How to Measure Marketing

One of the biggest challenges many marketers face in their work is the lack of tangible results of their work. They launch new products, concepts, and advertising campaigns into the world, and all they get in return are numbers on sales reports.

Many are forced to make decisions and act blindly because they do not have the tools to measure marketing KPIs: brand strength, recognition, image, customer loyalty, etc. Previously, these metrics were only available to large companies, but with the spread of the Internet, surveys have become cheap and accessible to the general public.

Nowadays, in order to carry out a quantitative survey, it is enough to send questionnaire letters to your customer base, place a survey advertisement on the Internet, or order a representative sample in one of the online panels. An online panel is a database of respondents who have agreed to participate in surveys on a regular basis.

How to do market research

The easiest way is to contact a specialized agency. But this is expensive, and you need to be sure that the results will help you make decisions that will pay back the investment. Unfortunately, this is not always the case, so let's consider guerrilla methods available to all.

Online panels (OMI, Tiburon), questionnaire designers (Surveymonkey, Survio, Anketolog) or advertising your own Google Forms questionnaire online are suitable for conducting surveys of your potential audience. Thanks to targeting, you can flexibly customize the parameters of the audience you're going to survey.

Getting feedback from your own customers is the easiest way to get it: email newsletters, short questions in messengers or phone calls - all this is almost free.

The only limitation in using external sources should be the cost of their extraction relative to the budget of the task at hand.

Quantitative or qualitative?

All market research can be divided into quantitative and qualitative.

Qualitative techniques (focus groups, in-depth interviews) answer the "how?" question. Quantitative (surveys, measurements) - "how much?".

Focus groups and interviews dive deep into details, but do not answer the question: how many people have the same opinion? If 9 out of 10 interviewees don't like something about your product, it doesn't mean that the same proportion will remain in the general population. Usually qualitative research is used to generate hypotheses, which are then tested with quantitative methods.

Sample size, quotas, and random selection

Quantitative studies use a sample of respondents whose opinions are generalized to all consumers with a small margin of error. In order to judge all consumers by a small fraction, three important aspects must be controlled for:

• A sample size of at least 100 participants;
• random selection of respondents;
• adherence to quotas.

Statistics work well when there are 100 responses. In truth, the law of large numbers turns on even after the 30-response mark, but samples that are too small involve assumptions and limitations that only professional statisticians are familiar with.

Randomness of selection is a second important principle. You can't judge the average height of Russians by the measurements of a hundred basketball players who practice in your front yard.

Perfect sampling is when every person has an equal probability of being among those surveyed. Completely random sampling is hard to achieve (it's very expensive), but it's something to strive for. Online surveys, by definition, weed out all the people who don't use the internet. If your target audience is media-users or retirees, then use a phone call.

Quotas are the proportions of traits that must be met in a sample. If you are interviewing a potential audience, make sure that men, women, people of different ages, and other subgroups are evenly represented.

The wording of the questions is very important

The wording of the questions has an impact on the answers - that's a fact. You can reduce the influence by asking questions that do not include built-in evaluations. If you include an answer option in the question, use all alternatives.

Bad: "Do you think more money should be allocated to road maintenance?"

Good: "Do you think more money should be allocated to road maintenance, the same amount or less than now?"

Be sure to rotate the order of the answer list. The top positions are more popular because they are the first to fall on. For single choice questions, add "Other" and "Difficult to answer.

Just calculate the frequency of the answers or the average to get the results. Consider subgroup scores by gender, age, and other factors. Scores can vary dramatically, and that's a huge field for marketing.

To make sure that differences between groups really exist and are not due to sampling error, statistical techniques are used. The most common are t-tests, chi-square, and analysis of variance. They are available in the Data Analysis tab of regular Excel.

Remember that the size of each subgroup must be at least 100 people (at least 75). For a study that plans to compare the evaluation of a new pet store concept among cat, dog or fish lovers, you'll need 300 respondents, 100 from each group.

Start tracking the numbers on a quarterly basis. Watch to see how advertising campaigns change brand image and awareness, and service improvements lead to increased loyalty.

5 useful techniques

1. Price Sensitivity Meter (PSM) or VanWestendorp method

Objective: to determine the level of expected prices.

What price do you think is high for this product, but at what price you can buy it?

At what price would you buy this product if you thought it was a bargain?

Once you have the answers, you can build a chart that shows the psychological

2. Attribute Analysis - Importance and Satisfaction

Objective: examine the strengths and weaknesses of the product.

Please rate the importance of "......" attributes?

Please rate "....." for each of the parameters?

Be sure to explain to respondents what scale to use for evaluation.

After receiving the responses, construct a graph where the X-axis is satisfaction and the Y-axis is the importance of the characteristics.

The upper right quadrant will include the strengths of the product that are important to consumers. This is usually what needs to be aired in communications.

The product features in the upper left area are important to the audience, but are not associated with the brand. Their improvement should be worked on first.

3. The power of the brand - conversion-retention

Goal: choose a promotion strategy (emphasis on reach or conversion).

Which "....." brands do you know at least by name?

Which "....." brands have you purchased in the last 3 months?

In addition to knowledge and preference, the ratio of these metrics is important. The Conversion-Retention chart shows how well a brand or product converts its own familiarity into sales.

The map of brand power is a frame of reference for management decisions. It shows in which direction marketing should work.

4. Image - multidimensional scaling

Objective: to examine the image and perception of the brand.

How much do you agree with each of the statements about "....."?

Respondents are asked to rate brands on a list of characteristics. For example: classic, traditional, kind, technological, aggressive.

The responses are summarized in cross-tabulations, which are scaled, that is, the data are mathematically transformed so as to display them on a two-dimensional (easy to perceive) scale with minimal loss of information.

5. Product properties - conjoint analysis

Objective: To obtain an unbiased estimate of the importance of product features.

Many respondents tend to distort the importance of features when asked directly. For example, many like to underestimate the importance of price.

In such cases, indirect methods are helpful. The most advanced of these is conjoint analysis. His idea is to bring the respondent closer to the real situation of choice. The target audience is asked to evaluate not a divided list of brands and features, but the probability of buying a particular product.

It can be a prototype, a rate card, or a drawing of a shelf in a supermarket. Each person estimates the probability of buying 5-7 items whose properties and price are different.

By correlating estimates, properties and prices using regression equations, you can determine the weight of each characteristic and its levels, as well as the synergistic effect of the interaction of properties.

In the automotive industry, the brand and color of a car are important attributes, but their power is tripled when it comes to a black Mercedes or a blue Subaru.

It's cheaper to measure than not to measure

By 2018, it has become much cheaper to collect and use consumer information than not to do so. Actual customers will be pleasantly surprised by the attention paid to their opinions. Potential customers are highly likely to follow your website after filling out the survey to learn more about the survey author.