (e.g. the Design Engineer); and if in doubt, the questionnaire should be aimed high in the management hierarchy rather than low. It is surprising how many managing directors pass on the questionnaires to the right person, and of course when this happens the chances are it will be diligently completed and returned.
The interest which a respondent has in the subject also influences the response. A postal survey about trucks will achieve a better response from road hauliers than will a survey on ball point pens amongst buyers in general.
The length and complexity of the questionnaire can lower the response rate, though the effect of this can be reduced by making it visually appealing with questions designed to maintain interest. Every attempt should be made to make the questionnaire as easy to answer and return as possible and for this purpose a pre-paid envelope is essential. The effect of a stamp as opposed to a business reply number makes a small difference (compare the accuracy of the sample frame which makes a very significant difference) and is tedious to organise in a large survey.
The presentation and persuasiveness of the covering letter helps considerably to improve the response rate, and for obvious reasons the questionnaire should not be timed to land just before, during or immediately after a holiday. Some researchers would argue that the day of the week that the questionnaire is received can affect the response as can the request for respondents to identify themselves.
A 30 per cent usable response would generate 300 questionnaires for analysis – that is £2 per usable reply. 100 interviews carried out by over the telephone (averaging 5-6 interviews per day) would cost around £25 to £30 if purchased from an agency. A face to face interview supplied by a market research agency would be £250 to £300 each.
It is clear that if the circumstances are appropriate for postal surveys, they have considerable attractions in cost compared with other fieldwork techniques. One person can easily organise the survey while visits and telephone interviews can be carefully selected to provide a balanced sample.