Cost Per Influence

Internet advertising was subjected to broadcast media metrics from the beginning. CPM, or Cost Per Thousand Impressions, was borrowed from print and was accepted by traditional advertisers as a measure of reach and frequency. Back then, if a company had a site to point to it was largely brochureware, a corporate identity on the web. But when the bubble burst its effectiveness beyond branding was questioned. The industry shifted to Cost Per Click around the same time that most companies had transactions available on their sites. An ad was effective if it drove transactions (Cost Per Action is another metric, a step beyond a click as a lead to an action as a sale). Consumers became sensitized to how broadcasted ads were trying to influence them. Google stepped in with a market for advertising, based on CPC, that rewarded effective narrowcasting. Both ads and sites are optimizing their messages for what people are looking for to gain traffic and transactions.

This model works fine with companies as the only influencers and the only ones with sites. But it ignores the influence of social networks. And what happens when consumers become users with their own identity on the web? When conversations influence attention?

I’ve suggested its time to explore new ad metrics:

What’s different with new media is simply that it’s not the number of impressions you make, but who you impress. In other words, instead of subscription counts, its the number of subscribers my subscribers have, discounted by the probability of my memes getting through. Cost Per Influence.

Jeff Jarvis comments:

You’re right: We need to define new metrics. This medium isn’t about impressions; it’s about relationships; it’s about conversations; it’s about influence; it’s about authority. We are starting to measure how many conversations a blog starts (or at least takes part in) with Technorati. But it’s just a beginning.

Michael Sippey agrees its depth, not breadth:

At some point a smart media buyer is going to wake up to this fact: that it’s not about how wide an individual meme maker can spread the seed, but how deep he can plant it into his/her first degree network, and those folks into theirs, and those folks into theirs, etc.

So what could this metric look like? Here’s a simplified example for an average blog that has 1000 monthly views and 100 inbound links:

On an impression basis, the blog hosting the ad gets $1 from CPM. Like broadcast, a blog with high traffic rates can generate a return. With an assumed 2% click-through rate for its 1000 views, the return of the CPC model is $10. For a blog that follows the law of the niche, if the language matches advertiser topics it generates a return. But traffic and topic do not represent influence and authority.

Assume the simple model of Cost Per Influence represents a blog that has 100 inbound links from blogs with a 2% probability per post that a new link will be generated, a return of $10. The virtues of this model is measures reputation, is simple and can be measured by Technorati today.

However, this simple example isn’t a good enough measure of influence. What matters isn’t just how many links a blog gets, but the reputation of blogs linking to it. Consider a CPI measure that represents two degrees of seperation: 100 inbound links from blogs that have an average 100 inbound links with a discounted probability that memes will cross two network horizons. The second degree of seperation is important because it factors in network structure in a simple way.

Fine theory, but what’s in it for advertisers? The meme being transmitted isn’t necessarily the message of the ad. In this case the measure is a proxy for the traffic it could generate, so why not go with the established metrics? According to an IAB Study, there is a clear trend towards performance-based measures, from 26% to 41% of revenue within the past year, and what has been outlined so far applies to impressions.

Google AdSense commoditized placement of CPC-based ads while putting the burden of performance on the ad writer. The problem is it treats my personal blog and Instapundit as the same for placement if we use similar language. This creates a niche for services like BlogAds. But a measure of influence remains implicit in valuing properties.

In effect, the same influence game is being played out implicitly with search engine optimization (NB! the biggest risk of a link-based metric is appreciating links fosters link spam, another reason why factoring in the second degree is important, its harder to game). But for advertisers, explicit metrics and mechanisms for placement need to be provided.

Where traditional metrics truly break down is with syndication. Two changes are afoot that I believe could enable explicit tracking of a two degree metric:

• RSS Commercialization — Users want full text feeds to manage in their news aggregators, especially with persistence. Advertisers want measures of how effective ads in feeds are, so today feeds are abridged and ads are served on sites. Expect a model that provides full feeds with trackable ads, everything you get on sites with more user control over subscription (portends a war with embedded ad blockers).
• Integrated Aggregation & Publishing — With Newsgator and Feedburner funded and on its way, expect more aggregtors to leverage the Atom API to enable integrated aggregation and publishing to tools of choice.
• Indexes Become Services — With Technorati funded and well on its way with a new Ad Mission, they could serve and track something different.

Transitive Ads

A new ad format is called for that works at the micro-content level, provide advantage to those with influence and follows paths of meme transmission. One major caveat before describing it, on a per-feed level, users can apply a block all ads feature to each feed. The ad is included at the bottom of a post, both on the site and in the feed. A basic one-degree CPI measure covers the value of initial impressions — ad buyers pay relative to established reputation.

A two-degree CPI measure is performance-based. In this case, performance isn’t a consumer clicking through. Performance is a reader that decides to write about the original post. Revenue is shared by both blogs, if the ad is pass through either through aggregation/publishing or index/service. The revenue share mix could either be weighted to favor the orignal poster (Instapundit picks up my post/ad, but I still get most of the return), or could be based on relative reputation (Instapundit picks up my post/ad and makes more money by adding his reputation). I would prefer the latter as its less of a Ponzi scheme, and remember that reputation isn’t built through ads but other more implicit means.

An important facet of this format is the amount of user choice. Users decide what feeds to subscribe to and ads to block. Bloggers should be able to choose what ads to both host and pass through.

I’m no Ad Man, and there are people with greater expertise in this area. What’s important is a conversation of metrics, mechanisms and formats to help us realize value from the network structure we participate in. If you have an alternative model, do post it, perhaps flush it out on this wiki page. But what ever it is, do base it on links and feeds, not impressions and clicks.

UPDATE: Chicago Times on Bloggers making a living with advertising