When a great salesperson is able to use an understanding of the types of buyers and their trigger points to close more business, we call him a "closer." We can't avoid great images of Alec Baldwin in Glengarry Glen Ross or half the folks in Boiler Room.
When a computer does the same, we use words like "segmentation," "optimization" or "personalization." Frankly, if I were a computer, I would be pissed. Seriously, would you rather that people referred to your charisma or your artificial intelligence?
To me, optimization is just a fancy word for making a selling process more relevant and engaging for your customer so that they make you more money. And the best tool lets a marketer adapt and learn, while the machine does the work.
Optimization Approaches
We see four major approaches to optimization: testing, targeting, predictive modeling and social. Each has critical value for the marketer.
Experimentation has a long and storied tradition in Western scientific thought. You pose a hypothesis, design an experiment that isolates the effect you are investigating, and measure the results. If the data supports your hypothesis, then you go with it. If not, you re-think and try again.
In online testing, the objective is to evaluate the effectiveness of various elements of an online offering based on response. Whether you are testing ad creative, navigation, or promotion, the basic approach is the same. We explicitly vary certain elements while controlling for others so we can clearly identify what has influenced a consumer.
The power of testing is in its simplicity - we intuitively learn through observation and experimentation. A great testing program is one where the process is fast enough that the marketer can get accurate answers before making any big moves.
Targeting is setting up rules to change the customer experience based on clearly defined situations. Most marketers are already doing targeting to some extent. Buying a keyword on a search engine is targeting, as in, "When the customer types 'Offermatica', show them my ad."
The bulk of landing page optimization is a combination of experimentation and targeting. When a marketer makes a decision about what page he wants his search traffic to land on (interior site page vs. home page), that's targeting.
Targeting's strength is its simplicity and transparency. At its most basic, targeting is about searching for groups of people that respond similarly. First time buyers may respond in one way, while those who buy several times a year respond differently, and those who have yet to make a purchase respond in a different way still. Successful targeting exploits the differences between those groups by showing different things to the different groups.
Targeting is by no means trivial. Behavioral targeting can be very sophisticated, from profiles or persona-based targeting to scenario-based approaches that can model a complete customer purchase cycle and target content by stage or maturity.
Predictive Modeling is really a refinement of targeting. The system or consultant essentially works to build a model based on prior behavior that can be used to predict future customer preference. With this approach, a broad range of variables (time of day, source of traffic, prior purchase) are evaluated against performance and a model is developed that is "fitted" to these conditions and that predicts how new customers will behave under similar conditions.
The benefit of these approaches is that the machine can potentially be more sensitive to changes in the environment and can determine correlations that a human would not have considered testing or targeting. The weakness is that predictive models tend to be expensive to maintain and create and are often opaque, leaving the marketer with little actual insight into what they should do next.
Predictive modeling is excellent for applications like search and cross-selling, where there are large sets of alternatives of a single type as well as a large number of interactions. The best users of predictive modeling tend to have very vertical marketing strategies.
The social approach to optimization harnesses the power of the community. One of the biggest challenges in measuring preference is gathering the data that helps you understand what a customer or potential customer wants. Through ratings, reviews, tagging and other forms of participation, marketers discover exactly what customers like and don't like. It's information that could never be gathered from watching clickstreams.
BazaarVoice, for example, offers a social approach to optimization by hosting the technology that allows consumers to post and read reviews, and to organize lists of top-rated products based on those reviews. Including a top-rated list that changes as ratings change is a form of social-based optimization.
Another good example is Digg or Reddit. There is very little magic math. You vote, the winner wins. The rules are comprehensible, and the marketer can apply and evaluate them. There are subtleties, but it is effective.
Bottom Line
So what is the best optimization approach? Optimization is just marketing with math. If your user base ratings improve the relevance of your search results, then do it! If testing helps to eliminate your CEO's bias towards acres of copy, do it! The marketing "mix" for optimization is going to take time to get right, but will yield tasty morsels of revenue improvement every step of the way.
And remember this: Marketing is done by marketers. Machines just help us listen and aim better.