Archive for March, 2008

Experts Agree – Online Brand Advertising is Broken

Wednesday, March 26th, 2008

There have been a flurry of great blog posts lately about how broken brand advertising is online, mostly because we are trying to measure it the wrong way. My own contribution to this effort notes that click through rates are the “hammer” of online advertising, and when all you have is a hammer, well, you know the rest.

Dave Morgan has a recent one about vertical publishers, where he talks about brand advertisers:

Major brand advertisers are starting to take online very, very seriously; just look at the announcement earlier this week that General Motors plans to shift 50% of its ad spend online within three years. Just look at all of the pioneering online ad work done recently by big offline brands like Coca-Cola, Pepsi, Snapple and P&G. These brands are now making online a big part of their overall marketing programs, and they are not just focused on direct-marketing objectives. No, while performance objectives will always be a big part of the online ad ecosystem, more and more marketers are looking to online ads to help drive consumer perception objectives. Yes, they are using online to drive brand-oriented objectives like awareness, favorability and purchase intent.

He then says “the portals and major online ad platforms are probably not going to be the ones selling it [ad space]”, but that vertical networks will fill the void.

John Battelle wrote on a similar topic, pointing out how technology is winning over talent in the brand wars – and this is not a good thing. He asks:

Do we sell inventory to the highest bidder via algorithms, automated processes, and platforms? Or do partner with marketers and creators of media to build brands – both media brands, and consumer marketing brands?

He then goes on to assert that the obsession with chasing Google has led the likes of Microsoft, Yahoo, and AOL to ignore their strength in brand building.

Cory Treffiletti wrote a good one, too, asking when we are going to stop using clicks as a primary measure? He has a good description of the difference between direct response and brand advertising:

Fundamentally all advertising exists to do one thing: increase sales and/or market share. If the goal is to generate an immediate response, meaning the acquisition of information or a customer in a single session generated via a click, then we can consider that to be a direct response effort. If the goal is to drive an increase in sales, market share or customers over a longer period of time (basically anything beyond that initial single session) we should likely consider that to be a branding campaign, or what is most typically now referred to as a brand response effort.

He then goes on to detail why click-through rates are not necessarily a good proxy for brand campaign success.

Finally, Eric Picard has an excellent interview with Jeff Einstein, where Jeff says:

We’ve been obsessed with our own ability to measure performance (regardless of the metric) since day one online. Our obsession with efficiency and scale all but eliminates the quality of the message from the consideration set, largely because quality is much more difficult to measure and formulize. We can tinker all we want with metrics and formats, but as long as we remain fixated on efficiency and scale as the keys to the kingdom, performance will continue to decline.

I couldn’t agree more. Reminds me of trying to use standardized testing to measure everything kids are supposed to learn in elementary school. Sure, it’s okay for basic math and reading, but our kids aren’t learning how to learn, because we only “teach to the test”. Brand advertising (and teaching) still requires a generous helping of creativity and talent.

UPDATE: Another great blog post by David Koretz asks if increasing CTR can actually hurt your brand:

If an advertiser’s goal is truly branding, then driving clicks by annoying the user is a horrible approach. Not only will it detract from the user experience, but it will also damage the very brand it was intending to build.

Well worth reading.

When all you have is a hammer…

Thursday, March 20th, 2008

Brand advertising on the web is suffering from a major problem: too much measurability. To be more specific, too much measurability of the wrong things.

Click-through rates were the first measure of online ad effectiveness, and seemed like the holy grail of ad measurement. Finally, advertisers could figure out which half of their ad dollars were wasted. I think the first well-known demonstration of the shortcomings of CTR as a measurement was the MCI “Shop Naked” ad back in 1996. The ad had tremendous click-through, but nobody actually bought anything.

This spurred innovation in measuring conversions. For direct marketers, measuring conversions really is the holy grail, becuase you can immediately understand if your ad is generating more margin than it is costing. The problem is, both of these metrics, CTR and conversion rate, are direct reponse metrics.

Brand advertisers typically have very different campaign objectives. Does Coca-Cola expect you to click through and buy a Diet Coke? Does P&G expect you to click through and buy some Tide? Of course not, but these campaigns are being force-fit into having direct-reponse objectives, because that’s what we can measure immediately on the web.

Campaign objectives for brand advertisers typically include affecting a customer’s attitude and propensity to buy, their liklihood to recommend to a friend, and simple brand awareness. CTRs and conversion rates are not good proxies for these objectives. Measuring the effectiveness of campaigns on the internet probably requires the same old-fashioned techniques as measuring the effectiveness of TV or radio ads: panels, questionnaires, and close analysis of the effect on local sales.

I say “probably” because there are always new measurements on the internet, and some may turn out to be good proxies for brand objectives. Look at some of the innovations that are coming along now: brand engagement metrics with widgets, for example, or some of the video ads that VideoEgg and others are touting. However, those have yet to really prove their effectiveness relative to the real campaign objectives (witness the number of games where the players can’t name the brand that sponsored them). And, they don’t typically have the reach that brand advertisers need to make meaningful buys.

What is particularly effective for brand advertisers on the web is their ability to target. While Oprah may be a pretty good proxy for Moms on TV, on the web it’s possible for Minute Maid to target exclusively Moms. And instead of limiting their reach to content that is aimed at Moms, they can use behavioral targeting and other techniques to find Moms wherever they are on a network of sites. This can make brand buys on the internet much more efficient, assuming they can measure them appropriately.

Another effective brand advertising technique that only the internet really offers is frequency capping. Most publishers and ad networks can offer frequency caps that allow a brand advertiser’s message to reach their audience an optimal number of times, without burning them out. We all have had the experience of seeing the same TV ad so many times you actually change the channel – with the internet, that doesn’t have to happen.

Brand advertisers need to start to embrace online advertising for what it can offer, targeting and frequency capping, and stop trying to use direct response metrics to measure their campaign effectiveness. When all you have is a hammer, everything starts to look like a nail. Click-through rates are the hammer of online advertising, and brand advertisers are getting nailed by them.

Vote for Bubble 1.1 as Yahoo’s Funniest Video!

Tuesday, March 11th, 2008

The Richter Scales recent video, Here Comes Another Bubble, was nominated for the 2008 Yahoo Video awards. Go and vote for us today!

Click here to vote!

UPDATE: We didn’t win, but we did get a nice T-Shirt

The 80/20 Rule of Online Inventory

Monday, March 10th, 2008

I’ve been talking to a lot of publishers recently, and I’ve gotten some very consistent feedback: a relatively small percentage of their impressions often make up a disproportionate percentage of their revenue. This is not news, it’s a classic 80/20 rule (although the numbers may differ – some are 90/10!). And every publisher knows what inventory I’m talking about: it’s perennialliy sold out, and the sales guys are all fighting over who gets to sell it next month. Here’s a slide illustrating what I’m talking about:

Inventory 80-20

Now this is oversimplified, but typically this “premium” or even “super-premium” inventory is sold to brand advertisers. Deals can be time-based sponsorships (typically with impression guarantees) or guaranteed-impression buys at relatively high CPMs. The remaining inventory is generally either sold on a CPA/CPC basis, at much lower eCPMs, or sent to a remnant network or ad exchange for a low CPM or revenue share.

What is interesting about this from an industry level is the amount of time and energy the industry has spent on optimizing the value of the 80% of the impressions that are not premium, and by comparison how little progress has been made on increasing the value of the inventory that makes up the majority of the revenue. Performance-based ad networks have sprung up everywhere, and ad serving providers like DoubleClick have products that are focused on increasing click-through rates, and therefore revenues, from performance-based ads. But these only help turn your $0.25 CPMs into $0.50 CPMs, they don’t do anything for your $10 CPM premium inventory.

The reason is not hard to guess: performance ads, because they are directly measurable, are much easier to optimize. Lots of people have had roughly the same idea: if I crunch enough data, I can more accurately predict who is likely to click on which ads, and therefore I can get more revenue out of less inventory.

Brand advertising is much less amenable to optimization in this same way, because click-through rates aren’t typically a good proxy for brand campaign objectives (look for a future post on this topic). To get more out of your premium inventory, you have to either make it more valuable to the advertisers, by adding more data and targeting more tightly, or you need to utilize it better, so you waste less through fragmentation and inaccurate allocation. Yieldex is focused on providing tools that help publishers maximize the value of their premium inventory, which is the 20% of the inventory that results in 80% of the revenue.