The 6 / 50 rule of internet advertising

comScore and behavioral ad network Tacoda  released a study last month that caught my eye.  It at least partially answers the question that I’m sure most regular internet users have asked themselves at one time or another: "who the heck actually clicks on these banner ads anyway?!?". 

Turns out that about 6% of users are "heavy clickers" in the study’s parlance.  These users generate about 50% of the total banner clicks.  The study points out that these users are heavily skewing banner click-through data.  They are also not representative of the overall internet population (heavy clickers according to the study are between the ages of 25 and 44 with household incomes less than $40k; they spend a lot of time on-line, but they don’t tend to spend a whole lot more than other users online – their clicking behavior isn’t indicative of spending behavior).Its worth taking a read of the press release which summarized the findings from their work – fascinating.

I think there’s a coming adjustment in the online advertising industry from a shift in the way that advertisers will be looking at their online spend (which overall, I – and pretty much everyone else – believe is increasing in the aggregate).  This shift will ultimately result in much better targeting and matching of users (and specific user demographics) to advertisers.  However overall, it will also result in the lowering of mean eCPM as "average" traffic proves to be less attractive to advertisers.  CPC traffic won’t go away, but clearly we’ll see more specific measurement of those clicks (and their resulting buying actions) that will try to separate the bad traffic from the good.

Hat tip: AdLab

  • Luke G

    I ran some comparisons of CPM and CPC data over the last 20 and 50 weeks (results are congruent, btw) for two types of online marketing: banner ads and paid search/SEM (Google, Yahoo, and MS). Paid search impressions are 10x-15x more expensive than banner ads, and SEM CPCs are 50%-60% more expensive.
    The real difference, though, comes in the conversion rate (actions per click). SEM ads convert 2.5X better than banner ads, and the cost per action (CPA) is 40%-50% lower for the SEM channel. The difference? Intent is a big part of it; more banner ad activity is window shopping. The problem is that for many of us, SEM isn’t scalable to the same extent that banner advertising is.
    The bigger point, though, is that both CPM and CPC advertising are poor approximations of what we’re looking for, which is the ability to minimize our CPAs – that’s how we track ROI. CPC advertising is closer, since a click is after all a type of action, but the (only) action we really care about is someone buying whatever it is that we’re selling.
    CPM, CPC, and CPA advertising fall along a scale, with risk-to-advertiser at the CPM end and risk-to-publisher at the CPA end. CPM advertising needs to die, and I think it will; CPA is a much truer measure of what’s going on. Rather than the burden falling on publishers, though, who will do whatever they can to minimize their risk and maximize their revenue, the risk should be greedily sought by ad networks and advanced technology providers. This is where the business is and will be. CPA is the future, dude.
    This is what Yahoo is after with its acquisition of Blue Lithium and Right Media, and their integration into its ad platform; they’re beginning to roll out the capabilities to execute CPA-optimized online advertising campaigns. As far as I can see, though, there is still some huuuge business here. Can we get some more math Ph.D.’s, please?

    • sethlevine

      great comment, luke! agree that cpm prices are dropping (not sure to zero, but definitely falling) and the cpc is a better approximation of real activity, although as this study points out there's gaming across the system. ultimately advertisers will insist on moving as much spend to cpa (which is completely measurable). And yes! bring on the Ph.D's

  • Ryan Hunter

    I can't remember the exact report, but Marketing Sherpa ran a study that showed a suprisingly high correlation (I think maybe 20-30%) between viewing (but not clicking) a banner ad and then later clicking through a text/search ad from the same advertiser. The moral is that a key value of advertising is awareness building. Rely too heavily on click counts to measure the effectiveness of an ad and you risk discounting the awareness value

    • sethlevine

      that's a tuff one, ryan – i think this sort of advertising is moving much more to 'actionable' results than brand building….

  • Luke G

    Here's a question for you: what effect will falling CPMs have on CPA ad-serving (targeting) technology? Sure, smaller revenues from impression-based advertising will make publishers more willing to adopt CPA campaigns, but I kinda feel like the lowered barrier to entry will help ensure that the quality of the CPA solutions will be lower, as well. Sustained high CPM-revenue advertising would (1) help fuel a CPA solution that's so good it can compete with higher-CPM campaigns (which I think is possible*), and (2) provide a useful wall for (cough) “stealth” startups (with those Ph.Ds) to hide behind while they're building their contraption.

    Another (inverted) way to look at it: what clean energy technologies can compete with oil at $25/barrel? None, right? But when we're seeing $100/bbl, plenty of alt energy sources become feasible; they get R&D/VC money and incrementally push themselves down the technology cost curve. Competing at $25/barrel is like competing with high CPMs – any solution has to be awesome – but algorithms are way cheaper than mechanical/chemical/electrical engineering.

    So what we'll probably see are the introduction of some pretty lame CPA solutions, which will slowly evolve. Still better than CPM deals, but way less exciting than the mad mathematicians solving everything. Startups don't play the competitive market game that well either, do they? I guess I'd rather see a killer app emerge than watch the same old saga play out, but that's just me.

    *If it's not possible, than CPA tech will shrink the market even more…

    • Aziz Grieser

      I've been researching e-commerce and advertising over the past 4 years, and I just learned a lot from the discussion above, thanks.

      One of Luke's points that I'll counter was that banner ads were ineffectual. I think that banner ads, as are most commonly used, are extremely ineffectual, and we're trained to not look at those part of websites we frequent. However, my theory is that banner ads are the most effective form of internet advertising, but you have to step outside of the box.

      Ever been to Just pretend for a moment that every product listed is actually a banner advertisement. In 2006, Ibis World's e-commerce industry report revealed that 60% of people that admitted to being online shoppers purchased a product that they had no intention of purchasing when they went to that website, within the last three months. What does that tell you? Window shopping sells, and big-time!

      Couple that stat with this fact: convenience is the # 1 reason that people reported for their online purchases, followed closely by price. Why does every store in your shopping mall have a display in front?

      We people, are visual. Look at something I noticed, Craigslist's new beta product listing style now has thumbnail pictures, smart!:

      The reason for banner ads having low click-through and conversion rates is not the fault of the medium. It is the fault of the content displayed. It's either inaccurate for the person watching it, the position of the ad has trained users to look elsewhere on the page, or the actual look of the digital print is cheap and “spam'esque”.

      Steve Jobs' i-everythings, Corvettes and Mustangs, new cheap yuppie homes, Zara brand clothes, and Target stores know this more than anybody else. Make it look purty and it'll sell. A text ad cannot compete with a hot visual. Reading is not a favorite past-time of the masses, just open a Wired magazine and try not to cover your floor with pretty advertisement litter.

      – My two cents. I write about this stuff on my blog, (Seth, you can edit that out if you don't fancy shameless promotion.)

      • Aziz Grieser

        I should perhaps add something regarding CPC, CPA, CPM, C-ME-P (Couldn't resist that last one!)

        Companies have always struggled with tracking advertising/marketing ROI, and Luke is correct, CPA is the only way to really know that your dollars are going to good use. The problem is that: A) calculation requires a bunch of Lukes, strong in the Math force, and B) Advertising technologists will make far less money to ever be motivated to create such solutions.

        You know what I think the solution will resemble on some level? Built-in value (tangible compensation) to the consumer for having followed that ad through to action. When I'm faced with 10 online backup brand advertisements, the one that actually gives me something to gain from using their service, as opposed to shoving their brand and some crap slogan about ultimate performance, or “perform like a Tiger” Accenture-esque BS in my face. However, there will have to be a multitude of value alternatives to the advertiser to choose from, and that will require much deeper integration with the medium.

        – My four cents now… (Seth, your ad-revenue should come up a bit now.)

        • sethlevine

          love the long conversation on this stuff, aziz. thanks for putting in your 2c.. seth

      • sethlevine

        btw, no worries on the site plug – with such thoughtful comments, you deserve to link back to your blog!

    • sethlevine

      Luke – The overall increasing online ad spend will DEFINITELY drive investment in this area (we're already seeing both an increased in the number of companies and fundings in “advertising”. I think we'll first see some people try to game the system a bit (and target the “clickers” differently than they target the rest of the population. Long term, I think there will continue to be a push away from transparency (the lack of which benefits the ad networks – think about what Google just did with its open ad server announcement – how would you like to have Google know what all the other networks are paying for traffic!). will be interesting to see where this really ends up . . .