Animas Weblog

The company I currently work for offers web-based advertising and lead generation for varied clientele across a specific niche. The great majority of our advertisers pay a flat monthly fee that scales according to the exposure of their ads on the sites (size, reach, exclusivity, etc.), rather than a CPM or CPA model. This is the standard in our industry, and it’s a fairly traditional-media revenue model for a purely Internet business. Much of our added value is based on aggregating traffic from a number of sources that would be cost- and time-prohibitive for a single advertiser, and making efficient use of this traffic to maximize advertising exposure and lead conversions for our advertisers.

One of the best things about web-based businesses, and particularly one with a virtual product (leads delivered, in our case), is that everything can be measured, quantified, analyzed, optimized. In theory. Interestingly, the key drivers for our revenue model are difficult to define, measure, and optimize. New advertisers and high retention of existing advertisers drive revenue for us. But the goals of our advertisers vary so greatly that there aren’t standardized metrics that will predict contract renewals vs. cancellations across the majority of our customers. Example:

1. Some advertisers have an army of sales people to chase the leads we send them. Some use sophisticated sales funnel automation tools to do some of this work with machines. These advertisers want lead volume rather than lead quality: throw as many candidates into the sales funnel as possible and use brute force to winnow this down to the ones that will close.
2. Some advertisers prefer quality to quantity. They work the funnel manually and expect much of the qualification to be done before they receive the leads.
3. Some companies have such a strong brand that they will receive overwhelming leads regardless of what we do to promote them or bury them.
4. Small advertisers just want to expose their new brand to our targeted user base. Although lead quality is important, neither lead volume or lead quality is as important as the branding opportunity they have with us.
5. Some advertisers want not only exposure on our sites, but also exposure on search engines or partner sites that we have access to by aggregating traffic. We are a gateway to a diverse audience, some of which is not ideally targeted, but all of which leads to additional brand exposure.
6. Some advertisers are overly protective of their brands and prefer fewer leads rather than suffering any brand dilution through branded keywords and the like.
7. Take any combination of two or more of the above, and it also probably fits some of our advertisers.

So, what metric do I pick that will provide a snapshot of my efficacy at pleasing these advertisers, and can be used as an indicator for the health of my projected profit/revenue a quarter or a year from now? One thing I can tell you for sure that does not work is managing to an aggregated cost per lead (CPA). Managers at our parent company stress this metric as crucial: it’s the easiest to understand and it’s something which is easy to compute historically. Unfortunately, in our business revenue does not correlate directly to leads; our revenue is two or three steps removed from leads. There is a correlation at a contract by contract level if you understand the goals and subjective aspects of each advertiser and match that up with the number and quality of leads provided to that advertiser. But that is pretty tough to quantify granularly and then aggregate into an overall snapshot.

In terms of the advertiser scenarios listed above, managing to minimize aggregate CPA will result in greedily pursuing leads from one or two advertisers who fit profile number 3. These advertisers will only represent about 5-10% of revenues, but all capacity to provide leads will be concentrated on these two advertisers. They will be flooded with more leads than they can possibly use, while the customers representing 90-95% or revenues will receive no leads and will cancel when their contracts are up. It almost seems like CPA is the pessimal metric to manage to.

It is overwhelmingly more likely (in our particular business) that the optimal distribution of traffic and leads across our advertising customers will result in a raised CPA and a disproportionately large increase in customer satisfaction and retention. In the short term, an increase in CPA actually correlates to profit growth because revenue growth far outpaces the increase in cost of traffic and leads. This is not an easy sell to management that expects conformity to a CPA trend. This principle is specific to my business, most likely not portable to yours. It also doesn’t scale forever, and is dependent on digging out of an efficiency-hole that might not exist in your business. And the lynch pin is the “optimal distribution of traffic and leads across our advertising customers,” which turns out to be what I spend most of my time working on.

What IS portable to your business is this: high level performance metrics are necessary to understanding the state of a business (or ad campaign, etc.). But in any business where success is won at a granular level (long tail applications like search come to mind), these high level numbers must be representative of the granularity that drives things. Just because you can measure something doesn’t mean an appropriate goal can be formed out of it. Think holistically about what would happen if you could move the needle to 100% on your most important performance metric. Shooting for a low CPA sounds logical enough, but in my case I would lose 90% of my revenue as I manage to CPA in a vacuum. Find other measures of the health of your business (or your campaign) and understand which of them balance each other out. Learn to keep things in proportion by pulling these levers until you get things pivoted correctly.