Have a question?
Boomerang Ideas is a simple and quick way of doing a temperature check or getting detailed quantitative feedback on any topic with your desired audience. It can be a simple poll on a very distinctive socio-demographic (for instance ask people from 25-34 if they’d pay for CO2 compensation on their next flight) or a global validation of a certain naming or feature, so that you don’t get lost in translation, like the Mitsubishi Pajero debacle.
In order to do this, we have built a marketing automation that allows us to post on all major social media channels such as Facebook, Instagram, Snapchat and LinkedIn. Using micro-targeting technology, in the form of online ads, combined with our intelligent algorithms, our technology ensures that you get feedback from the right people. Once all the data has been collected, our business intelligence consolidates the findings, including individual comments.
In as little as 1 week (often less), you’ll receive a detailed report about your question from the people you wanted to get feedback from. This will include a full break-down with details such as location, age group or gender (subject to availability and offer) in order to see the opinions of several demographics.
- More accurate, as people are on their favourite channel. We rely solely on their curiosity as incentive (panels attract only a subset of the population and often have wrong incentives that lead to biases source)
- More affordable: Thanks to the accurate micro-targeting filters, screening questions are unnecessary, which will therefore save you money.
- Data entry is super simple and hassle-free.
Our “dynamic social media panel” is less biased (as most people have never been polled before), faster (it only takes less than a week) and cheaper (no screening question necessary thanks to micro-targeting filters).
It’s the most effortless feedback-mechanism you can get. No need to get your feet wet. You just give us your questions, and we’ll take it from there. The tests happen fully autonomous. Our marketing automation ensures to ask on all relevant social networks such as Facebook or Instagram.
Our methodology has been validated in a pre-study in collaboration with the University of Zurich. A full study with publication in 2024 is currently ongoing.
You can consider us a dynamic social media panel that always connects you with your desired audience. However, we only connect with social media users, thus there are a variety of biases or limitations to be considered:
- All answers will – obviously – only come from people on social media channels. If your required age range is below 13 (or below 18 in EU/CH due to Meta's new regulations as per November 2023) or way above 65, or you want to reach absolute non-tech people, social media is probably not the right channel to use. Social media surveys do not cover the entire population due to the nature of the channel, but do provide a good sample.
- A small fraction of responses (<1%) come in the form of comments as part of social media posts. It is the nature of posts that all past comments are visible if someone within the same micro-group wishes to see them, hence a potential bias might occur within this method. However, our Representative sample plan separates these posts into hundreds of posts specifically targeted at small micro-groups (e.g. same gender in same age group in same language and region), so it is less likely to happen.
- The same method also can lead to a minority of responses coming from respondents outside the targeted group. This happens when someone engages (share, like, comment) with this post.
- Your poll cannot include content that is prohibited (illegal, discriminatory, weapons, tobacco, adult, fake-news) or restricted (political content, alcohol, dating services, gambling, crypto). This one’s out of our hands, as it’s the social networks that make these rules (for a reason). Check out the full list from our friends at Facebook. Other networks follow a similar approach.
- All sampling methods allow representative weighting, which weighs the results from all clusters based on actual socio-demographic attributes of a population (so that, for example, no age group is overrepresented). While this increases the accuracy of the poll substantially, there are other attributes such as education level or political opinion which we currently do not take into account when creating our sampling and therefore may be unequally represented in certain groups.
- Currently, our surveys are limited to 6 questions.
Both Quick Check and Deep Dive Boomerangs give you access to our “dynamic social media panel” and are based on representative samples.
Quick Check allows for a representative temperature checks with your target audience. The less granular stratification give you faster results at lower costs.
Deep Dive allows for a more granular stratification of the desired population and therefore more accurate results.
A poll at Boomerang is a 1 question survey with 2-4 possible answers and up to 2 follow-up questions, while a survey can include up to 6 questions with several answer formats.
This option ensures a more detailed recruiting on every micro-cluster and a more accurate weighting of the data. Recommended for extra accurate and representative surveys.
In order to represent the target population as close as possible, we use a survey sampling method called stratified sampling – the gold-standard for sampling. This method divides the population into several groups, which are based on specific characteristics.
Characteristics such as:
- Geography (f.i. Greater regions within a country, such as Lake Geneva region, Zurich region etc.)
- Gender (f/m)
- Age group (in available cohorts of 13-17*, 18-24, 25-34, 35-44, 45-54, 55-64, 65+)
Then, within each group, respondents are randomly selected to participate in the poll (random sample) until we have enough responses. The groups findings are weighted according to official ground-truth data from each country's statistical office.
For in Switzerland, Germany and Austria we offer cross-quoted samples (also known as crossed quota samplings) which ensures to keep the ground-truth data not only on the overall level (simple weighting or "gleichverteilte Gewichtung" which ensures representative age-group distribution, representative regional distribution and representative gender distribution) but on all sublevel as well (e.g. representative age-group distribution within a region and within every age-group representative gender distribution). These samples offer the maximum of accuracy but, due to their complexity and higher costs, are only recommended for scientific or publication purposes.
- With regard to representative data we disclose the following: Representative on 3 dimensions (geography, gender, age group), weighted by ground-truth data from the statistical office and conducted with respondents on the social media channels Facebook, Instagram, LinkedIn, and Snapchat.
- Age group 65+ on social media: Micro-targeting does not allow granular age targeting beyond 65+ (e.g. 85-94 years olds) and therefore includes all persons above the age of 65. The group however is weighted according to the population of the age group 65-74.
- While 3 dimensions increase the accuracy of the sample substantially, there are other attributes such as education level or political opinion which we currently do not take into account when creating our sample and therefore may be unequally represented in certain groups.
* Unavailable in EU/CH due to Meta's new regulations as per November 2023.
Short answer: Generally, it's always an approximation, and it depends on the definition. We have to live with the sampling error, that we can only reach people who are online and on social media. The question is, what’s the alternative to get the opinion of a population?
- Telephone surveys, in which most often the landline is called. But who picks up the phone these days? As, mobile phones are mostly not integrated into the method, and even if they are, they are not the best alternative since some people have a personal and business phone which would over-represent them. In addition, people with more time – the ones that do not hang up – are also overrepresented (source: http://theconversation.com/heres-how-to-make-opinion-polls-more-representative-and-honest-117405).
- And then there are online panels* which are a fair way to compensate. However, wrong incentives (like monetary incentives) and also the uncommon interest in regularly participating in polls, lead to sampling bias, which isn't particularly ideal.
Back to social media, where all people that have Internet and a device (phone, computer) can participate on their favourite platform, just like that. In Switzerland, that’s currently 82% of the population daily, which isn’t too bad. To make sure we capture all needed parts of a demographic, we are using stratified sampling (the gold standard for representative surveys) in combination with micro-targeting on a variety of dimensions like age, gender and region.
* Online panels differ a lot in terms of quality and methodology. There are better ones like SurveyUSA or YouGov, average ones like Google Surveys or failed ones like SurveyMonkey according to FiveThirtyEight. If you’re into panels, we recommend visiting ProjectFiveThirtyEight or downloading their powerful comparison right away here.
We currently allow representative surveys in the following countries:
- Czech Republic
- South Africa
- South Korea
Representative surveys (with cross-quoted sampling) are available in the following countries:
Please find all the regional codes here.
Please contact us at firstname.lastname@example.org to get your country listed in our next release.
Our curated Global 10 list includes responses from the following countries:
- South Africa
We are planning to offer a feature where you can select your own 10 countries. If you’re interested, give us a shout at email@example.com.
As all surveys are done on social networks such as Facebook or LinkedIn and we have to comply to their rules. Profane language, incorrect grammar or prohibited or restricted content (drugs, alcohol, weapons, adult content, gambling, some crypto, dating) is not allowed and will stop your survey within several hours (with money-back of course).
In addition polls about social issues, elections or politics are also not allowed (except for Switzerland where we are allowed to do so; subject to timing and availability). Social issues are defined as civil and social rights, economy, environmental politics, immigration, political values and governance, security and foreign policy.
However, you don’t need to worry too much about this, as we automatically refund your credit card if this happens.
You can find more about the review process at the Facebook policies.
In the digital age, ensuring the integrity of online platforms and providing a safe environment for users is a priority. This is achieved using several strategies and tools. IP checks are utilized to identify and block users engaging in fraudulent activity or rule violations, as they allow detection of multiple accounts being used from the same address. Google's reCAPTCHA service is employed to protect against spam and abuse by distinguishing between human and bot interactions.
Additionally, platforms monitor user behavior to detect speeders and straightliners - respondents who complete tasks at an unrealistically fast pace or choose the same response for every question, respectively. These behaviors usually indicate a lack of engagement or an attempt to exploit the system. Lastly, individual links per cohort are used and they are closed as soon as the cohort is finished, preventing any form of late or inappropriate entries.
We are currently supporting payments from all major credit card issuers in CHF only (AMEX, Mastercard, Visa) as we are based in Switzerland.
Registered Boomerang accounts have the possibility to change to EUR. This can be done by changing the country in your profile.
Our surveys are based on feedback from real people from all demographics on their favourite social media channels. While this makes it extremely valuable and accurate, it also exposes our method to human uniqueness and therefore unpredictability.
Based on our experience, the Boomerang engine is well oiled to reach 100% of the promised amount of responses (e.g. 500 answers for the Representative Sample in less than a week) and this rarely is a problem.
If your survey doesn’t work (low engagement and therefore no poll results), we’ll stop the campaign after 24hrs and give you a full refund.
Data Privacy & Data Processing Agreement
By "new subprocessors", we're referring to the additional external companies we're collaborating with to carry out our research. Currently, we're working with a variety on subprocessors worldwide. For a current list see the Data Processing Agreement which is part of the terms & conditions.
These subprocessors help us in our research efforts by providing advanced capabilities. They enable us to leverage their technologies to derive more meaningful and robust insights.
With the help of these subprocessors, we can provide you with more robust insights and a better user experience. We believe these collaborations will significantly enhance the quality of our research, which in turn will benefit you.
We will only be sharing anonymized, non-personal data with our subprocessors. This means that the data has been stripped of any information that could potentially identify individuals, ensuring your privacy is maintained.
Absolutely. We stand by our commitment to data minimalism and security. Only non-personal, anonymized data will be shared with our subprocessors. Your personal information will never be shared, and your privacy will always be protected.