CPC, CPA, lead generation does not work

Idea
This is the idea that the old CPA and lead gen models are inadequate measures of performance and therefore their pricing needs to change.

Problem (s)

The problem here is knowing when a user clicks on an ad, and then buys a product, if it is as direct result of that ad.

What lies behind these issues is that click-through does not correlate to purchase.

It is worth a brief diversion to say a bit about why click-through does not correlate to purchase:

1) Clickers
Research shows that some users are just “clickers” and will click on anything. Clickers make performance marketing less accurate. Clickers are not stupid, they are your wife.  And they skew results.

2) Non clickers
Non-clickers make performance marketing less accurate:

“respondents said they were twice as likely to notice a web ad, not click on it, but visit the advertised site later (61 percent) as they were to click on an banner ad to reach a site (30 percent).”

DoubleClick Touchpoints survey, 2006

3) Click before purchase

The last click before a purchase is not the reason why that purchase was made.  Intuitively this makes sense: I am personally more likely to click on an ad for Dixons than I am for ABC Online electrics ltd. This is a result of 20 years of brand-building from Dixons.
The question then becomes: why should I pay a lead-gen fee or CPA to an affiliate if they have not done the work? This is the most potentially disruptive one to all those elegant business models that we are backing.

Solutions

How are people solving the above problems?

1) Clickers & non clickers

Agencies are trying to solve the problems posed by clickers but it’s not easy. One of the most compelling is that from Double Click who offer ”view-throughs”.  With View-throughs, lead generation / CPA is paid even if the browser buys within 30 days of viewing.  The problem is still far from solved.

2) Paying rewards based on effectiveness

If advertisers and their agencies are able to come up with one generic and widely accepted measure of effectiveness, then results can be paid.  This is a fascinating area and I have covered here  .

3) Media mix optimization
Due to the quantitative holes in performance marketing, some advertisers are seeking to attribute value to “total exposure” rather than pay for specific clicks: paying for exposure across the media mix.
NBC is trying out a metric called “total audience measure”, which tallies TV and online impressions: “you have to understand, every medium has a different effectiveness level”. Beth Comstock, president of NBC Universal Integrated Media

Opportunities for growth companies

The big issue for many start-ups with a revenue-model based on CPA or lead gen is the question of whether the last click before a purchase is the real reason for that purchase being made. I personally believe that the current system is the least-worst mechanism that we have and therefore PRICING WILL STAY THE SAME:  The downward pressure on CPAs that would come from recognition of additional brand building will be netted-off by the overall shift in spending from mass to performance marketing.

It doesn't concern me too much that the last click before a purchase is not the reason why that purchase was made.  Why?  Because a great web app will act as a "noise reduction agent" to cut through the advertising clutter. This is one of the compelling things about Moveme.com (Advent Portfolio company): there is a clear development towards reducing the noise and concentrating on signal strength from publishers.

Affiliate Networks themselves are seeing some questions from advertisers about the true effectiveness of affiliate network campaigns . I have previous posted on this.

All web properties; from small blogs, through growing media-networks (like glam.com) to massively trafficked ad-driven models like Dailymotion will have to continue building bigger and more comprehensive databases of their users: profiles, demographics, likes, dislikes.  Advertiser demand for this data will only grow.

I am also convinced that we’ve not seen enough from companies targeting specific influencers in key communities.  I’m excited about businesses helping advertisers go after influencers in specific niches or providing horizontal tools to go after influencers in multiple verticals.

Finally, the enormous shift away from mass marketing and into analyzing performance marketing and targeting will create fascinating opportunities for growth:
•    Managing customer data-bases
•    Analysing customer data
•    Automating the measurement and performance of advertising dollars on-line and offline

Comments

[Jens]

I think what would be really useful in this context would be to enable advertisers to benchmark their campagins against those of other companies. Let me explain why.

Fundamentally, there are several issues here. To start with, there is the effectiveness of a website for a particular company/product when compared to other types of products. Then, there is then the effectiveness of various websites when benchmarked against each other. Finally, there is the question of the effectiveness of the campaign. If you could measure that - the horror! performance-based pay for marketing agencies...

Anyway, after all, this is all just a question of price. If I could dynamically bid on a bang-for-the-buck basis (based on a mix of clicks and impressions) across a wide range of websites, and had benchmarking data to compare my campgains with other companies (on an anonymized basis)that would be great.

[Vitaly]

First off, let me ask 2 questions I have regarding this post. I hope you can make it clearer.
1. What exactly do you mean by "noise reduction agent"? I don't see how a site dedicated to helping people to move their homes from one place to another does this. Especially since you mention publishers.
2. Then, how do you mean "target influencers"? How this can be done? Who are influencers? Why they should be targeted?

Also I would like to share some thoughts of mine on this problem (I have been working with and developing ad serving solutions since 2002 so I believe I can give some insight on the technical details of action-tracking).

The problem is as old as the Internet. Technically it's very similar to the spam problem. If you read about some of the fanciest ideas on how to solve the spam problem, what types of anti-spam solutions are out there you can figure out what types of solutions can be applied here as well. Action-tracking hasn't been as popular a problem as fighting spam.
So I would break down the solution into 2 parts: technical and social. Socially you can force the user to write down the coupon ID after the click, so next time he comes to this site he gets some discount by providing the ID. Technically you can drop a cookie for 30 days and keep your fingers crossed the user is not paranoic and doesn't go Tools>Clear private data...>Check Cookies>Clear every now and then. Personally I clear my private data from time to time but not cookies (because I don't want to input passwords each time).
Statistical approach is a good one. But technically it can become very complex and end up inefficient. Because with new data coming in new rules would have to be built. There are exceptional cases in advertising too (like Christmas holidays when suddenly CTR jumps up - my guess). So even statistics meant for noise reduction would need noise reduction too.
And I don't think advertiser would prefer to see 'approximately 10.345 conversions' instead of '7 real conversions' in the reports. It's fooling yourself. You get ROI when the customer pays, no approximation here.
Advertising startups are popping up in all possible niches. I especially hate those new double-underline popup ads that are messing up with real links. What I see as a niche is solving old problems the new way. Eg. using Flash StoredObject instead of cookies - a solution that came up only recently after the Flash player has become ubiquitous. After all Google founders did not invent web search, they just handled the problem differently.

[Oz Har Adir]

Its a good post and so is the one above.
I believe that in the travel industry in general and in the price comparison industry in particular (and at best when the two meet) 'accurate info' leads to very useful clicks that are directly tied with purchase. With the 'commodotization' of some products into services, accurate info in the shape of 'full variety' of options for purchase and 'best price' found through comprehensive comparison are enough to answer a very specific customer need in a perfect way.
To give a practical example: if Kayak.com would have indexed all possible flight options from Amsterdam to London and find the best price available for each of these flights, then any potential buyer of this trip who searches on Kayak would have been very likely to purchase following the link to the supplier website.

This example is very true for specific industries and would not pass any of the problems you mentioned in other industries, but I think that its a shame that because of the current acceptance of the CPC & CPA models, even firms such as Kayak prefer to drive as many inaccurate leads in favor of providing valuable search and valuable traffic.

[Fabio DB]

Just made a few comments on this post here:
http://blog.veedow.com/2008/07/11/lead-generation-does-not-work-or-does-it/

I tend to agree on everything except the fact that affiliate "don't do the work" and therefore their CPA is not entirely deserved. I think that the point is, Dixons and other businesses build their brands and then other people convey traffic to their sites. Some websites just try to generate leads and don't care about their quality, while other services (moveme.com or my startup veedow.com) help making the signal to noise ratio better and crossing the chasm between products/services and consumers.

[Glasses]

agree that CPC does not work - but a good way for a company to get some potential visitors, and try to drive customers '' through the door''

[Brian]

This made me smile "Research shows that some users are just “clickers” and will click on anything" The person who discovers what makes them do it will be as rich as Bill Gates.

I could really do with a couple of them every day clicking away.

[Lunchthem]

Democratic Engineering,how cash sport possibly test structure wind individual growing whereas park course suppose panel recognise version assume ministry invite inform absence grey religion plastic realize interesting offer judge author sense busy religious sight basic labour object soldier invite she pleasure reform source rock scheme future other double promote or ever equipment month success play bit category head care dinner whether impression bill treaty requirement most on troop vote church hospital thus attractive long individual belief formal worker critical direction impose whose personal community press context