Web Analytics: Personalize Your Site With Five Levels Of Optimization
We spend $1 out of every $100 of our media budget on the “last mile”: converting visitors who come to our site into customers. There are a few reasons for this lopsided ratio. One reason may be that we feel by filling up the funnel we will drive more traffic. However with only a small portion of that traffic converting, we believe it becomes simply a “numbers game” of pushing more volume. Another reason may be that we are simply not convinced or have not been exposed to the right set of tools designed not only to help increase site conversions but increase the quality of those conversions.
So how should you be investing in the “last mile”? Deploying a personalized site experience will provide the biggest opportunity for improving the visitor experience, thus helping to increase your qualified site conversions. A large part of this strategy is leveraging the right set of tools to convert online behavioral data into executable targeted strategies.
I propose five levels of site optimization to create a personalized site experience:
1) Multivariate testing – optimize the base construct of the site to define the winning combination of creative, content, and placement.
2) Onsite behavioral targeting – align content based on a prospects’ site browsing behavior.
3) Propensity targeting – target content to individual prospects based on likelihood to convert.
4) Prospect profiling – profile anonymous visitors based on pre-site behavior, and personalize content based on profiles.
5) Creative alignment – use website data to inform media as yet another valuable data source, with full visibility into how each new visitor performs once they land on the site by segment, to better align messaging and content.
Think beyond just site KPIs and consider revenue, loyalty, and profit as benchmarks for your optimizations. Do this by connecting to your CRM systems and databases to allow for better customization of the site experience and media to attract your most valued customers. Through the use of website behavioral data you will be able to better align your media creative and ensure sequential/consistent messaging is targeted to lookalike prospects.
Reaching an optimal state of site personalization is not enough, since we all know behaviors can shift over time. It is essential to employ a continual process of testing and optimization to guarantee a consistent level of performance. Create a process to address market fluctuations, creative changes, and website behavioral shifts. The process consists of six steps:
1. Audit – Conduct stakeholder interviews to gather historical performance data.
2. Define – This step is designed to translate the business objectives (e.g., account conversions) to key performance indicators (e.g., application starts). You will need to define existing and/or intended segmentation requirements (e.g. Urban Hippies, High-Rollers, etc.) and outline platform requirements necessary to track results, test, and optimize.
3. Plan – Design the optimization plan using defined objectives, and draft technical requirements.
4. Implement – Take the optimization design and implement it in-market.
5. Optimize - Once the optimization strategy is designed and implemented, this step assures ongoing and continual optimization against defined objectives.
6. Validate – Work to validate all optimization activities include comparisons versus forecasts and control groups. Once sufficient data is available, leverage web performance data to forecast the expected conversions. This could provide early warning signals of unexpected performance.
I feel that Define is one of the most important steps; this is where you select the appropriate optimization tool. This can be a complex task, given the number of potential tools in the space. Evaluate each tool on ease of implementation and estimated value.
Another critical step is Plan; this is what gets the process off the ground. Ensure that each test has adequate time in-market. This will allow you to get an accurate read on results without wasting effort and resources just to realize your sample was inadequate.
Last but not least, use experimental design in your testing to ensure you have a clean read, and control for external factors so you can attribute success to the right set of activities.