I’m a bit surprised nobody asked what kind of sites we kept during the $155 to $5 per new customer optimization of the Lab Store AdSense Campaign. But I guess I should not be, as folks probably just assumed it was a straight ROI thing. And it was.
But often, I think people miss out on really learning something because they just follow the numbers and never ask why. Why does the optimization work like it does? Because if you understand the Root Cause, then you take away learning that you can use the next time around. This result as opposed to robotically optimizing the same stuff over and over without learning anything transferable.
For example, what is the psychology or sociology behind why certain optimization choices worked? What conscious or unconscious mechanisms cause people to do what they do?
Many folks interested in “why” might then go to surveys, but the problem with that approach is people often find it difficult to tell you why they do what they do. This is where a basic knowledge of Psychology and Consumer Behavior can be extremely helpful.
So which sites did we keep?
There are 3 answers, depending on your point of view; yet all 3 answers below apply to all the sites that made the cut:
1. Optimization – the sites with the highest ROI
2. Consumer Behavior – the sites delivering an experience most like Pay-per-Click ads in the search engines
3. Psychology – the sites where the audience is in a Listening Mode
Reasons 2 & 3 are why Reason 1 happened. Reasons 2 & 3 are why chat boards and similar sites delivered the highest ROI. Where it gets really interesting and drives a theory, is when you toss in that Social Media had the worst ROI.
I mean, what’s the difference between a Chat Board and MySpace, a difference that produces a cost per new customer of $5 versus $155? One could argue they are similar – communities, connections, etc. First there were online Discussion Groups, then Chat Boards, then Social Media. Same idea, just with progressive technology improvements.
Right?
I don’t think so. I think there is a clear reason the results are so different. One of the above is not like the other two, from a Behavioral and Psychological perspective. Here’s my personal theory why in general, AdSense in Social Media sucks so much compared with AdSense on topical Chat Boards:
Discussion Groups / Chat Boards are about “how can I help you?”; Social Media is about “how can you help me?”.
Discussion Groups / Chat Boards are more about Social; Social Media is more about Media.
Discussion Groups / Chat Boards are about Learning; Social Media is about Connecting.
Therefore, Discussion Groups and Chat Boards are a much better environment for AdSense, because people engage these environments to absorb information and listen; people engage Social Media to broadcast information and speak.
And an audience that wants to listen is simply a better environment for display ads than an audience that wants to speak.
If the above is true, there are enormous implications for creating successful products, ad copy, and landing pages for these different environments. And critical implications for entire business models.
Thoughts? Please feel free to poke holes in the above theory, explain the same theory in a different way (perhaps using words that make more sense to you), or provide an alternative explanation!
The next post in this series is here.
Jim another fascinating set of postings. Thanks for that.
The first thought I had re the Social sites being “bad” is that perhaps they are simply the wrong demographic for your products?
I was discussing loosely around this topic with Avinash K a few weeks back. He hit me with a Q: What about the twitter/myspace etc crowd? I replied that they are not even remotely our target audience. Those who do inhabit those sites just aren’t interested in what we offer, and probably won’t be till they get another 10+ years older.
Confessing that this is my *observation* of the inhabitants of those social sites. I have zero hard data to back that up. Beyond a complete lack of any referral traffic from any social sites. Of course. :-)
Is it perhaps as simple as that? The social site visitors *want* to buy, but can’t? Or perhaps you just create really compelling banners to click on?
;-)
Cheers!
– Steve
Steve, spoken as a true analyst. I am hoping people will fire hard at these ideas, because there is quite a lot riding on the conclusions we come to as a group…
As far as demos, we know from the posted (by MySpace members) activity that many of these folks are into exotic pets. There is tons of material up there. Further, the folks that own an exotic pet create lots of interactions with each other on health, feeding, etc. They’re “engaged” in the topic. So absent any other info, I don’t think “demos” are the issue.
Seems to me it’s more about how people come to the environment, what their mindset is (Psychology), and what their intentions are (Behavior). Clearly, AdSense might have some contextual targeting problems on MySpace but there is nothing I can do about that, so the issue is moot.
As far as the ads go, well yea, I hope I can create a compelling ad / landing page combo by now…
But since I did not specifically bring it up, let me explicitly state the test used exact copies of both the ad and landing page for all AdSense sources. Also, we used the same pricing across all AdSense impressions. The idea was to find out at a base level if we could optimize AdSense into our financial model. You can always tweak later, but there’s no point in even trying if you can’t get “in range”, if you know what I mean.
By the way, the same creative was used for AdWords, which on ROI smoked everything on the AdSense side, Chat Boards included. There is nothing like being there when the question is asked directly. On the Chat Boards, we get there indirectly.
But there is a different texture to the Chat Board audience, they are higher up in the funnel than the pure searchers (we can tell from the search terms they use and URLs), so we’re pleased we can finally reach them economically through AdSense.
Hi Jim,
OK, another brick in the pond! Having read you for some time now, I know you’re not very much into demos in order to explain behavior. Are you here suggesting that “mindset”, or “experience dynamics” (push vs pull, information gathering vs broadcast, community vs individual, etc), would be a way to segment, and that could have some predictive value?
If so, hmm, very interesting. And, yes, the implications would be huge in how we evaluate sites in a placement strategy.
Jacques, thanks for the comment. You know, I’m not very good at explaining myself, am I?
Or, probably more likely, I use language that is familiar to me and not to most other people. So despite how hard I try, I really do need folks to interact in their own words so we can expand the knowledge base. Thanks for putting this idea into different words that make a lot of sense.
Yes, I am absolutely suggesting mindset / purpose as a way to segment, and a much more accurate and actionable way to segment at that. If you just think this idea through, it makes so much sense.
Actions and context speak so much louder than demographics ever will. People are what they do, not what their age, sex, and income are. And what they do needs to be interpreted in context.
So, if I am in *my space* at MySpace, I have certain tasks I am trying to complete that are very ego or network -centric and I am not paying attention to or interested in display advertising of any kind, no matter how targeted. I just don’t care. I might pay attention to a widget, because (hopefully) it helps me accomplish a task faster, better, etc. The widget is helpful, advertising is not.
When I am on a Chat Board, I am in Learning or perhaps Teaching mode if a moderator or old-timer. Questions and answers. Seeking knowledge, exchanging ideas, as opposed to transacting social business. A widget is probably not helpful here – I don’t know what to do with it. But a well-written display ad could possibly be helpful in this environment.
Further, if you understand the context, you would create ads specifically for this context, as we did. Our copy for this test was not about “buy now” or “best prices” or anything like that. It was about “do you need more information?” which worked superbly for the chat boards and miserably for MySpace.
But I submit to you that the MySpace issue is not a copy problem, it’s a context problem. The copy we used should be generally applicable to the MySpace environ – not pushy, helpful, friendly.
But it didn’t work.
And I think the reason is there is no “audience” for ads there.
We might try to create a different kind of ad for MySpace in the future, but here’s the thing. A display ad that would work would be probably something about “connecting” or a widget or whatever and that simply doesn’t have enough value to be worth the effort.
In other words, it’s simply too expensive (and an incredible waste of resources) to let people know about our Brand until they actually need it. And when they do need it, they will find it through Search or topical Chat Boards.
And then, they will have such a powerfully positive customer experience that they will come back again and again.
That’s Relationship Marketing. Pull, not Push.
I’m not into interrupting people in a personal space with display advertising, and I don’t think most MySpace folks like it either.
This is the paradox of the display ad-supported Social Media business model; the more successful you are at building a thriving “connection exchange” the lower the value of display advertising is to the audience in the exchange. Success is self-defeating.
Interactivity is different. The context is different. The behavior is different. Why do people continue to use the old ad model?
If you are into the “awareness” thing, which truly is critical for certain Marketing situations, buy the real Mass Media, not MySpace or FaceBook or whatever the newest twist on Social is.
Social is Social, it’s not a “media”, except where it might apply specifically to the context of Social.
In case it’s not clear to other folks reading this, I think apps like MySpace and FaceBook and all the related ideas in progress are incredibly useful to a broad range of people. I’m just questioning the value of running display ads within these environments.
Some day, the Marketing people who buy these display ads will measure outcomes correctly and realize the same thing. Reach and Frequency – the old media GRP thing – are simply not relevant in the context of Social environments.
Hi Jim,
Your ideas are both excellent, and very dangerous. You are putting your finger on a dirty little secret that is becoming less so day after day. Interestingly enough, I was catching up with my magazine reading this morning when I came across a very interesting article in the February 18 edition of Business Week (pp.54-55). Allow me to quote from it:
“Many of the people who hang out on MySpace, Facebook, and other sites pay little attention to the ads because they’re more interested in kibitzing with their friends. Social networks have some of the lowest response rates on the Web (…)”
And the article goes on explaining that the big ones recognize that situation, and are planning to allow marketers to place ads in hundreds of categories of interest. Which won’t fail to make those ad programs even more agressive, and frustrate users.
Typical, ain’t it? If demos don’t work, then psychographics will. However, and this is where what you’re saying fascinates me, it is a lost battle. It won’t work, because they are going to frustrate their users. Why? Because it is not in the DNA of those sites, which call for a certain type of interaction, i.e. “purpose” to use your word, and that ad interruption is counterproductive to the users.
So, here’s to you: should we assume that some sites are jut not MADE to have ads? That the main reason why users are attracted to them just does not fit the model? If the answer is yes, well, MySpace and Facbook don’t have a business model, or at least not the ad supported one.
If “purpose” is then a strong predictor of ad responsiveness, could we be able to segment/classify purpose(s), and predict which type of purpose generate which level of responsiveness?
I mean, my God, this is very powerful, and this brings the question of proving it. I have absolutely no doubt your testing was rigourous, but if I understand you correctly, we need to extand that way more.
> should we assume that some sites are jut not MADE to have ads? That the main reason why users are attracted to them just does not fit the model?
Yes, and further, the main reason people are attracted to them is actually a negative for ads, so the more intense the attraction, the less effective the advertising will be.
> could we be able to segment/classify purpose(s), and predict which type of purpose generate which level of responsiveness?
I don’t know about predicting response, but you can predict ad effectiveness. Let’s go the opposite way. Think of fashion magazines. People subscribe to a large extent *because of* the ads. That’s a heck of a business model, one that really works for advertising, don’t you know.
Context is everything in advertising. The less the ad is “in line” with the context, the less effective it will be. Online display advertising is very frequently out of context because it’s not mated to topical material. PPC ads are almost always in context by nature; they are *designed* that way.
AdSense is a hybrid. It tries to be in context but often fails.
Unfortunately, the failures tend to be on very high volume sites, and the successes are on small, very targeted sites.
Makes sense, right?
If I can remove the failures (Social) and have enough good stuff left (topical chat boards) I can make AdSense work.
And it’s not that the folks on the social networks don’t buy, they don’t even click through! That should tell you something. Either:
1. The context is poor. This is a targeting problem, my ads don’t show up on pages with related content. Possible, but hard to fathom given all the personal info present in the pages. I searched for MySpace pages on our topics and generally saw our ads within those contexts. What I don’t know, of course, is how many ads ran where. If 1% of our ads ran in context and 99% did not, well, that’s a problem I can’t fix. Weighted, the entire MySpace site would then be “out of context” from my perspective.
2. MySpace participants simply don’t click on ads because the entire concept of an ad – any ad – in a personal space is out of context. It doesn’t matter if the ad is attached to great content, because the attention in a personal space is completely absorbed with other activity.
See GeoCities and Tripod. We have been through this before. People will argue “there’s no comparison, the new social has so many cool tools, it’s like night and day versus the old social stuff”.
Right. More tools. More to do, more reasons to not pay attention to ads in the space. The tools argument is actually an argument in favor of why the ads don’t deliver any benefits.
In this case, “Engagement” is not good for the business model, because the higher you drive focus on activities, the less effective the advertising will be in the space.
I come into the space, I complete my tasks, I leave. What ads?
See also the free online storage business model. They show ads when folks are uploading files. Does anybody think people are paying attention to those ads? Really? Dream on.
Advertising works best when the audience has *nothing* to do except watch TV, listen to radio, read a magazine or sit in traffic / on a subway. Said another way, advertising works best when you have nothing else to do but pay attention to it.
Makes sense to me.
Unless, of course, the advertising is about a topic people are specifically researching at the minute the advertising displays. Online, only PPC ads and display ads in tightly targeted contexts (topical chat boards and pure play B2B sites, for example) fit that model. Your attention is already on the subject matter and the ad is part of that experience – like a fashion magazine.
Jacques: I wonder if the article quote is in fact wrong? Given Jim’s experience? In that lots of traffic came (ie. The ad worked in that sense) but in the sense of getting results beyond a “click” it failed. So much traffic came that it was too expensive and the “wrong type”. ???
Which in turn Jim leads me to some additional thoughts:
* You mention re-tuning the advert for the myspace audience. You mentioned initially that you did get some ROI (just very poor) from the myspace crowd. Is there a sweet spot? Was the rate of sell to myspace folk enough that there is an opportunity to create a better advert for myspace – I want to say: “do a bad job”. :-) ie Create a less compelling advert so you get less clicks, but get the ones you want? The ones that buy.
* The whole “people don’t pay attention to ads” maxim. Some of us have been that way since last century. ;-)
I got so fed up with ads continually flashing at me and interruptus maximus in my tasks of using the ‘Net as a work tool; that I used all my technical skills for evil – I block every last little advert I can. Originally via dedicated proxy server(s), now inline to the browser itself.
The only ads which actually survive to be seen – and rarely acted upon are the google results and gmail style ones.
ie text only; and out of my way except when I actively look at ’em.
* I also wondered what those people coming from myspace but not buying are doing. They’re obviously interested enough to pause their myspacism, but {….}???
One is left thinking that there is *some* disconnect there. Is there opportunity? Dunno….
Fascinating stuff! Especially when you get me thinking about this while I do the dishes etc. ;-)
Cheers!
– Steve
Well, taken in the context of the Natural Born Clickers study we might say that most of the people that clicked had no intention of doing anything else…they were bored.
Either way, the bottom line is the general experience of folks running display ads in Social Media has been poor. And I now have a specific example of my own. So one has to wonder why the only ads that “work” on Social Media are ads for tightly related concepts like widgets and so forth – products or services directly related to the social experience.
Makes sense to me. It’s a “Me, Inc.” kind of space.
So either:
1. The audience sucks. I don’t think this is true.
2. The environment for ads sucks. This I think is true, see comments.
3. Something is severely broken with the targeting system. I have no evidence for this, but it’s certainly possible that the majority of ads run completely out of context (as much of display advertising does) so is viewed at best as irrelevant and at worst intrusive and bothersome.
The only way to resolve this I can see is to give me control of which profiles my ads run on, probably by keyword phrase – just like PPC.
By the way, the PPC ads we run on MySpace through Google Search Syndication?
They suck too. See #2 above.
Hi Jim,
Very interesting read. I got here through a referrer (his name is Steve, could be the same as above), who posted this on a mailinglist on a related topic.
Which left me wondering. I get heaps of catalogs and stuff in the mail everyday, and while a “no ads” sticker could fix that, I do want to see it sometimes. Just not always. In fact, most of the time it goes straight to the bin.
It’s pretty much the same deal with online ads. Most of the time i just don’t care. But every now and then, and need something. And then I want ads, as they tell me about sales, allow me to compare, etc.
So what if there were an ad portal, where sellers could place their products (along with the competition) and where consumers can search through them looking for the product they need? Would that work for you as a seller?
Ideally we’d cut back on paperbased catalogs, although I don’t think that’ll happen any time soon.
I kinda like the idea (as a consumer) where I can browse the ads that are relevant to me.
Cheers! Olger.
Olger, thanks for the comment. It seems to me every time someone tries to do that model, it doesn’t work out. If you think about it, Search kind of serves that purpose, though there is the Serendiptity problem – if people don’t know about a product, they don’t search for it.
Perhaps new product launches are what the mass media is for?
Otherwise, on the web, I think it boils down to the fact the web will never be a “mass media” and when you treat it like one it simply doesn’t work. Smaller campaigns / Better targeting is the answer, including making display advertising more intelligent and sensitive to visitor intent.
So for example, in MySpace, differentiating the ads shown to people actively managing their accounts and people looking around. The page views associated with managing an account probably have a much lower value than those associated with casual browsing.
Question is, what % of page views are associated with managing an account? Could be substantial…meaning this is why ads tend to perform poorly in that environment.
A fascinating conversation. Jim, in terms of putting the discussion in to our own terms, I always talk to people about the concept of immediacy among prospects.
Assume for a moment, that we remove the demographic question. What if we only considered the universe that overlaps in chat forums and social networking sites. An identical demographic for our test audience.
The chat forums would still be more effective (given appropriate contextual ads), presumably, because of the immediacy of the need.
The example that I often used for my clients was in the valuation of rented opt-in lists versus keyword advertising (this is going way back). The members of an opt-in list have indicated a predisposition to receive information about a certain topic, say exotic pets, or web analytics tools. But when those same indviduals have an immediate need, they actively go out looking for the solution. They can suddenly “see” the ads that they would normally filter out.
Just like my inbox, or even my rss feeds. I read them when I have spare cycles and am interested in generally expanding my knowledge. I almost never go to them when I am actively searching for a specific solution. (I may have the memory filed away, and recall it – then go searching and click, or be reminded, by a chance stumble across an ad, which accounts for the occasional success).
So, in order to Myspace, or Facebook to have a successful revenue model, they need to identify the immediate needs of their audiences when they are interacting in the space (like widgets).
Yes. “Display” isn’t the issue, because you can make it work if the audience is focused enough. You have to create the same kind of contextual environment for display as you get in search to get similar, or even better results.
The problem with most social media as a display ad concept is you can’t do this, because the aggregation of Attention is along the wrong vectors – except for products directly related to social media. So what these guys need to do is create “channels” or “verticals” so the context is created as part of the resaon to be there in the first place.
The problem is this means they will be much smaller businesses than they are now. There is not a general advertising market for the “gross Attention” they are currently aggregating.
The web will never be TV, it’s just not built that way.