Omnichannel shoppers have been disrupting retailers for years, and its likely to top the industry’s agenda of challenges for years to come. But optimization, an omnichannel analytics technology, can help harness the positives of omnichannel retailing and minimize showrooming. Consider this everyday retail dilemma: E-commerce sales are growing, but in-store […]
One of the most powerful sales tools is often something that you can’t foresee or control. Even though customers read papers, visit websites and talk with a salesperson, another factor can make all the difference – a referral from a friend or coworker.
Think about the way that sites like Google, Yelp and others have changed the way consumers make everyday decisions, such as choosing restaurants. You can go to the restaurant nearest you or one you’ve visited before. Or, you can try something new by looking at your smartphone to see which dining spot has the highest ratings or the best reviews. Why? People show a preference for the personal experience of those in their networks.
For business-to-business software companies like SAS, the impact of customer advocacy is critical. These influencers can set the tone and provide a consistent positive influence throughout the customer journey. Unfortunately, this type of advocacy is tough to measure and hard to predict.
The challenge: Acquisition and retention
Although a customer may be a single record in your database, she doesn’t exist in a vacuum. Each contact has a connection to others within her business or the industry. Understanding and fostering good relationships can have a huge effect on your retention and loyalty efforts.
During our effort to map a modern customer journey, the SAS marketing team focused on different phases of this cycle. The customer journey contained these phases:
- Acquisition – which includes need, research, decide and buy.
- Retention – which includes adopt, use and recommend.
On the retention side, the team knew from anecdotal evidence that some SAS customers were advocates of the technology and for the company overall. In fact, several SAS regional offices and divisions had data confirming the idea that finding and rewarding high-value customers led to big returns. What was lacking was an overarching program for getting customers to advocate for SAS technology.
For a larger effort, the team assessed the customer behavior data, examining those who attended events, provided feedback on surveys, sent ideas to R&D, and generally stayed engaged with the company. From a revenue standpoint, those people were often the ones advocating for the use of new SAS technologies or the expansion of existing deployments.
What was less understood was the reach of these influencers and how their activities affected others. With that information, SAS could identify more advocates and nurture that behavior.
The approach: Identify advocates by scoring BFF behaviors
The SAS marketing team members started by digging into the data that they had on customers. They first identified a segment of the top accounts that contained more than 20,000 individual contacts and the team began to examine the behaviors exhibited by that group including:
- Live event attendance.
- Website traffic.
- Technical support queries.
- Customer satisfaction survey data.
- Customer reference activity.
- Webinar attendance.
- White paper downloads.
This information provided a better understanding of the range of activities that customers undertake. However, simply cataloging the behaviors wasn’t enough. The team applied a scoring model for different types of interactions. This allowed the team to weight certain activities, helping to further identify which customers were the best advocates—“BFFs” (best friends forever) as the marketing team began to call them.
The results: Advocacy campaigns that matter
SAS marketing used the information to create a model that is the foundation for customer-focused data exploration. The initial effort helped shed light on how influential advocates can shape retention and additional sales. As a result, sales and marketing worked together to highlight BFFs within key accounts in an ongoing effort to foster better relationships with those key individuals.
Initiatives to locate and encourage advocates used the model to identify the likely candidates within customer organizations. The team then designed campaigns and outreach efforts to give these advocates the tools to foster and expand their influence.
The marketing team now focuses on advocacy campaigns that target potential BFFs. The goal is to build more SAS advocacy during the recommend phase of the customer journey.
Acquisition and retention campaigns begin by doing advanced segmentation in SAS Marketing Automation. Campaign workflows are created that are backed by analytics, ensuring that communications to customers are appropriate and relevant. Through the collection of both contact and response history data, attribution can be performed in SAS Visual Analytics that allows marketers to see correlations and cross-promotion opportunities.
Interested in learning how to leverage SAS Marketing Automation techniques for advanced segmentation? Explore our SAS Marketing Automation: Designing and Executing Outbound Marketing Campaigns and Customer Segmentation Using SAS Enterprise Miner course offerings.
Editor’s note: This post is part of a series excerpted from Adele Sweetwood’s book, The Analytical Marketer: How to Transform Your Marketing Organization. Each post is a real-world case study of how to improve your customers’ experience and optimize your marketing campaigns.
1. Don’t be afraid to launch a product without machine learning.
2. First, design and implement metrics.
3. Choose machine learning over a complex heuristic.
4. Keep the first model simple and get the infrastructure right.
5. Test the infrastructure independently from the machine learning.
6. Be careful about dropped data when copying pipelines.
7. Turn heuristics into features, or handle them externally.
8. Know the freshness requirements of your system.
9. Detect problems before exporting models.
10. Watch for silent failures.
11. Give feature column owners and documentation.
12. Don’t overthink which objective you choose to directly optimize.
13. Choose a simple, observable and attributable metric for your first objective.
14. Starting with an interpretable model makes debugging easier.
15. Separate Spam Filtering and Quality Ranking in a Policy Layer.
16. Plan to launch and iterate.
17. Start with directly observed and reported features as opposed to learned features.
18. Explore with features of content that generalize across contexts.
19. Use very specific features when you can.
20. Combine and modify existing features to create new features in human understandable ways.
21. The number of feature weights you can learn in a linear model is roughly proportional to the amount of data you have.
22. Clean up features you are no longer using.
23. You are not a typical end user.
24. Measure the delta between models.
25. When choosing models, utilitarian performance trumps predictive power.
26. Look for patterns in the measured errors, and create new features.
27.Try to quantify observed undesirable behavior.
28. Be aware that identical short-term behavior does not imply identical long-term behavior.
29. The best way to make sure that you train like you serve is to save the set of features used at serving time, and then pipe those features to a log to use them at training time.
30.Importance weight sampled data, don’t arbitrarily drop it!
31. Beware that if you join data from a table at training and serving time, the data in the table may change.
32. Reuse code between your training pipeline and your serving pipeline whenever possible.
33. If you produce a model based on the data until January 5th, test the model on the data from January 6th and after.
34. In binary classification for filtering (such as spam detection or determining interesting emails), make small short term sacrifices in performance for very clean data.
35. Beware of the inherent skew in ranking problems.
36.Avoid feedback loops with positional features.
37. Measure Training/Serving Skew.
38. Don’t waste time on new features if unaligned objectives have become the issue.
39. Launch decisions are a proxy for longterm product goals.
40. Keep ensembles simple.
41. When performance plateaus, look for qualitatively new sources of information to add rather than refining existing signals.
42. Don’t expect diversity, personalization, or relevance to be as correlated with popularity as you think they are.
43. Your friends tend to be the same across different products. Your interests tend not to be.
Last week I wrote about the 10 most popular articles from The DO Loop in 2016. The popular articles tend to be about elementary topics that appeal to a wide range of SAS programmers. Today I present an "editor's choice" list of technical articles that describe more advanced statistical methods in SAS.
I've grouped the articles into three categories: statistical graphics and visualization, statistical computations, and matrix computations. If you are a SAS statistical programmer, these articles deserve a second look.
Ten posts from The DO Loop that deserve a second look #SASTip
Click To Tweet
Statistical graphics and visualization
SAS ODS graphics provides an easy way to create standard graphs for data analysis. The graphs in this list are more sophisticated:
- Have you ever struggled to specify the order and colors of categorical variables? This article provides a general technique that is often useful: insert special observations at the top of the data before you create the graph. A related technique is to append observations at the bottom of the data so that you can visually represent special values in a graph.
- If you want to visualize complex regression models in SAS, you MUST learn about the EFFECTPLOT statement! An effect plot shows the predicted response as a function of certain covariates while other covariates are held constant.
- For time-varying processes and iterative methods, animation is a valuable visualization technique. Learn how to create an animated GIF is SAS by using the BY statement in PROC SGPLOT.
These article show helpful statistical techniques that you should know about:
- Confidence intervals are an essential tool in inferential statistics. This article uses simulation to answer the question "What are confidence intervals?" A related post shows how to compute confidence intervals for a multivariate mean.
- Most SAS procedures include a CLASS statement for handling the analysis of discrete classification variables. However, for some advanced statistical methods, you might need to generate a design matrix, which is a set of variables that represent categorical variables and interactions in a regression model. This article describes four SAS procedures that can generate a design matrix.
- Some statistical algorithms (such as clustering) rely on computing the distance from an observation to its nearest neighbor. This article shows how to compute nearest neighbors in SAS. A related article shows how to compute the distances between observations in one group and observations in a different group.
The SAS DATA step is awesome. For many programming tasks, it is an efficient and effective tool. However, advanced analytical algorithms and multivariate statistics often require matrix-vector computations, which means programming in the SAS/IML language.
- SAS/IML 14.1 introduced packages. Packages are a new way to share SAS/IML programs. You can watch a video presentation about how to use and create packages.
- Statistical programmers often use the SAS/IML language to run custom optimizations. To help you navigate common pitfalls, this article presents a checklist of 10 tips to ensure that your optimizations are correct and efficient.
- SAS/IML is often used to carry out efficient simulation and bootstrap studies. This article describes how to implement the smooth bootstrap method in SAS/IML.
- Because SAS/IML is a matrix language, it is the ideal environment to implement Markov transition matrices. A related article shows that certain probabilities in Markov chains can be computed in terms of properties of the transition matrix.
There you have it, 10 articles from The DO Loop in 2016 that I think are worth a second look. Did I omit your favorite article? Leave a comment.
Not too long ago (last month), in a theater not too far away (10 minutes from my house), I saw Rogue One: A Star Wars Story. According to CNN Money, the standalone Star Wars movie brought in $29 million at the domestic box office on opening night, making it the […]
Streaming technologies have been around for years, but as Felix Liao recently blogged, the numbers and types of use cases that can take advantage of these technologies have now increased exponentially. I've blogged about why streaming is the most effective way to handle the volume, variety and velocity of big data. That's […]