Steps to design a neuromarketing project following scientific methodology
Neuromarketing has been the subject of several blogs, videos, conferences, etc. However, with the exception of a few scientific studies and courses, there is limited information on the methodology employed to carry out a neuromarketing study correctly. At the time of executing a neuromarketing study, you must consider the following factors:
Briefing with the client
To satisfy expectations with a neuromarketing study, the first step is to understand very well the objectives of the client. The fact that neuromarketing is a recent field must be taken into account, and it is probable that the client will not even mention neuroscience, rather focusing on marketing, business and/or strategy.
More familiarized with traditional market research, the client will probably focus on aspects such as:
Background information and strategic motivations
Main objective of the study
Secondary objectives of the study
Profile of the desired sample
This is more than enough for a good market research professional to propose a plausible scenario and elaborate a research methodology, a calendar, and a budget. However, in neuromarketing, it is recommended to obtain more information so that the proposal is well adjusted. It is especially interesting to know:
Type and amount of materials/experiences to evaluate
The number of stimuli to evaluate will determine the duration of the study, the effort required to interpret the results and, therefore, the budget.
Deadline for the delivery of materials to be evaluated
In a neuromarketing study, it is necessary to design the study with software and is always highly recommendable to carry out a preliminary pilot study with 2-3 participants. It is not possible to change a study once it is launched, and therefore preparation is a key aspect. Having the materials beforehand is necessary for good preparation.
Possible segments to analyze
On one hand, enlarged samples are recommended to obtain representative results per segment. On the other hand, each segment can almost double result interpretation efforts. Both points directly affect the execution calendar and the budget.
Deadline for delivery of results
Neuromarketing studies generate a high amount of data (at least one datum per 0.2 seconds of exposure for each metric). If the delivery times for the results are very tight, it is advisable to shorten the scope of the study or the number of metrics to be obtained, and then meet the deadline for presentation of results without compromising the quality of interpretation.
Although hard to achieve, an approximate range of the budget that the client is willing to invest in a study can reduce considerably the number of interactions between client and provider. Usually, with such a new discipline, clients are not aware of what can be provided by a study, nor of the costs associated with adding a new objective or segment. In traditional research, this is something that would not generate any impact on the budget, but in neuromarketing this could mean the addition of new technologies or tests. If the provider counts with a tentative budget, the client can be guided to focus on the results that will provide more value while reaching the objectives, and discarding those results that are “nice to have” but increasethe budget considerably.
Sample to include
Similarly to traditional market research, the characteristics of the sample are set according to the interests of the client. There are two typical questions about the sample in a neuromarketing study:
Is it necessary to exclude people with some type of mental disorder and/or who have undergone pharmacological or psychological treatment previously to the study?
Depends basically on the brain decoding algorithms (computer programs that translate brain signals into emotional and cognitive metrics):
If the algorithms are generalist (assuming that all brains are similar), then it is recommendable to exclude any type of profile that would provide a different brain response. However, in practice, almost all people have different brain responses and therefore these algorithms are not recommended unless very large samples are employed (+400 people).
If the algorithms treat data individually for each participant (such as Bitbrain’s machine learning algorithms) then there are no issues at the time of including any types of profiles, as the algorithms adapt to the brain response of each participant when collecting data.
What is the minimum recommended sample size?
Depends on the decoding algorithms. If algorithms are generalist, very large samples are required (+400 people) to achieve reliable results. Nevertheless, when machine learning algorithms are employed, the sample can be significantly smaller. To better understand the effect of the sample, let’s take a look at the following figure, which shows the effect of sample size (groups of 10, 20, 30, 40 and 50 people) on the replicability of the computation of emotional metrics, in an image visualization paradigm with 30 images.
Graphic obtained from GSR results utilizing the Usenns database (participants distributed around the world), visualizing the same calibration images.
It can be observed that the replicability of results in groups with 10 people is generally low, with very variable results in the majority of images. This is similar to what occurs in groups with 20 people. With 30 people, slightly higher replicability begins to be observed, and with 40-50 people, variability is quite acceptable. It must be highlighted that the images evaluated were calibration images. In other words, these images were selected so that all participants responded in a similar emotional manner. Therefore, with 40-50 participants homogeneous responses were captured in the population, but this homogeneous response could not be present in specific stimuli and samples.
Technologies to include
At the time of selecting the technologies to include in a neuromarketing study, we have to consider these main aspects:
Quality of technologies: The neuromarketing techniques that register physiological responses (electroencephalogram, galvanic skin response, eye-tracking, etc.) are like a “microphone” that we place on our bodies. If the “microphone” has low quality, we are not able to capture the “sound” with quality, and therefore the metrics obtained and the consequent interpretation are not reliable. Aspects such as quality of the components (electronics, cables, sensors…) are fundamental for the technologies to measure correctly. The number of sensors, and especially where they are placed, is also crucial. For example, in neuromarketing it is not necessary to place sensors on the motor cortex because it is not interesting to decode aspects such as intention of movement. However it is necessary to place sensors on the frontal, parietal and occipital locations to capture emotional and cognitive reactions.
Briefing with the client: it is necessary to comprehend very well the technology characteristics, strong points, and limitations to have a clear understanding of which metrics can be provided by each one. For example, the following figure depicts the most popular technologies included within Bitbrain’s laboratories, with the metrics provided by our software:
Maybe this is one of the most complicated points at the time of designing correctly a study, as it is necessary to have basic notions of experimental design in neuroscience. The approach requires a good understanding of the objectives of the study, as these objectives will condition the design. This is why briefing is so important: the design will have to answer a series of questions, and incorporating news questions later is not as simple as in a traditional market research study.
Moreover, there are some basic design aspects to be considered:
Controlled versus real surroundings
Will the study be carried out in a laboratory (controlled context) or in real surroundings (real context)? If in a real context (such as a supermarket), how will it be carried out? As previously mentioned, the more ecological the study is, it is more difficult to know why the results are produced, and although it is possible to isolate different variables, this makes the study more expensive.
What type of materials are being evaluated (spot, packaging, product, jingle)? Are they very similar or different? Processing one type of information or the other is different within the brain (for example, auditory and visual information). Therefore the design is also carried out differently. If materials are very similar (for example, two identical spots with different songs), special care must be taken during experimental design to control learning biases, which could occur when the participant sees for the second time a very similar stimulus. Again, these biases are controllable but complicate the study (also making it more expensive).
Tasks for the participant
What is the attitude of the participant during the study? Is it passive (only listening or visualizing something)? Or is it active (he/she must carry out a task)? When participants have a passive attitude during the study (for example, watching a spot, look at packaging or listen to advertisements), there is complete control over the interaction, which is comparable across participants. In this way, it is possible to aggregate results for the entire sample and simplify the analysis of results. However, when each participant has to carry out a different task (for example, navigating the internet or stroll through a store), it is necessary to analyze the behavior individually, increasing the complexity of the study, the analysis, and costs.
In general, market investigation institutes do not count with neuroscience experts that can design good experimental protocols from scratch. However, there are laboratories such as the ones provided by Bitbrain, that include templates and pre-designs for the most common neuromarketing studies and experimental protocols. This free the researcher from concerns on this aspect.
Organization of fieldwork depends considerably on the internal capacity of the research institute, but it is perfectly feasible to carry out a study with 40 participants in approximately one month, considering:
One week for recruitment
Note that recruitment occurs similarly to recruitment in a qualitative study, and higher gratifications are not necessary. Moreover, during this week the study is designed within the software and a small pilot study is carried out (2-3 people) to confirm that duration, protocol, and technologies all function correctly.
One week for field work (estimating one hour per participant, although times can be shortened if, for example, the institute counts with several laboratories and the study can be carried out with several participants simultaneously).
One of the advantages of neuromarketing is that field work does not necessarily require highly qualified personnel. In qualitative research, the experience of the moderator is crucial to obtain valuable results. However, within a neuromarketing study, field work consists in setting up the technology and verifying that everything runs correctly. With the most recent and advances neuromarketing laboratories, this work is very simple because the software indicates if the technologies are correctly placed, and manages the entire study. If any issues occur, the technician is warned. Therefore, little expertise is required to carry out fieldwork for a neuromarketing study.
Two weeks to obtain metrics, interpret results and elaborate the report. This is a very important step and is explained in details in the following two sections.
This step can be very simple or very complex. Basically, it depends on the type of laboratory employed by the research institute. If the laboratory does not provide the metrics automatically, this step requires specialized personnel to analyze the physiological signals. However, if the laboratory computes the metrics, there are no special requirements. It must be highlighted that the quality of metrics is very important, and laboratories with custom decoding algorithms and adequate experimental protocols are preferred.
Interpretation and elaboration of results report
Along with briefing and experimental design, this is one of the most relevant points in a study and probably one of the most complex. Neuromarketing provides objective metrics on the emotional and cognitive reactions of consumers to different marketing stimuli (spot, packaging, brand, etc), and all these data can fill sheets and sheets of paper. However, to provide true value to the client, the researcher must be able to:
Understand the information globally and be capable of summarizing it.
Understand why the results were produced taking into account emotional aspects or possible biases that affect our behavior and decision making.
Comprehend whether the results are positive or negative for the marketing strategy.
Detect areas of improvement and learnings for the future.
In summary, although neuromarketing is said to be an objective research technique, the truth is that, despite the objective data produced, interpretation depends heavily on the researcher. Therefore, at the time of selecting a neuromarketing company to carry out a neuromarketing study, brands should not only worry about the technologies utilized by the institutes, but also consider the knowledge, experience, and ethics of its researchers.