4 mistakes to avoid when representing data in content
/
If you’re in the business of creating content, data and statistics are your bread and butter (or, at least, they should be). The analysis of data is what enables you to draw insights about your target customers, your market, and trends in your industry, among other things.
If you aren’t generating your own data through proprietary research, you at least need to use third-party or secondary data to back up any claims you make in your content.
How you gather these insights and what you do with them are ultimately what matter most. When using your research to create data-driven content, it’s important not only to use it correctly but also to use it in good faith.
Unfortunately, not every business knows how to use data properly in written content, especially when using proprietary data to back up their own arguments and observations.
Without due diligence, it’s easy to misrepresent your research accidentally, draw erroneous insights, or worse — give in to the temptation to make the data say what you want it to say, even if it doesn’t.
Since we’ve been writing data-driven content for years, we’ve picked up on a few of the most common mistakes people make when trying to represent data in content. Here’s what to look out for.
1. Drawing Conclusions Based on Assumptions
Be careful not to stretch your data simply to confirm your own assumptions and biases. Just because the data implies something you believe doesn’t mean it fully supports it. This disconnect will be noticeable to your readers.
Let’s say you run a survey of millennial consumers and the results from one question suggest that 23% of the people in this age group still live with their parents. That’s a significant number which, if correct, indicates this age group is having vastly different life experiences than their predecessors.
However, the media often takes such statistics and uses them to prop up claims they don’t fully support: ‘23% of millennials live with their parents, which suggests this age group is lazy and entitled,’ is not a reasonable claim. While there may be some millennials who could be qualified as ‘lazy’ and ‘entitled,’ and who also happen to live with their parents, the latter does not automatically signify the former assumption.
Instead, you must dig deeper into the results of this data. What do the answers to other questions in your survey reveal about this phenomenon? If you don’t have any data that is applicable, you could draw conclusions based on secondary sources about this age group to come up with insights.
Zillow actually ran the numbers and came up with the ‘23%’ figure in a study of census data from 2005 to 2016:
(Source: CNBC)
The CNBC report that covered this story drew upon secondary data to reinforce the argument that financial challenges like massive student loan debt and high housing prices are what contribute to this trend. There was no mention millennials’ laziness, or lack thereof, in the report.
2. Obfuscating Your Findings To Support Your Argument
Misleading people with statistical information and data is nothing new, as almost anyone can attest.
Darrell Huff published the book How to Lie with Statistics in 1954, and almost everyone who has ever lived in a consumer economy has been subjected to a sales pitch involving questionable data.
Because a world of information is now available at the click of a button, anyone can conduct their own fact-checking session to determine if an article or salesperson is telling the truth. Nonetheless, people still try to bend, alter, and obfuscate data to support their arguments in the hope that their audience will take them at face value. Most often, they are eventually called out on it, and it damages their reputation.
Obfuscating your findings is pretty easy to do. Take this graph of the Global Land-Ocean Temperature Index from NASA, for example:
(Source: NASA)
Based on the information presented, you could make the argument that global temperatures were relatively stable between January 2000 and January 2010. Without any further analysis, you could even state that global temperatures often have long stretches of stability, implying that much of the fuss about the climate crisis is overblown.
But if you zoom out on that same graph, you get a very different picture:
(Source: NASA)
This graph covers a period roughly between the Second Industrial Revolution and today, and it shows a very clear warming trend. In fact, the short period that we referenced before is almost inconsequential by comparison.
Although the graph in the first example contains legitimate data, and although saying “the global climate has long stretches of stability” is technically true, you’d be obfuscating the truth from your audience by including this information in your report.
You can do this in a business context as well.
For example, let’s say you learn from a study that 90% of companies are planning to invest in a type of workforce management software in the next year. Since your company sells workforce management software, this information could present your readers with a sense of urgency, boosting your sales.
However, if in another question 87% of your respondents tell you they’d only be interested in such software if it is mobile-optimized, these results could harm you if you’re still in the process of developing your mobile solution. You could leave out the data about respondents’ desire for mobile-ready software, but you’d be misleading your audience.
This isn’t just dishonest; it’s a poor use of your resources.
If you’re investing time and money to do research, you should use that research to its fullest extent, even if it gives you results that don’t favor your original argument. In fact, sometimes results that don’t favor your original argument are the best results of all. Your research has helped you realize something you didn’t know previously, and your audience will no doubt want to know, too.
3. Delivering Your Findings in Bad Faith
Although obfuscation is an example of delivering findings in bad faith, messaging can also play a role, as can the actual structure of your research.
Any good content marketer will tell you that the needs of your audience (and by extension, your customers) come first — always. Although your end goal may be sales and revenue, you can’t create awareness-level content from a sales perspective. Instead, you must provide your audience with information they can actually use.
When doing research which will yield data to be used in marketing content, this means asking respondents the questions your audience would want to ask — not just the questions you want to ask.
Once you’ve acquired your data, you must look for insights your audience would find interesting.
For example, your survey may suggest that people with college degrees earn more in their lifetimes than people without college degrees. This is interesting information and worth knowing, but if you’re delivering your findings to undergraduate students at your school who haven’t decided a major yet, wouldn’t it be pertinent to tell them which types of college degrees lead to the highest lifetime earnings?
(Source: Visual Capitalist)
Not every student will make their degree choice based on salary alone. But knowing this type of information would help any student make a more informed decision.
4. Presenting All Results and No Analysis
Lastly, it’s essential that you not only present your findings, but also the conclusions readers can draw from them. Drafting a report is hard work, and your audience will certainly want to draw their own conclusions based on your findings, but it’s your expert analysis of the data that adds real value to your content.
Make your content easy to digest by including a “Key Takeaways” section either at the beginning or at the end of the report. When possible, you should use visual information, callouts, and sub-sections to communicate important information in the report.
These are just a few of the challenges we’ve come across when writing data-based content. If you’d like to learn more about how we turn data into actionable content, don’t hesitate to contact us.
