a digital product firm

The ‘Analysis’ Category

Groupon, the Super Bowl, and Buzz

Thursday, March 3rd, 2011

This is a companion article to “Groupon Really a Super Bowl Traffic Loser?”.

In a previous article, I took a look at the assertion that Groupon’s traffic lift from its Super Bowl ad was inconsequential.

Here I want to take a look at a different part of the Fast Company article on Groupon’s Super Bowl performance: Buzz.

The article shows a graph put out by Nielsen showing that Groupon had a very large amount of buzz relative to other digital brands that advertised during the Super Bowl. While I’m not a huge fan of sentiment (it’s still very much an inexact science), it seems clear that some concept of sentiment is missing from the FastCo article. If the majority of buzz is bad, then it wouldn’t be surprising to find that actual direct response to the ads was low.

More importantly, however, this reinforces how important it is to not live by buzz metrics alone. It’s easy for people to say something about your brand one way or another. What is more telling is what they actually choose to DO. Lots of talk (even very positive talk) doesn’t guarantee any impact on your brand. People may have had an interaction, fired of a tweet about it, and then totally forgot about it. Sure, there would appear to be more opportunity for people to do something, and certainly more opportunity for brand impressions on a person’s followers, but the relationship between buzz and actual response is not straightforward, and likely differs not just between brands, but between kinds of efforts inside a brand.

It’s best to measure not just what people say, but what you are trying to get from a campaign so that you can get a sense of what kinds of buzz actually move the needle on your business, and what kinds are just hot air. And, of course, if you can’t do that, we can do it for you ;)

Groupon Really a Super Bowl Traffic Loser?

Thursday, March 3rd, 2011

I was just reading this article on Fast Company comparing the response to Groupon’s now infamous Super Bowl ads to that of other advertisers (notably GoDaddy), and thought it was worth pointing out a couple of things.

First, the article uses % increase in traffic week-over-week as the primary metric. That certainly makes for a dramatic looking metric, but it probably doesn’t give you the insight into “success” that you might think it would. First of all, for this kind of performance, you need to be looking at raw numbers. Percentages are great for comparing things that are directly comparable, but if we’re looking at direct-response-type metrics (which site visits would be), then what you care about is how much you had to spend to get a volume of response. So knowing that Groupon’s 3% increase – which seems paltry next to HomeAway.com’s 27% – accounts for 420,000 visits (as pointed out by commenter Rob Day) compared to HomeAways’s 230,000 provides a more meaningful comparison.

And if you’re looking at direct response, then what you need to know is what the cost of the effort was versus its value.

Assuming that the cost of a Super Bowl commercial is $10 million (and I have NO idea what the actual cost was), then we’re looking at about $29 per visit. Is that worth it? Well now we’re into the realm of things we probably can’t know. To know for sure, we’d need to know what percent of visitors Groupon is able to convert on their site, and what the lifetime value of each conversion is.

By way of a hyperbolic example, let’s look at Groupon’s numbers versus GoDaddy’s. GoDaddy (again, according to commenter Rob Day) gained 4,100,000 new visits compared to Groupon’s 420,000. If they were both able to convert 10% of visitors, then GoDaddy will have gained 410,000 customers, where Groupon gained only 42,000. But if the lifetime value of a customer is $10 for GoDaddy, and $100 for Groupon (again, numbers pulled out of thin air to make a point here), then GoDaddy has added $4,100,000 in revenue, and Groupon has added $4,200,000, making their overall performance about equal, despite very different seeming traffic percentages.

A final note on this: If they each paid $10 million for their spots, then their ROI is going to be the Return divided by the Investment, or $4,100,000/$10,000,000 for GoDaddy, and $4,200,000/$10,000,000 for Groupon. That’s 41% ROI for GoDaddy, and 42% for Groupon. If neither got any additional value out of the ads, then they are both underwater on this effort – that is, they spent more than they earned – making this not a good candidate to repeat for either one of them next year.

I also want to talk about the “buzz volume” metric in the FastCo article, but I’ll do that in a separate article.

The Naked Numbers

Tuesday, August 3rd, 2010

This is part of a series of articles about how Fight is approaching using some free advertising. It all kicks off here.

The advertising portion of the Naked Campaign has now drawn to a close.

We divide the execution of a project (versus the strategizing portion) into 5 stages: Execution Planning, Running the Project, Analysis, and Adjustment.

Since we’ve finished running the project, it’s time for some analysis. Here’s a quick it on the base metrics that we were tracking, with more to follow.

Unfortunately, we ultimately had to throw out all of the Google AdWords data from before 20th as it looks like the click fraud that we were experiencing went all the way back to the beginning of the campaign :( So here’s the adjusted data starting at the 9th of July for LinkedIn’s DirectAds, and the 20th for Google’s AdWords:

DirectAds Spend: $199.24
DirectAds Impressions: 104,229
DirectAds Clicks: 38 (0.036% Click Rate)

AdWords Impressions: 116,092
AdWords Clicks: 78 (0.067% Click Rate)

Twitter Followers (current, all types): 295 (+14% since start)
RSS Followers (7 day avg): 15 (-17%)
Unique Visitors (30 days): 547 (+2.4%)
Comments (campaign, total): 0

I’ll do a deeper dive into the numbers over the next few days, and then we’ll go into the Adjustment stage and see what this all means for Fight moving forward.

DirectAds Finally Works!…Sort of.

Tuesday, July 27th, 2010

This is part of a series of articles about how Fight is approaching using some free advertising. It all kicks off here.

After 9 days of running ads on LinkedIn’s DirectAds, we’d seen a total of one click. This despite the fact that more than 16,000 impressions had been run, the fact that we took our top-performing ad from Google’s AdWords and put it into rotation on DirectAds, and the fact that we wrote the most pragmatic ad that we could think of.

Here we had free money, and we couldn’t spend any of it.

So we did what we do here at Fight. We looked to see what we could do differently.

After a brief brainstorm, we decided to open up the audience target. We had been targeting people in Marketing job functions at medium-to-large companies and were likely to have control of a budget. Since we had to spend the money or lose it, we decided to go wider; much wider.

We blew away all of the audience targeting options; anyone at on on LinkedIn could see our ads, moving us from about 500,000 potential audience members, to over 70,000,000. Our justification for this (aside from having to spend money by July 31st) was that it’s possible that marketing professionals just don’t use LinkedIn that much (seems unlikely, but our low number of overall impressions is suggestive), and that if we reach beyond the primary network of the people that we’re directly targeting, into the secondary network of people in a company who might refer us along, and the tertiary network of influencers outside of a company, then DirectAds could still be a benefit to the Naked Campaign.

Here’s what happened:

There was an immediate uptick in both impressions and clicks. The click through rate is modest, and the bounce rate for people arriving at the Naked landing page from DirectAds is about 67%, which means that there are likely SOME visitors who have a legitimate interest in what Fight does.

And what ads are getting all of these new clicks?

Maybe it’s no surprise to you, but it’s the top performing ad from the AdWords campaign and the pragmatic ad written specifically for the LinkedIn audience.

What the Click?!?

Friday, July 23rd, 2010

This is part of a series of articles about how Fight is approaching using some free advertising. It all kicks off here.

Last Friday I checked in on the performance of our AdWords ads and things seemed to be going well. There was even a gentle increase in the total number of clicks. When I checked back in on AdWords this Monday the number of daily clicks had more than doubled!

Hooray, right?

Not so fast.

Driving traffic to your site is not just an issue of quantity. Even people casting the widest nets still want to know (or SHOULD want to know) how the traffic is performing. So I went to Google Analytics to start with the Bounce Rate. The Bounce Rate tells you how many people made it past the first page of your site. It is a good first indicator that the traffic that you’re getting to your site is valuable.

And this is where the mystery began.

The traffic wasn’t there. AdWords showed 20 clicks on July 17th, and 25 on the 18th. Google Analytics? 7 and 12 visits respectively. Clearly something was wrong here.

It turns out that most of our clicks were coming from Google’s Display Network, and mostly through what shows up in the reporting as “AdSenseForMobileApps.com”, which is how Google reports ads that were displayed on mobile devices inside of apps. Sadly for us, the apps that were the biggest click generators (by far) have all been suspected of click fraud in one forum or another. Click fraud is when a site (or app in this case) generates clicks as Google AdWords would define them, but doesn’t actually send real traffic to the site that paid for the ad.

Fortunately, Google tries to keep on top of this, and there’s even a report (if you dig for it) that will show you how many clicks Google thinks were invalid and that it didn’t charge you for. They also seem pretty open to you reporting times when you think there is fraud, and crediting you back for disputed clicks.

For our campaign, however, the problem is more than a monetary one (since we’re working on free ads in the first place); they’ve destroyed our data. This issue calls into question much of the data that we’ve been gathering, since we can’t tell for certain how long we’ve been getting fake clicks. Since there’s no easy way to see what ads were served because of which keywords onto which placements, it makes it near impossible for us to determine if ads that we thought were high performing actually were, or were just lost into the honey pot.

So, what to do?

In AdWords, you can exclude sites from carrying your ads. You can even do this on an individual app or URL basis. For us, we excluded the entire AdSenseForMobileApps.com category, because not only do we have a data problem, but we also now have a trust issue. We can’t be sure what is a valid source of traffic from apps, so we’re shutting them all down. It means starting over with our data, but it may at least mean we have clean data.

It’s been a couple of days since we shut down the apps (we put a halt to it on Monday the 19th). What was the effect?

Clearly we have some work to do.

The SEO Effect

Tuesday, July 20th, 2010

This is part of a series of articles about how Fight is approaching using some free advertising. It all kicks off here.

One of the benefits that we factored in when we considered doing a campaign of this sort is search engine optimization. Doing the Naked Campaign meant that we could write about a lot of the things that we talk about internally in a public-facing way. Naturally that means that words that we care about would show up related to us in search engines.

There is, of course, no guarantee that we will show up high enough in any particular search to make it worth the effort from that point of view alone (and certainly there’s more to ranking high than just having particular words in your content), but the potential side effect was definitely worth factoring in.

So, is it working?

In short, yes. Since the Naked Campaign started, keywords that show up exclusively in articles related to the campaign have accounted for 20% of our organic search traffic.

Here are the phrases that have sent us traffic so far:
Adwords
madebyfight.com/naked
invalid http response code adwords
free $250 dollar google adwords
keyword suggestion tool
linkedin directads coupon

Especially interesting is the fact that the most popular keyword (AdWords) also had the lowest bounce rate at 40%. This means that 6 out of every 10 visitors from this search term went on to read further into the site, which is extraordinary for our site generally speaking, much less for search terms. This isn’t particularly surprising as our program is built around AdWords (and the lesser-searched-for DirectAds), but it is encouraging to know that the targeting appears to be effective.

Perhaps you’ve come to this page via a search? Let us know your thoughts in the comments below. What caught your interest enough to click, and did you find what you expected to find?

First Stats and Ad Optimization

Wednesday, July 14th, 2010

This is part of a series of articles about how Fight is approaching using some free advertising. It all kicks off here.

It’s time to check in on how our ads are performing. Let’s take a look.

LinkedIn has served 8,095 impressions and delivered 0 clicks. We pay only for clicks, so this would be ok if we saw an uptick in traffic that we could attribute to LinkedIn. But we don’t. So this is a problem. The problem here is that we may not be being pragmatic enough with our ads. The closest one to speaking directly to what we do in the most straightforward sense is “We Are Fight”.

DirectAds only lets you run 10 ads as part of the same campaign and we’re using all of them, so we’re going to take the ad that seems the farthest from the mark and replace it with one is trying to be as on the mark as possible: “Marketing Strategy. We do this every day. Follow the Naked Campaign to see how.”. The ad that seems like it would be the least effective is: “Do you read these ads?”. Nowhere in the ad does it say what we do, so it can easily come off as a research panel.

Of course, DirectAds has no easy way to retire an existing ad (you can turn it off and hide it, but it still takes up one of your 10 buckets), so we’re going to have to edit the ad that we want to get rid of to have the copy for the new ad we want to put in. We’ll see how that effects the stats for the ad (we updated the ad when it had 871 impressions and no clicks), but it’s too bad their tool wasn’t a bit more capable. I’d love to have the option of turning the old ad back on if the new ad doesn’t perform.

Also, since the ads have gotten no clicks, we’ve spent none of our credit, so we’re raising our daily spend from $10 ($250/25 days) to $13.38 ($250/18 days). We’re well above the suggested CPC bid range at $5, so we’ll hold steady there and see what happens.

By contrast, AdWords is ticking along steadily, serving slightly fewer impressions but garnering 11 clicks (for a click through rate of 0.14%). It’s too early to say what ads are performing well as we’ve had clicks across 9 of the 22 ads in rotation. Of those 11 clicks, 8 of them came from Google’s Display Network, which is comprised of all of the sites that use AdSense. If you are letting AdWords manage which sites get to see your ads (as we are), then you need to have a minimum level of volume before Google will tell you what sites got how many impressions. We’re not yet to that point, but when we see where they’re going, we’ll let you know.

Of the 3 clicks that came from the SEM proper, somewhat surprisingly they came from some of the simplest, broadest terms that we’re using (we’re currently running 87 search phrases): “analytics”, “campaign”, and “AdWords”. In some senses, based on the content of our ads, this is not surprising, and Google agrees sending these terms substantially more impressions than most of the rest of them.

For AdWords, we’re going to let things sit for a while longer until more obvious trends emerge, but here are some early returns on the best performing AdWords ads sorted by CTR:

The top performer will probably jump out at you right away as it took only 13 impressions to get a click! It will be interesting to see how that one continues to perform. Worth noting is that the reason that it has received so few impressions relative to anything else, was that it originally had “Google” in the title, and was rejected for it, so it didn’t start running until very recently.

Goal Stats

DirectAds Impressions: 8,095
DirectAds Clicks: 0

AdWords Impressions: 8,013
AdWords Clicks: 11

Twitter Followers (current, all types): 255 (-1.2% since start)
RSS Followers (7 day avg): 17 (-5.6%)
Unique Visitors (30 days): 541 (+1.3%)
Comments (campaign, total): 0

Collapse Is Simply The Last Remaining Method of Simplification

Friday, April 2nd, 2010

Clay Shirky Fight Portland
Clay Shirky

We are very lucky to have Clay Shirky amongst us as he brings an incredible knack for being able to explain, very simply, seemingly complex problems in the digital arena. His latest post, The Collapse of Complex Business Models, covers his thoughts on the transition to the web for TV companies and producers. Before addressing their concerns he takes a moment to reflect upon Joseph Tainter‘s book, The Collapse of Complex Societies where Tainter looked at several ancient, sophisticated societies that suddenly collapsed. As it turns out, it was bureaucracy that ruined those societies – “In such systems, there is no way to make things a little bit simpler – the whole edifice becomes a huge, interlocking system not readily amenable to change.”

He then turns to the TV producers, news content providers and their issues with the web:

About 15 years ago, the supply part of media’s supply-and-demand curve went parabolic, with a predictably inverse effect on price. Since then, a battalion of media elites have lined up to declare that exactly the opposite thing will start happening any day now.

To pick a couple of examples more or less at random, last year Barry Diller of IAC said, of content available on the web, “It is not free, and is not going to be,” Steve Brill of Journalism Online said that users “just need to get back into the habit of doing so [paying for content] online”, and Rupert Murdoch of News Corp said “Web users will have to pay for what they watch and use.” Diller, Brill, and Murdoch seem be stating a simple fact—we will have to pay them—but this fact is not in fact a fact. Instead, it is a choice, one its proponents often decline to spell out in full, because, spelled out in full, it would read something like this:

Web users will have to pay for what they watch and use, or else we will have to stop making content in the costly and complex way we have grown accustomed to making it. And we don’t know how to do that.”

How Mobile Phones Are Changing Social Media

Wednesday, March 31st, 2010

Resh Sidhu Flowtown Infographic
Click on image for full size view.

I came across this infographic via Interactive Art Director, Resh Sidu, who had posted it to her blog. Hat tip to @simonmainwaring for tweeting it up.

Resh points out an interesting stat: “25% or more than 100 million Facebook users access from a mobile phone, and those who do, are twice as active on social networks compared to people accessing from a computer. The 35-54 year old bracket is the most active mobile social users, bet some of your clients wouldn’t have expected that.”

Where Are the Basic Twitter Tools?

Tuesday, March 9th, 2010

Deep down I’m a numbers guy.

When Fight does something in the world (either for ourselves or for our clients), I want to know what happened, and, to me, that means numbers.

What I can’t get my head around (through no shortage of banging my head on it) is why there aren’t any basic numbers around stuff we do on Twitter.

I’m not looking for anything particularly complicated, and maybe that’s the problem.  I just want Impressions and Reach.  But I’ll be darned if I can find them anywhere.

Impressions – Impressions is the number of time that your message was seen by a person. It doesn’t matter if it’s the same person seeing it over and over, or if it’s one unique person for each time the message is viewed. For this reason (among others), I’m actually not a big fan of this metric, but it’s simple, and it has equivalents across all kinds of media.

On Twitter that would be (for any given tweet), the number of people who follow you directly, plus the number of people who follow lists that you are on, plus the number of people who follow anyone who retweets (either in the traditional or new fangled way) your tweet, plus the number of people who follow lists that your retweeters are on.

This will give you an idea of how many people could have seen your message (not all of them did, of course, and others will see your tweet without following anyone (like, through search, or the like)).

Reach – This is the total number of individuals that saw your message. It removes any times where a person saw your message more than once. In the Twitter world, you would keep track of every unique individual you run into when counting impressions.

Reach is especially handy when applied to Twitter as a person who follows you directly might also follow, for example, a list that someone who retweeted your message was on.

More than abstract measures of “Klout” or the like, these numbers tell you what actually happened when you tweeted something, which is the gateway to understanding which things that you tweet are resonating with your audience, and which are not.

But the trouble is, I can’t find them anywhere – at least not calculated like I would. And beyond these are yet more interesting metrics that could be generated, but aren’t (again, as far as I can find) and a whole awesome concept of Twitter-based CRM that I’ve yet to stumble on (though, admittedly, I haven’t looked recently).

Am I just digging in the wrong place here? Do tools exist that measure this and I’m just missing them (@twitalyzer? @webtrends? Speak up if I’m missing it)?

Can anyone help a numbers guy out here?