Audi's Social Media Fail

Mar 03, 2011

By most accounts, global carmaker Audi's high-profile Twitter campaign was a success. Great press, thousands of Twitter mentions, $25,000 going to charity, and glowing quotes to put in a press release. But for some of us, it was a failure.

Which is a pity, because it could've been something truly innovative and meaningful. Hopefully, it can at least be a case study for other big brands to learn how to truly engage social media.

Background

Hawaii designer John Garcia was a finalist in the contest, and I interviewed him as things were just getting started. But after holding the lead through the last three days of the five-day contest, John came in second place — news that took several days to announce, despite previously daily updates.

It's hard not to see my disappointment, or that of his many other fans, as simple sour grapes. There's probably an element of that involved, and no one likes a sore loser. But sore or not, this loser can't resist making a case for what went wrong... in the hopes that next time, everyone — brands and fans — can do it right.

What Should Have Happened

Audi was very, very proud of its social media plans. They said their Super Bowl promotion was making "social media history." A Twitter hashtag in a Super Bowl ad was to launch Twitter and Facebook contests that would "allow fans to join the movement to redefine luxury."

Ten finalists were selected, and the "Finalist Round" involved, basically, promoting the contest. Like any marketing campaign, the objective was to get people talking about the brand across social media platforms. So a straightforward point system was set up:

Retweets on Twitter: one point each.
Views of a Twitpic photo: two points each.
Comments on a Twitpic photo: one point each.
Views of a YouTube video: three points each.
Comments on a YouTube video: one point.
Blog post about the campaign: ten points.
Comment on a blog post: one point.

Audi was trying to spread its message across several channels, awarding points for promoting them, and points for evidence of engagement from everyone's followers.

In a perfect world, the contest would've kicked off a symphony of multimedia, fueling a thriving conversation about Audi and progress and how the two went well together. And the points would reflect how successful the finalists were in promoting the contest.

What Actually Happened

John kicked butt. He got thousands of retweets on Twitter. He put together great, striking YouTube videos that got hundreds of views and dozens of comments. He posted blog updates, and several local bloggers also wrote about his efforts. We featured his campaign on Hawaii Public Radio (which Audi lauded as the contest's first "old media" coverage), and his success merited after-the-fact coverage on KITV. Someone even wrote him a musical jingle!

There was content all over the place, cheering John on, and promoting Audi in the process. The contest closed with a flurry of Twitter posts, and with John still at the top of the leaderboard.

It took a long time to get the official word, but the official word was that John placed second. Audi's fans gave a polite cheer, and moved on. But John's fans were left scratching their heads. Nobody can argue that giving $25,000 to charity isn't something worth celebrating, nor that John's second place prize won't still do a lot of good. But after several hopeful conversations about how a big brand was recognizing the value of social media, it seemed like something wasn't right.

Many of us wanted Audi to release the numbers on which they based their final rankings. The official rules were straightforward, and the point system was clear. By our reckoning, the math was simple, and worth sharing... both so we could understand what happened, and to use as a valuable resource for developing future social media campaigns.

But after dozens of happy Twitter posts encouraging all the finalists and bragging about the contest, Audi fell silent on the matter. Finally, Audi social media manager Laurie Mayers asked Russ Sumida to follow her so she could send him a private Direct Message. And that message was:

"Hi, I appreciate your passion, & @johngarcia did a great job, but @audi is not obliged to release scores & doesn't plan to."

That was that. It was left to us to figure everything out, and that meant we could only speculate and analyze the results based on the information available to us.

And according to my math — and that of most observers — it all came down to Twitpic views.

Number Crunching

Here are some rough numbers to illustrate the point, covering the top three finalists. Note that these are based on the current information, almost two weeks after the contest ended, and are only as accurate as the tools used (like Topsy). But this way you can check the data yourself, and I'm confident in saying the information is representative of how things were going and felt during the contest.

Again, the numbers are not accurate, but should be comparable relative to each other. Without the official tally, this is about as good as it's going to get.

Here's Twitter:

@jetsetbrunette (2,144 followers, Klout 50): 89 mentions.
@johngarcia (4,562 followers, Klout 63): 2,621 mentions.
@tommytrc (101,570 followers, Klout 71): 593 mentions.

Here's YouTube:

@jetsetbrunette: 1 video, 63 views, 0 comments.
@johngarcia: 2 videos, 759 views, 54 comments.
@tommytrc: 7 videos, 579 views, 18 comments.

I'm not sure how many blog posts, and thus blog comments, were written by or about each contestant, so let's just use flat out Google result counts for a general sense of the visibility of each contestant's entry:

@jetsetbrunette (#ProgressIs #RoastPig): 1,970 results.
@johngarcia (#ProgressIs #DeskDiary): 3,990 results.
@tommytrc (#ProgressIs #OldJacket): 995 results.

Finally, Twitpic:

@jetsetbrunette: 11 photos, 221,109 views, 20 comments.
@johngarcia: 1 photo, 102,984 views, 199 comments.
@tommytrc: 1 photo, 49,799 views, 15 comments.

Whew. Easily a 100,000 point difference from the Twitpic view counts alone, which pretty much wipes out any other metric used to measure success. @jetsetbrunette won by a country mile. The question of how John came in second is settled, as far as I'm concerned.

But there's a lot more worth discussing.

Why View Counts Suck

It takes a fair amount of effort to create a video, or write a blog post. A tweet isn't hard to put out there, but you have to have engaged and supportive followers to pass along the message to their followers in a retweet. People who comment on videos, blog posts, and photos at least care enough to leave their mark... even if it's just a "Yay!"

But a Twitpic view means simply that someone looked at the page. They might not have cared one bit about the contest. They might not have even known what they were going to get when they clicked the link. And, yes, you can rack up views simply by hitting "reload" on your browser over and over again. Or set your browser to refresh the page automatically.

There was a lot of reloading and refreshing going on in this contest, but frankly, it wasn't entirely clear whether that was even a problem. It was clear early on, though, that Twitpic views were going to be the way this contest was going to be won.

After the first day of the contest, I was stunned by the fact that one Twitpic had 18,343 views. Even more stunning was the rate at which it was climbing. Five minutes later, there were 18,633 views. That was about 58 views a minute. About an hour later, it had hit 20,000 views.

Eventually, I posted the third comment to the photo: "5,000 views in the last two hours? That's... something!" The contestant then explained that her parents had forwarded the link to their family business' extensive network. "Theyve tweeted it to every vendor/customer!" Sure enough, she was in first place at the next update.

Today, the photo has 93,369 views, and seven comments — one from me, and three from the same person.

In my opinion, comments, not views, are a much more useful metric. They're a great measure of engagement, which is what really counts. Though it was not a factor in Audi's point system, here's another way to look at it:

@jetsetbrunette: 1 comment per 11,055 Twitpic views, 0 YouTube comments.
@johngarcia: 1 comment per 517 Twitpic views, 1 comment per 14 YouTube views.
@tommytrc: 1 comment per 3,319 Twitpic views, 1 comment per 32 YouTube views.

The Aftermath

Today, at long last, John Garcia posted a detailed recap and analysis of what happened. You should read it. He remains gracious, and positive, and hopeful. I want to be John when I grow up. But in his post, he notes that Audi did, in fact, issue a warning about inflating page views:

Please note, it has been brought to our attention that Twitpic views may have been manipulated to increase page views. If evidence of this practice is discovered, you may be disqualified, in Sponsor’s sole discretion.

This demonstrates that Audi was very familiar with a key way one of their metrics could be gamed, and warned the finalists not to exploit it. We don't know exactly what practices were frowned upon (was hitting "reload" over and over again okay, but using browser scripts forbidden?), or whether anyone was disqualified.

But Audi clearly knew that Twitpic views were the weak point in the contest, yet it was the primary way the winner was determined.

That's a failure.

Audi reaped the benefit of thousands of messages sent on its behalf by contestants and their individual communities of friends and followers, but fell silent when reasonable questions were raised. The campaign was a "success" under the old way of thinking, but not in terms of engagement, responsiveness, accessibility and transparency.

That's a failure.

After all the great advance buzz about Audi and its Superbowl hashtag, my view of the Audi brand has instead dimmed. This could've been a great exercise for both big business marketing and grassroots movements.

Meanwhile, I want to continue to support local contestants in national or global competitions, something Hawaii residents are always eager to do... but this experience makes me wonder whether it's worth annoying people who are uninterested in such things with hundreds of posts and plugs. I feel bad that I might not cheer quite so hard next time for an otherwise deserving Hawaii candidate.

Kudos to John, at least, for staying positive through it all, even when his friends and fans so passionately felt that something was not right. I look forward to helping him do good with the prize he did win... and meanwhile work alongside him and everyone else in the local social media scene to continue to demonstrate what progress really is.

Lessons Learned

For all that went wrong with this contest, I hope other companies continue to explore social media and engage audiences through fun contests or other initiatives. I would offer the following pieces of advice to avoid Audi's mistakes:

Use meaningful metrics. Pageview counts that can be inflated simply through reloads or scripts are meaningless. YouTube videos and blog posts, on the other hand, show at least some level of thought and effort. Comments illustrate reach and how engaged someone's followers really are. And as with any online popularity contest, there are simple technical measures you can take to minimize cheating — limiting votes or view counts by IP address, requiring a login or email confirmation, etc.
Be willing to admit mistakes and adapt. Audi knew that Twitpic views opened the door to abuse, and we don't know if they ever acted on the threat to disqualify people. Rather than being rigid and focusing on "the nuclear option," they could have simply changed how Twitpics views were weighted, or thrown them out. Write contest rules that allow for on-the-fly adjustments.
Be responsive. Respond to comments, monitor and participate in discussions, and don't disappear once everything is over. Ignoring negative messages and pushing people to "take it offline" seems suspect, and deprives you of the chance to demonstrate publicly your willingness to hear reasonable comments (and let critics demonstrate whether they're being unreasonable). And flooding a channel with promotional messages, only to withdraw once you've gotten what you were after, is like attending a party only long enough to shill for Amway.
Be transparent. Bolster your credibility by making data public and allowing independent auditing. Especially if it's a points-based contest, numbers shouldn't lie. If you want to have some discretion for choosing a final winner that better fits your objectives, simply include that in the way your promotion is structured.