We love data. It helps inform everything we do, backs up everything we say, and helps demonstrate our success. But often the challenge is how to gather that data in the first place. Analytics is a great start, but what if you need more qualitative data?
Most of the data we can collect with web tracking is quantitative: bounce rates, conversion rates, shopping funnels, etc. All of these numbers are created by analysts for analysts, but they make normal marketers confused and uncomfortable. They want to focus on quality and experience – the human factors – not numbers and statistics. So they end up relying on their gut-feeling. In comparison, qualitative marketing research are seen as the bearers of human attitudes and opinions.
But what if we merge qualitative and quantitative data about the same customer? What can statistics and human attitudes tell the company about the perception of their brand, special offers, trust to marketing channels, product preferences, loyalty?
That is exactly what The Economist set out to do with their latest online survey aimed at collecting the qualitative data about their customers’ behaviour. By merging this data with on-site behavioural analytics, they can predict the likelihood of different potential customers buying different products.
When I saw a pop-up offer on Economist website to take a part in their qualitative survey, I started answering questions purely out of curiosity. I went through 20 or 30 questions just answering the questions like normal people would do. But then I reminded myself that I am an analyst, so I went back through the whole survey again to better understand marketing and data intentions behind the survey questions.
Here are a few observations I would like to share with you.
- Survey pop-up.
- Prize as an incentive for participants to complete the survey.
- High-level customer segmentation based on the perceived likelihood to buy
The first question segments responses based on their likelihood to buy with the current packages and offering. This will help them understand where there is an opportunity to upsell and what might create new subscribers, as well as assessing the value of individual responses in developing new products and packages.
For example, if a respondent said that they have never and would never consider subscribing to the Economist, their responses will be less useful in devising new subscriptions packages – they’re not going to subscribe anyway!
- Investigating product preferences
The value of print publishing is declining – the costs are high and the readership steadily declining. Yet some people still much prefer print, deriving value from the tangible. With this question, The Economist can spot trends in publishing, helping them to assess the value in continuing their print magazine.
- Evaluating the level of trust in different marketing channels
As preferences in the delivery method (print or digital) changes, so too do the preferences for promotion. It’s at this point that analysts will be able to draw connections between different segments of respondents. For example, do people who do or would subscribe and who want digital versions of the magazine therefore prefer email contact? Or perhaps their inbox is oversaturated and they would rather a self-service approach via social media?
- Customers are invited to take part in product development
- Assessing the perceived value of different packages
In contrast to the first question (current likelihood of subscribing) this question measures the allure of the potential new packages. They might not buy based on the current packages, but now they know a bit more about what each package and service provides, and have had the chance to create their ideal package, they may be much more likely to buy.
- Customer perception of The Economist’s brand
Prestige looks to be the central brand offering and USP for The Economist. This question assesses how well that brand positioning is currently conveyed. It would also be interesting to see the impact of perceived prestige on the likelihood of a respondent subscribing, or whether it has any impact at all. Is prestige a useful brand position in terms of converting subscribers?
- A set of socio-demographic questions
Further segmentation and insight can be generated with the use of demographic information. Notice that these questions are included at the end of the survey. Asking these types of questions too early can cause concern, while asking at the end generates a higher response rate simply because the respondent has invested time into answering the previous, more interesting questions.
- And, finally, call-to-action for participants to leave their e-mails
By gathering email addresses The Economist’s analysts can attach them to customer purchase history and make a very detailed audience segmentation for up-sell/cross-sell according to customers’ likelihood to buy. They also know exactly how each respondent prefers to receive promotions, so they can target people who prefer emails via their email address.
The incentive helps here by providing a good reason for gathering email addresses: to inform the voucher winner. If you’ve completed the survey purely for the chance to win the voucher, you will still be as keen to provide your email as those hoping to receive new promotional packages from The Economist.
It’s a simple model with great potential value.
There are, however, a few considerations you should make before running anything similar:
- Questions perceived as boring will make people leave without completing the survey. You have to keep them engaging, interesting and/or of potential benefit to the respondent.
- The number of people who have completed the survey should be sufficient for further analysis. The response rate for surveys can be very low, so if you’re only getting a few thousand unique visitors per month, you might not generate enough data to provide useful analysis.
- People like me who click on the survey out of curiosity and mess all of the results should be filtered out based on anomalous engagement (time on site/page). The issue of clean data is one of the most consuming in data mining & predictive analytics so shouldn’t be written-off as irrelevant in the planning stages.
Want more insight? Give me a shout – I am more than happy to discuss anything and everything analytics!