Do social media ads really work? We put them to the test!

Twitter: Clickworthy yet costly

Twitter has just beefed up its ad analytics.

Setting up ads on Twitter is fast and painless. I submitted a single promotional tweet linking back to my home page, with a $25 per day budget and a per-click cost between a penny and $1.50.

My ad attracted two new followers.

Twitter's fees are set dynamically, but the variability built in to the rate is confusing and disarming. My first click cost $1.30, but I'm not sure why, and Twitter doesn't do a good job of explaining how rates are set. However, Twitter did suggest that my ad might generate between 62 and 104 clicks per day. (I skipped the option to pay more to promote my Twitter account directly.)

My ads began showing up within minutes. There doesn't appear to be any manual approval process. Reporting is simple and straightforward.

Twitter results.

Ultimately, Twitter did not perform as well as promised. After four days, the ad received just 67 clicks, under a quarter of the 248 minimum clicks promised for a four-day run. The ad was promoted 7529 times, which works out to a great clickthrough rate of .89 percent. That's double the rate I got with Google AdWords, and at a CPC of 94 cents, it was the least expensive investment of the bunch on a per-click basis, too. (I also got two new followers on Twitter along the way.)

The verdict: The total expense of $62.87 is pricey for such a small number of impressions, but the clickthrough rate is still impressive.

StumbleUpon: Cost-effective without clickthrough

StumbleUpon's ad results were surprising.

At StumbleUpon you don't bid for impressions. Instead, you pay a flat rate to have your ad displayed to "stumblers" as they use the website. You don't write ad copy and you don't even have to rely on an ad being clicked upon. The page you're advertising just appears in the Stumble flow of users as they click along. Ad impressions cost a flat 10 cents each, an amazing bargain.

I paid $100 up front and set a $25 daily limit for my ad to be displayed to users with an interest in business topics, then submitted my Null Media home page for review. StumbleUpon's evaluation took about six hours, at which point my ad began displaying. (Make sure you submit each page you're advertising to the regular StumbleUpon system as well, because the ad reporting tool tracks both paid and organic Stumbles.)

StumbleUpon ads: A sleeper hit?

Because no bidding system or variability is involved, StumbleUpon's tracking is very straightforward. Over four days, my ad was displayed 1001 times (or 250 times a day at 10 cents each), and my website received a surprisingly high 397 organic stumbles on top of that. While I paid $0.10 per paid ad view, the "effective" cost per view including the organic stumbles was just $0.072 per view.

The verdict: These are very cheap ads, but the effectiveness of advertising a B2B service on a heavily consumer-focused service is also questionable. Click the Reports link under your Account profile to generate a more detailed look at who looked at your ad and for how long.

The takeaway

First, the results of this experiment are hardly scientific, given the small budget and limited amount of time. Plus, my ads targeted business buyers, not general consumers. A test of ads promoting a hangover remedy product or a novelty wristwatch would likely have yielded far different results. That said, here are the most helpful observations:

  • Facebook ads appear particularly ineffective at getting clicks, possibly because users have already become accustomed to ignoring these ad placements due to their location. Experiment with low CPC bids on Facebook when getting your feet wet.
  • LinkedIn is easily the most expensive service on which to advertise, and clickthrough rates are low.
  • Twitter's almost 1 percent clickthrough rate is double the rate of AdWords, and far and away the best I encountered. Ads are surprisingly cheap. It's difficult to target your ads on Twitter, but they still appear effective despite this limitation. Perhaps that's because promoted tweets are virtually impossible for users to ignore.
  • StumbleUpon is a great option for extremely inexpensive ads, and if you are targeting a broad swath of consumers, it's definitely worth a look.
If you're sold on social media advertising, consider these tips before you begin.
  • Start small. Set a very low budget for a CPC (not CPM) ad, and let it run for as long as you can to get a sense of how it's performing. Tweak the ad frequently and track your results. Some services let you run multiple versions of the same ad, so you can compare results among them.
  • Clicks aren't everything. Despite getting hundreds of clicks over hundreds of thousands of impressions, my business didn't net any new clients from these ads. Understand what value your clicks actually bring to you, and make sure your ROI is positive after you have sufficient history to examine.
  • Think about your target market. It's an old adage that "Twitter is for business" and "Facebook is for fun." My experimental results bore this out, so think carefully about what social networks your potential customers are likely to be using before charging ahead.

How I Tested

How effective is a social media billboard?

The budget for these tests was small and the timeline compressed, so consider these results to be anecdotal instead of scientific. It's impossible to run the identical ad across all social networks, but I tried to keep as many variables as consistent as possible. For example:

  • I allocated a maximum budget of $100 for each social network, to be split up over four consecutive days, $25 per day. (Most services did not use the full budget.)
  • I ran a baseline of four days of Google AdWords advertising, with the same budget, before running the social ads. I then shut off AdWords.
  • I designed all advertising to send traffic to the home page of my business, Null Media, and not to various social network profile pages, in order to better compare those results with AdWords' results.
  • I tested each social network consecutively, and ads were programmed so they did not overlap.
  • If an ad network did not use up its ad budget for a day (due to the vagaries of the way ads are presented and bid upon), I did not extend the length of the experiment. Shortcomings in budget expenditures are noted.
  • I tracked results using the services provided by each social media site.
  • When at all possible, I used both the same criteria for the desired viewers of an ad (such as location, demographics, and occupation) and the same ad copy and imagery.
  • I kept all social media accounts active by posting fresh content, two to three times daily, across each account. Some content linked back to the Null Media target domain; some did not.
Shop ▾
arrow up Amazon Shop buttons are programmatically attached to all reviews, regardless of products' final review scores. Our parent company, IDG, receives advertisement revenue for shopping activity generated by the links. Because the buttons are attached programmatically, they should not be interpreted as editorial endorsements.

Subscribe to the Daily Downloads Newsletter