Follow me on Twitter: @pricingright
Every time I have a conversation with someone who supports and strongly believes in a freemium model, I hear two main revenue streams attributed to freemium customers: [tweetmeme http://klck.me/Agx source=”pricingright”]
- Initially customers do not know they need this product but when then start using it they fall in love with it and upgrade from free version to paid version.
- Customers that adopt early and fall in love generate considerable WoM marketing for the product
Since there are two levels (o and 1) for these two states, any user can be in any one of four following states.
After a few months, anyone continuing to use the service and not generating any WOM revenue is a freeloader. Only when your users start paying they become your customers. When freeloaders generate WOM marketing (preferably revenue) they become do-gooders. When users become customers and generate WOM marketing, they are your apostles. These state transitions and triggers are shown in the next image.
Here are my questions to you:
- Do you know what percentage of your users will end up in each state?
- Do you know how long anyone will spend in the freeloader state?
- Do you know how you can facilitate the triggers out of Freeloaders state?
- Do you have a model for WoM marketing revenue that a Do-Gooder will generate?
- Do you know how to keep customers and apostles in their states without sliding back?
If the answer to any of these questions is
- anything that is not expressed in numbers (I will take 90% confidence numbers )
- not specific to your business
- based on cost arguments
- something that uses Evernote or Dropbox numbers
Are you sure you should be adopting this model, or even undertaking this venture?
Now my selling points:
– what is the information worth to you before you invest resources?
– what if I told you I have a predictive model for finding the probability distribution of where a user will end up?
[tweetmeme source=”pricingright”] This is the results from the consumer behavior experiment I conducted.
Footnote: The deck could definitely use improvement and you will find it hurried. I found it better to share the results right away and improve the deck iteratively. I still need to work on the marketing implications of these findings.
Please let me know of methodology errors , flaws and or other comments.
For Forrester’s report on which companies are recommended most see here. If customers do not know he extent to which their purchases are influenced by their friends, should business continue to invest their marketing dollars on WOM?
[tweetmeme source=”pricingright” single_only=”false”] Effect of Word of Mouth (WoM) marketing is intuitively accepted by most and a few peer reviewed research in the journals of Marketing Science and Marketing Research found evidence supporting the predictability (somewhat weak however). There are two key shortcomings I see with many of the popular Word of Mouth Marketing metrics:
- One that is most acknowledged in any marketing research study is the gap between customer’s stated attitude and their actual behavior. When the customer answers the question on paper (or screen) they are asked to answer about their expected behavior but the answer is still rooted in the attitude. For example, when you ask a customer about likelihood of purchase, paying for services or recommend the product/service to others the customer states their current attitude and not what they will do when presented with the exact situation in the real world.
- Two is trying to compute a net metric based on the stated attitude level. They result is data loss (because some responses are thrown away) and the need for extremely large number of samples to get statistically significant results. According to research published in Marketing Research, (Summer2007,Vol. 19 Issue 2) up to 26 times more samples are needed for net metrics.
I propose here a newer and actionable metric – iTRM (iterative Total Recommendation Metric). This addresses both the shortcomings listed above. First, it gets closer to actual customer behavior when it comes to recommendations than any of the metrics. Second there is no data loss, all responses are counted. Here is how the metric is calculated.
Customers are specifically asked about their actions, how many times they will recommend. The scale is -5 to 5. Negative values mean customers will be “de-recommending” your products. iTRM offers multiple advantages over existing metrics:
- The time horizon is specific and limited, 1 month. The time limit however does not imply that the measurement must be repeated every month but if the metric is as a predictor of sales it requires monthly measurements.
- The customers do not have to make subjective judgments about rating scales. For example a 8 for one could be 6 for another in many ordinal scales measuring likelihoods. Here it is the exact number of actual recommendations.
- This is simple to administer and measure. It can be either a simple sum or can be continuously improved to be a weighted sum that is different for different industries and even for specific brands based on past data. (The word iterative stands for both my brand and the iterative process by which the metric evolves).
- The limit of 5 is not a shortcoming given the specific action and the time horizon. If for a particular brand we find the need for higher scale, it can easily “iteratively” improved upon and still be consistent with past results.
- It is easy to validate its predictability and improve on the weighted calculation.
Overall, I believe iTRM provides a better measure of WoM with potential to use and easily validate as a predictor of growth.
How do you measure your WoM?