This article discusses the potential of segmentation in Facebook advertising.
Why is segmentation needed?
Segmentation is one of the most fundamental concepts in marketing. Its goal is to identify the best match between the firm’s offering and the market, i.e. find a sub-set of customers who are most likely to buy the product and who therefore can be targeted cost-effectively by means of niche marketing rather than mass marketing.
There are some premises as to why segmentation works:
- Not all buyers are alike.
- Sub-groups of people with similar behavior, backgrounds, values and needs can be identified.
- The sub-groups will be smaller and more homogeneous than the market as a whole.
- It is easier to satisfy a small group of similar customers than to try to satisfy large groups of dissimilar customers.
(The list if a direct citation from the Essentials of Marketing by Jim Blythe, p. 76.)
While segmentation is about dividing the overall market into smaller pieces (segments), targeting is about selecting the appropriate marketing channels to reach those customer segments. Finally, positioning deals with message formulation in the attempt of positioning the firm and its offerings relative to competitors (e.g., cheaper, better quality). This is the basic marketing model called STP (segmentation, targeting, positioning).
How to apply segmentation in Facebook?
I will next discuss three stages of Facebook campaign creation.
1. Before the campaign
There are a few options for creation of basic segments.
- generate marketing personas (advantage: makes you assume customer perspective; weakness: vulnerable to marketer’s intuition, i.e. tendency to assume you know your customer whereas in reality you don’t)
- conduct market research (advantage: suited to your particular case; weakness: costly and takes time)
- buy consumer research reports (advantage: large sample sizes, comprehensive; weakness: the reports tend to be very general)
- use Facebook Audience Insights (advantage: specific to Facebook; weakness: gives little behavioral data)
The existence of weaknesses is okay – the whole of point of segmentation is to gather REAL data which is stronger thana priori assumptions.
Based on the insights you’ve gathered, create Saved target groups in Facebook. These incorporate the segments you want to target for. If you are using an ad management tool such as Smartly, you can split audiences into smaller micro-segments e.g. by age, gender and location. Say you have a general segment of Women aged 25-50; you could split it into the following micro-segments by using an interval of five years:
- women 25-30
- women 31-36
- women 37-42
- women 43-48
The advantage of micro-segments is more granular segmentation; however, the risk is going too granular while ignoring the real-world reason for differences (sometimes the performance difference between two micro-segments is just statistical noise).
After creating the segments in Facebook (reflected in Saved target groups), you want to test how they perform — so as to see how well your assumptions on the effectiveness of these segments are working. For this, create campaigns and let them run. In Power Editor, go to the Custom audiences (select from the sliding menu), select the segments you want to test and choose to create new ad groups. (See, now we have moved from segmentation into targeting, which is the natural step in the STP model.)
NB! If you particularly want to test customer segments, keep everything else (campaign settings, creatives) the same. In Power Editor, this is fairly simple to execute by copy-pasting the creatives between ad groups. This reduces the risk that the performance differences between various segments are a result of some other factor than targeting. Finally, name the ad sets to reflect the segment you are testing (e.g. Women 25-31).
2. During the campaign
After a week or so, go back to check the results. Since you’ve named the segments appropriately, you can quickly see the performance differences between the segments. To make sure the differences are statistically valid (if you are not using a tool such as Smartly), use a calculator to determine the statistical significance. I created one which can be downloaded here.
When interpreting results, remember that the outcome is a combination of segment and message (and that the message is a combination of substance and tone, i.e. what is said and how it is said). In other words,
Result = segment x message, in which message = substance x tone, so that
Result = segment x (substance x tone)
Therefore, as you change the message, it reflects to performance across various segments. This means that you are not actually testing the suitability of your product to the segment (which is what segmentation and targeting is all about), but the match between the message and the target audience. Although this may seem like semantics, it’s actually pretty important. You want to make sure you’re not getting a misleading response from your segment due to issues in message formulation (i.e. talking to them in a “wrong way”), and so you want to make sure it reflects the product as well as possible. Ideally, you’d want to tailor your message based on your ideas of the segment, BUT this is prohibited in the early stage because we want to make sure the message formulation does not interfere with the testing of segment performance.
How to solve this problem, then? Three ways: first, make sure the segments you are testing are not too far apart – i.e. women aged 17 and men aged 45 subjected to the same message can create issues. Second, try to formulate a general message to begin with, so it doesn’t exclude any segments. Third, you could of course make slight modifications to the message while testing the segments — here I would still keep the substance (e.g. cheap price) stable across segments while maybe changing the tone (e.g. type of words used) depending on the audience – for example, older people are usually addressed in a different tone than the younger audience (yo!).
Finally, one extra tip! If you want more granular data on how different groups within your segment have performed, go to Ad reports and check out the data breakdowns. There is a wealth of information there which can be used in creating further micro-segments.
3. After the campaign
What to do when you know which segments are the most profitable? Well, take the results you’ve got and generalize them into your other marketing activities. For example, when you’re buying print ads ask for demographic data they have on readers — it has to be accurate and based on research, not guesses — and choose the media that matches the best performing segments according to your Facebook data. In my opinion, there is no major reason to assume that people in the same segment would act differently in Facebook and elsewhere (strictly speaking, the only potential issue I can think of is that Facebook-people are more “advanced” in their technology use than offline-people, but this is generally a small problem since such a large share of population in most markets are users of Facebook).
There you go – hopefully this article has given you some useful ideas on the relationship between segmentation and Facebook advertising!
I’m into digital marketing, startups, platforms. Download my dissertation on startup dilemmas: http://goo.gl/QRc11f