Big data, especially the right data, has the potential to completely transform how companies communicate with their customers and fans. With new technology and tools like sensors and beacons, we can track every aspect of a customer’s online and offline interaction with a brand, and use that data to customize and curate content and promotions. Many customers are willing to share their data with brands in return for personalized experiences and offers while still being respectful of personal boundaries.
In a recent survey by SDL, 79% of respondents said they’re more likely to provide personal information to brands that they “trust.” So how do brands ensure they remain trustworthy in the eyes of customers who are becoming more concerned about their privacy and the collection and use of their data every day? For starters, don’t let your marketing be invasive, incessant and downright creepy. There’s a fine line between being pervasive and being invasive.
Second, always be transparent when collecting data and let customers know what data you’re collecting, why you’re collecting it and how you plan to use it. And, always give them a choice to opt out of it. Most importantly, though, once you have their personal information, make sure you secure it and protect it from hackers and data breaches.
Third, don’t assume your big data is always right. Bias, inaccurate or low-quality data, unreliable data sources, technical glitches, not seeing the big picture, and a lack of tools and resources to analyze large volumes of data often lead to data blunders that shatter customer trust in a brand and make for extremely damaging PR debacles.
Here are 7 examples of big data blunders that made news over the last couple of years. While many of these examples are cringeworthy, there are important lessons that can be learned from them to help prevent data fails for your company.
1. Target Figures Out Teen is Pregnant – Before She’s Told Anyone
Target set off a maelstrom of outrage and privacy concerns a couple of years ago when the retail behemoth angered an unsuspecting father by sending discount coupons for cribs and baby clothes to his teenage daughter who hadn’t yet revealed her pregnancy to her parents.
The New York Times (via Forbes) spoke to Target’s statistician Andrew Pole (before Target asked him to stop speaking to the press) who explained how big data helped them sort customers’ purchases to figure out if they’re pregnant even before they actually had a baby.
It came down to a unique ID number linked to a customer’s credit card, name and other info, which saves deep data on the customer’s purchases. That data is then mined to look for patterns (for example: pregnant women bought unscented lotion at the start of their second trimester) to then send them special deals and coupons for baby items. When Target’s data mining system analyzed customer purchase data based on 25 products that pregnant women frequently buy, it could assign a pregnancy prediction score to each shopper and estimate her due date so Target could send her relevant coupons for various stages of her pregnancy.
Deep data on customers has the potential to be creepy and unnerve them. Companies need to find ways to reduce the “stalker” factor in their marketing. Following this incident, Target started mixing up their customized offers so a discount coupon for a coffee maker showed up right next to one for a crib and an ad for wine glasses showed up next to a coupon for baby clothes, so someone looking at it couldn’t tell if the recipient of the deals was pregnant.
2. Bank of America Credit Card Offer Calls Customer a “Slut”
Lisa McIntire, a journalist living in California received a credit card offer from Bank of America that was addressed to “Lisa is a Slut McIntire” and sent to her mom’s address. After her mom saw the mailer and texted her to let her know about it, McIntire posted pictures of the letter on Twitter and it went viral in minutes. The credit card offer was sent by Golden Key International, an honor roll society that McIntire had joined, which is also an affiliate marketer for Bank of America.
It’s important for marketers to qualify their data before they launch campaigns. Don’t blindly trust that the data is clean, problem-free and accurate. Don’t be in a hurry to just use a segmented contact list without doing some due diligence first to scrub the list and look for potential issues before you make a fatal mistake that can hurt the business in a big way.
3. OfficeMax Sends Mail to Father Mentioning That His Daughter Died in a Crash
A month before Bank of America sent the credit card offer above, OfficeMax sent discount coupons to a customer and the envelope was addressed to “Mike Seay, Daughter Killed in Car Crash.” Seay’s 17-year-old daughter Ashley had died in a car crash along with her boyfriend a year before. A company spokesperson doubted Seay’s claims initially, until a local TV station aired a story on it. After the story went viral on social media, hundreds of people criticized the stationary giant. Company executives didn’t apologize directly to Seay until the Los Angeles Times posted a story online two days later. Even after a company executive apologized to Seay, they still didn’t explain why an office supply company knew that his daughter had died. Seay felt it really wasn’t any of their business to have this deeply personal piece of information on a customer.
In addition to being ethical and transparent about data collection, it’s equally important to only collect the data you really need. To avoid haystacks without needles, have clear objectives for your data collection. To begin with, much of the data we analyze already rests on erroneous models making mistakes a common occurrence. When we collect huge volumes of data we don’t really need, the errors that are made can be exponentially damaging. And, if you do make a mistake, apologize immediately. Customers’ trust is almost impossible to win back once it’s lost.
4. Romney’s Orca System Did The Campaign More Harm Than Good
Big data was the buzzword of the 2012 presidential election and both campaigns tried to out do each other in launching sophisticated data analytics systems designed to measure everything, power data-driven decisions and give them a competitive edge in the race. Obama’s campaign hired an entire team of data crunchers who worked out of “the cave” in Chicago, headed up by chief scientist Rayid Ghani.
Mitt Romney’s campaign deployed Orca, a much-hyped mobile data analytics platform that was supposed to give them deep insight into what was happening at polling stations, which in turn was supposed to help campaign volunteers direct get-out-the-vote efforts in battleground states like Iowa, Pennsylvania, Ohio and Florida. Named after a killer whale, Orca turned out to be more of a beached whale, plagued by technical issues and bugs. While Orca isn’t the primary reason that Romney lost the election, the technical problems and huge waste of money and resources did the campaign more harm than good. Many volunteers who used the system complained that it was built in a hurry, confusing to use, mostly unstable and that they couldn’t get it to work from the field in many states as the system crashed repeatedly. One website even reported that there was, in fact, “a massive suppression of the Republican vote by the Romney campaign, through the diversion of nearly 40,000 volunteers to a failing computer program.”
The Romney camp denied that there was any “massive failure” except for some minor glitches and that Orca had no relation to the outcome, but that’s hardly the right response even if it were true. Considering the money, resources, volunteers and time that was put into this project, they should have invested some of those resources to test the system and make sure it was foolproof and bug-free before they unleashed it into the wild. The 2016 presidential election is around the corner and there are valuable lessons to be learned here for both parties.
5. Nude Webcams & Facebook Ads Teens Weren’t Supposed to See
Earlier this year, Facebook made news for showing highly inappropriate ads and content to teens, some of whom were only 13 years old. Some of the ads promoted an app called Ilikeq that lets users rate other users’ “attractiveness” with sleazy ads mostly featuring semi-clad women. Facebook’s automated, data-driven ad system configured some of these ads to reach young teens. Fourteen-year-old Indiana resident Erica Lowder’s picture ended up on display to adult men online who were able to vote on it and users of the Ilikeq app could click through to the teen’s Facebook page. Lowder’s mother and thousands of people were up in arms about how Facebook was putting their children at risk by selling ads to “weird apps and sites that open kids up to terrible things.”
Due to the sheer volume of ads published daily, Facebook’s social ad system is mostly automated and users’ social data populates millions of ads every day. Advertisers can create campaigns and target ads to reach users based on a wide range of user information collected including age, location, relationship status and interests. When systems like this use incredibly large volumes of data, it’s almost impossible to identify problematic content and ads without more manual checks and balances in place.
6. CNN Reports 110% Turnout in Scottish Independence Vote
While the expected turnout for Scotland’s independence referendum last month was pretty high, CNN took it to a whole new level. Many of CNN’s computer generated interactives and graphics for their live event coverage are powered by data analytics and in this case it looks like the data was overly optimistic. During a report on the referendum, CNN displayed a graphic that showed that 110% of Scotland’s population had been polled. According to the graphic, 58% of Scots had said “Yes” and 52% voted “No.” The graphic was obviously off on the total voter percentage, but it was also wrong about the fact that the majority of Scots actually voted “No.”
The erroneous graphic quickly went viral on social networks with Twitter users mocking the network for the miscalculation. CNN updated the graphic later to a more accurate one showing 52% voting “No” and 48% voting “Yes,” but the damage was already done. Quick lesson here: check and double check your data and don’t just assume it’s accurate. Performing some basic data quality checks before displaying information might mean that you don’t hit the airwaves first, but at least you’ll get it right.
7. Pinterest Accidentally Congratulates Single Women on Their Weddings
Pinterest is a huge favorite with brides looking for wedding planning ideas and inspiration. Many brides use the site to visually curate everything including their wedding gowns, make-up, hairstyles, decor, cakes, table settings, bridesmaids dresses and invitations. And they often share these image boards with friends and family, too. While most social websites now collect some amount of data on their users to serve up relevant content and ads, Pinterest may have taken that to an awkward new level. The website accidentally emailed some users congratulating them on their upcoming weddings and offered a special on wedding invitations. Except, many of these women weren’t getting married, and some of them were actually single. Many of the unhappy recipients took to Twitter to share the erroneous email.
The email subject line was “Hundreds of wedding invitations (P.S. You might find the one)” and the email itself said: “You’re getting married! And because we love wedding planning — especially all the lovely stationery — we invite you to browse our best boards curated by graphic designers, photographers and fellow brides-to-be, all Pinners with a keen eye and marriage on the mind.” A company spokesman issued an apology citing bad data and that they were sorry they came off “like an overbearing mother who is always asking when you’ll find a nice boy or girl.”
Big data still has big issues. While big data is supercharging marketing campaigns and helping drive conversions, it’s important to win your customers’ trust and ensure the data is reliable and accurate, that the data analysis isn’t biased or missing the context, and that the tools and technology you’re using to do this are foolproof and glitch-free. Mistakes are inevitable, but it’s important that we learn from them and find ways to improve how we communicate with our customers.
Data collection is a two-way road, helping a company better monetize and serving personalized experiences to users. Don’t let poor data, unethical collection or lack of due diligence create a data memory the web will never erase.