Contact Details
Telephone
650.804.2884
Electronic mail
Fax
206.282.7930
Mailing Address
2117 W. Ruffner St.
Seattle, WA 98199
USA
Projects
Our specialty is working with clients to design and execute a statistical analytical approach to meet their project goals in the most efficient way possible. Here are some of the projects and kinds of questions we have handled for our clients:
- Business Analytics
- Evaluation and Metrics
- Investment analysis
- Product and Manufacturing reliability
- Biostatistics
- Risk management and fraud detection
- Injury and Damages Assessment
- Survey and Sample Size Analysis
We also offer seminars specializing in using statistics as a tool for decision making.
Business Analytics
This field includes a wide variety of analyses - from statistical forecasting and customer analytics to improving advertising mixes and analyzing price elasticity. As with other areas of our work, if you don't find an exact match here, please contact us to discuss your specific project concerns.
Predictive Modeling
Suppose you want to provide your employees with an internal tool that helps them predict various metrics? Or say you have a website on which you would like to provide a service that predicts your users salaries, home prices, risk of heart disease, or helps you determine what products or services they are likely to purchase from you. All these various services at their heart a statistical model that can be developed on existing data, and predict outcomes based on new data.
For example, based on information you collect about your customers, or their behavior as they browse through objects in your store or website, the model can give likelihoods of which products or services ehy may be most amenable to buying. It provides you with invaluable insight into how to approach your customer. The forecasting model makes this level of individualized prediction. People being people, the model may one time out of a hundred ask you to sell a Hello Kitty ceramic figurine to a grizzled biker, but most of the times it will make more accurate predictions. And who knows, maybe the model picked up on something.
For example, you may want to provide a service to your customers or organization where they can get accurate estimates about your services or products. For example, salary estimates, property prices, changes in crime rates, radon levels, etc. All these can be put into an interactive web form where your users input their data, and the web calculator will provide an estimate specific to their situation. The underlying statistical model is something we can develop and help you implement.
Case Study
An national organization wanted internet to be able to figure out their salary potential on their website. They performed a detailed survey of its members. The survey data was used to create a statistical model that was used on the website to give each user personal reports based on their inputs:
- 80% of the organization's members paid their dues so as to be able to access the calculator, the majority of these were new members.
- Visits to the website surged initially by 50%, settling to 35% after 6 months.
- Prediction was within 10% of salary + other sources of income for about 90% of users.
Customer analytics
Often marketing and sales work better if you know how your customers break down in terms of demographics and your own industry-specific characterizations. But how do you know which segments are statistically reliable and stable over time? Knowing (rather than hoping or guessing) the answers to these questions can save you lots of marketing dollars over the long term.
We can also combine your survey data with your sales, and mine it to determine what types of customers are likely to gravitate towards your main products and services. This type of statistical analysis has rich potential and can provide actionable insights into how to make your products or services more appealing to under-serviced markets.
Marketing analytics
A lot of marketing dollars are wasted on blanket advertisements: why spend hundreds of thousands of dollars on a TV spot advertising your teen clothing brand when only about 20% of your audience is a potential customer? (Why spend anything pitching a new blackberry flavored bubble gum to a 65 year old man with diabetes?)
If you're dissatisfied with your advertising outcomes and have a lot of data, we can help you develop a better advertising strategy based on statistical analysis. For example, we can segment your populations, identify which campaigns or promotions have worked in the past, help design methodologies for measuring the effectiveness of new campaigns, build predictive models for pitching certain products or ads at specific populations and build short and long term measurement models for tracking ongoing performance.
Advertising mix
Determining which advertising channel has the best ROI is an important first step. Sometimes, the answer isn't always the most expensive medium (making COO's the world over ecstatic). Having tried all possible channels, how do you determine which one actually worked best? After determining your hierarchy of goals (increase market share, revenue, etc.), we can analyze your advertising data to determine what works best for you.
Case Study
A national real-estate company was designing a new TV ad campaign, and wanted to figure out what the potential return in revenue would be. A secondary goal was to see how a campaign with mixed radio and print ads would increase revenues.
The initial statistical analysis showed:
- TV actually had very little impact on revenue increase, but it did increase their brand awareness.
- Print ads seemed to be their primary driver of extra revenue.
These conclusions led to a more complex analysis investigating the mixture of ads.
- Online advertising came out with the top ROI.
- The effect of print ads had a 7 week lag, while TV's best effect was 1 week, but the revenue impact remained unchanged
Price elasticity
Pricing of products and services is often a major uncertainty for companies. It is also one of the levers you have most control of so determining it with evidence and certainty can dramatically impact your bottom line.
If you are charging too little for your products or services, you are losing out on revenue today that you may never recover, or will be less valuable if you manage to get it at a later period. But if you are charging too much, your customers will go elsewhere or decide they don't need your services.
Finding the sweet spot, the price at which you are maximizing your revenue, does not have to be a blind process. We can bring a lot of analytical support to help you determine how demand and revenue may shift with changes in price. We can use your historical pricing data, or pricing data of products and services similar to yours, to determine where revenue is maximized. This process takes into account seasonal factors, unusual events, promotions, etc.
Don't worry if you don't have all the data you need yet. We can also advise on methods for collecting data to determine price elasticities down the road. We've had many clients argue it was the best investment they ever made.
Case Study
A retailer in California seeking to enlarge its online sales commissioned a study to predict the revenue effects of raising prices of certain categories of products. As the analysis progressed and intermediate results were communicated to the client, changes were made to the project plan to focus on the most promising geographical markets.
New methods were also developed to account for temporal trends (the company was experiencing growth, and was observing seasonal fluctuations) and findings included:
- Their elasticities were seasonally dependent - so that a small decrease in price during the back-to-school season and Christmas resulted in a much larger increase in customers.
- One category was price inelastic, prompting the company's management to consider making this their primary product.
- Subsequent data showed that the decisions based on the predicted price elasticities resulted in an increase in revenue that was within 3% of predicted revenue. This allowed them to create forecasts and budgets with a very high level of confidence even as their company continued its rapid growth.
