In-Stream Twitter Advertising: Does It Work?

As Twitter continues to boom, Twitter advertising is slowly starting to emerge as a hot button issue, at least when delivered by third parties.

Marshall Kirkpatrick on ReadWriteWeb on April 10 covered advertising through one of a number of services that is offering Twitter advertising, Be A Magpie. Marshall originally attributed the ads to major companies, which Allen Stern at Centernetworks pointed out was incorrect: the ads highlighted for companies like Apple were affiliate ads (RWW has since amended the post.)

The notable thing is that affiliate marketers are using Twitter advertising to begin with. But does it work? I signed up for a Be A Magpie advertiser account to find out.

Before we start

In testing Be A Magpie, I’m not endorsing the service nor the form of advertising. I haven’t signed up to run ads on my Twitter stream, nor do I have any intention of doing so. I don’t think spamming your followers is a clever thing to do if you value your Twitter account, but having said that there’s a fine line between spam and self promotion; that’s not an argument I’m going to cover here.


Signup is straight forward. You add your email address and a password, and you have an account. To start a campaign, you’re asked to nominate a budget, provide a Tweet for the campaign, select keywords and set a maximum CPM. After you’ve set up a campaign, you fund your account via Paypal (in Euros only) and wait for approval.

The test

I submitted two campaigns for a total spend of 50 Euros (about US$74): a tweet for a bacon wallet on Amazon, and a tweet for Quantum of Solace on iTunes. I targeted the keywords fun and movies.

After the first day it appeared that the Quantum of Solace campaign wasn’t converting as well as the bacon wallet campaign (the figures following show I was wrong). I also knew that Amazon sets cookies that result in subsequent sales; that is, if someone visits the bacon wallet, then acquires another product (for memory over 7 days) I would receive a cut from those sales because they had the cookie, so I flipped most of the ads to Bacon Wallet.

The results

The CPM is based on the number of people who potentially read the Tweet. For example one tweet on an account with 1000 followers = seen by 1000 people.

Quantum of Solace had an exposure of 3652 followers for 119 clicks, or a CTR of 3.25%
Bacon Wallet had an exposure of 13022 followers for 344 clicks, or 2.64%

Number of direct sales on both: 0.
Number of indirect sales: difficult to tell. The Quantum of Solace links did nothing at all, where as my Amazon account shows just over $10 in affiliate payments this month, although at least half are not related to this campaign. Let’s say $5 in affiliate sales.

Result: doesn’t work.


Testing an affiliate program on Be A Magpie has its hazards: do you go low and hope that the chances of landing a sale are higher, or do you go for a high value product and hope that you might land a small number of sales, but get more for them.

I went low, so I can’t vouch for a high end ad. However if the aim is to convert tweets to sales, your chances have to be better with a lower price product (spontaneous purchase.)

There were other problems with the way the tweets went out:

1. Some users label the tweets as ads, some don’t.
From an advertisers perspective, the ones not labeled ads work better (and the stats showed higher CTR on the accounts that didn’t label the tweets). However, disclosure of advertising is a good thing in terms of transparency.
2. There were quality control issues with the accounts used
At least one account seemed to exist only to show ads, and the follower count was at best questionable. Another account wasn’t in English, so the Tweet likely hit people who don’t speak English, or at least as a primary language. This is a problem to considering that the target market for these links were Amercians; there’s no option so far to opt for Twitter accounts by country.

One positive: in one case, a Tweet was retweeted driving more clicks; that though means that the CTR could be lower because the system doesn’t track (from what I can gather) retweets by non advertisers (except where you use their tracking, I didn’t). It does mean you may get extra exposure though.


The mere presence of affiliate marketers on Be A Magpie suggests that there is money to be made using the system; well, either that or there’s a lot of people testing it. I can’t say that it doesn’t work for everyone, but it didn’t for me in this test.

The cost isn’t appealing: roughly 13.6 cents per click, and that’s not taking into account that the retweet of one ad may have delivered more clicks that would have been had from the paid tweets. Note I made both my ads conversational and appealing, a lot of the ads going out on the system look like ads and would likely have lower CTR.

If you simply wanted to drive traffic to a site, StumbleUpon offers 5 cents a page view, with the added benefit that if the page is voted positively, you often end up with more page views than you paid for (so it’s cheaper again.) StumbleUpon might not do affiliate links, but offers far better value if you’re seeking traffic to a page.

For Twitter users, the economics of using your account for advertising using Be A Magpie are a little strange: An account with 4,600 followers cost me €10.29 to advertise on, and yet Magpie offers a variety of compensation models, from pay-per-sale, pay-per-lead, pay-per-click or pay-per-view for Twitter users. Only pay-per-view was available to me as an advertiser, so I wonder if they’re used to pay considerably less than 50% of the amount charged to the advertiser to the owner of the Twitter account (note: the 50% share is a guess, they simply don’t tell you what the cut is upfront.) Either way: you’ve got to question whether potentially losing followers by spamming them with advertising is really worth it.