One of our junior programmers shot this one over after fighting with Proc Export for some time to remove the quotes from a string.
His need: export a single column of data without headers or quotes.
How he did it, quick and easy.
Sometimes the simple solutions are right in front of us.
Ever since you started with SQL or learned SAS, you've been warned about Cartesian products. But sometimes you need them. Shocking, we know.
Quick note of warning, when creating all possible pairs, make sure you've narrowed down your dataset to only contain a distinct list of those values you want to use in the pairs. Extra values make it more likely that you'll take the system down...
Here are some easy ways to produce all possible pairs of data.
example 1 - Using dataset and point
do _n_=1 to nobs;
set value_y nobs=nobs point=_n_;
example 2 - Using proc sql
create table all_pairs as
example 3 - Using hash
data all_pairs (drop=_:);
if _n_=1 then do;
set dataset_w_distinct_values (obs=1);
dcl hash h(dataset: 'dataset_w_distinct_values', ordered: 'a');
dcl hiter hi('h');
set dataset_w_distinct_values ;
do _rc=hi.first() by 0 while (_rc=0);
Often as marketers and analytics professionals, we are striving to measure every dollar. Yet, it may be equally important to get a read on things prior to the dollar figures. What about sentiment? Facebook knows. Google is learning. Pandora built a business around it. What about your org?
Sometimes we need to measure things faster and the dollars are more grey than black or white. Our advice? Improvise. Ask people what they think in 1 click. Yes or No? Like or dislike? Up or Down?
How is your organization leveraging sentiment to improve your marketing cycle?
Regardless of the industry or customer type, everywhere we go our discussions with clients gravitate towards the topic of real-time offers.
First, let me qualify what I mean by real-time.
Real-time is not:
- Picking from a group of offers available and presenting the one most relevant. There are many ways to accomplish this and although not every company has this ability, it is available.
- Putting an offer on your customer UI or website.
- Simply serving a predefined offer on a mobile device.
Real-time is augmenting an offer to be the right value proposition for a given customer at the time of interaction.
Customer X is going to make a purchase and is currently investigating offers from various organizations. Internally customer X knows that this purchase will bring him some measure of value. Customer X would purchase and commit to a 2 year contract with your company if you offered him 20% off instantly.
Customer Y is making the same purchase decision. Customer Y would also purchase from your company if you waived the startup costs however is unlikely to commit to more than 1 year of service.
Real-time is looking at historical patterns and augmenting the offer to meet the customers needs when a purchase decision is made. Predicting that Customer X should be offered 20% off for 2 years instead of 25% off.
The combination of profiling customers into complex sub groups based on historical preferences and performance and augmenting value propositions to meet the needs of the customer is much easier said than done. But companies that deliver this value and do it in multiple communication channels simultaneously take market share.
The first step to REAL-TIME is creating a partnership between your analytics group and technology teams. We stand in the gap between technology and analytics. We understand complicated data structures and the technology limits your organization has to work within.
If your organization is looking for ways to make a significant marketing impact with their marketing technology, contact us today.
Here is the scene. 3 development groups within the company all working on inter-related applications. Each group wants to make a name for themselves and get the cool projects from then on. Each one builds their software with marginal rates of success.
Then something else happens. Each one starts to add 'the grey area' to their app. What is 'the grey area'? It's those features that aren't clearly one apps responsibilities. So in an effort to provide great software, the team adds those features. But so do the other teams.
Additionally, those other features also create new problems. Business users aren't clear where to go to perform those 'grey area' tasks. Each unit is telling them to come to their software to do that. Unexpected results are produced by the software because each is going above and beyond their expected roles.
This scenario gets played out in company after company. This scenario is what makes enterprise architecture so important.
As a solutions designer, it is critical that the roles and responsibilities
for each piece of software be laid out as clearly as possible. Does that
mean 900 pages of requirements? NO!
It's means setting a clear mission for the software with clear integration points. It means collaboration between teams. It means less
competition between teams.
Clean Abstraction is a design goal for marketing systems. It means efficient development (no wasted time on redundant features). It
means clear business value. It means modular software that is
reused and reintegrated instead of rebuilt.
If you are building marketing systems / software from scratch or off the shelf, focus on clean abstraction. Marketing changes too quickly to waste development time on unwanted, or unnecessary features.
When we use words like analytics or algorithms, marketers eyes glass over. However, when we use words like dynamic content and positioning, they get very excited.
In reality, we may be talking about the same thing and it isn't brain surgery.
So, from a marketing / technology / analytics point of view, here are a couple of simple steps to improving engagement:
1. Build media outlets / channels (think website, mobile app, LCD touchscreen in your lobby) so that you can know who is interacting and possibly why (what do they want to know?).
2. Arrange content into categories such that they align with the who and why from step 1. This will make life much less complicated. Also note, content may be duplicated between categories.
Channel / Who / Why
- Primary Website
- Gold user in US
- Price shopping
- Main image1
- Store locatorMain image2
- Price shopping
- Gold user in US
3. Come up with a measurement strategy. An easy way to keep measurement simple is choose an engagement metric. This will be one number or metric that will tell you if users are interested in your content (maybe clicks, data entry, or downloads)
NOTE: Even on the best channels, not all content spots are going to engage users. Make sure that you are maximizing engagement when and where the users desires.
Most of all KEEP IT SIMPLE. This is the 'walk before you run' approach to dynamic content and analytics. The next level includes more words like algorithms.
One of the issues with SAS for many IT departments is the lack of integration with service oriented architecture (SOA). The good news is with many new features coming online with versions 9.X+ are service oriented.
Lets look quickly at the Proc SOAP procedure now available.
For those SAS programmers out there that aren't familiar with SOAP or services, get your basis here: SOAPUser-Basics
In a nutshell, SOAP is transporting XML data via a HTTP Post. In order to make a successful SOAP call from SAS you need a couple of things.
1. a request XML file
2. a repsonse XML file
3. a webservice URL and WSDL (Web Service Definition Language) -Think webservice users manual
Here is a simple example of a SOAP call we use on a daily job.
filename rqst_xml 'some file system reference';
* Create the XML;
if first.records=1 then do;
put '<?xml version="1.0"?>
if last.records=1 then do;
filename rspns_xml 'some file system reference';
proc soap in=rqst_xml
Upon executing the call, you can read in the rspns_xml data with the SAS XML engine.
This is meant to be a simple example with very limited scope. Service architecture can quickly get complicated with error and condition handling. Please let us know if you need help with your SAS architecture or coding.
Do you think your group is forward thinking and tech savvy?
How about your customers? They are probably smarter than you think..
Can you describe cloud computing? What about when you were 10?
Our friends over at Accenture put this together.
As we begin to add posts in 2011, we stepped back to look at our business and our blog. We realized that the two are perfectly aligned. Although the majority of our focus is within the SAS product suite, our focus is marketing analytics and operations.
Given that, it is time for us to expand our commentary. So, in the coming weeks you will begin to see more content regarding some of our other technology partners including Teradata, Netezza and Unica.
If you have a question about a particular technology or topic, we'd love to hear from you. Comment below or email us a firstname.lastname@example.org .
There are few things that seperate the good marketers from the bad. In our opinion contact strategy is one of those things that few organizations understand and all should consider. When you are trying to decide who you should contact, through what channels and how many times, here are a couple of things to think about.
1. Not all customers are created equal - Nor do they have the same appetite for your marketing efforts. We recommend you spend time looking at your customers and how many times you contacted them in the past. A good cluster analysis will begin to help you understand that some groups prefer 10 'sale' messages and 2 'partner' messages before opting out. Put this cluster with other customer breakouts and you have the basis for your contact strategy.
2. Not all messages are created equal - As we eluded to above, you can not simply look at the recency and frequency of your messages to determine how much is too much. You need to understand the message types or objectives. Customers have varying appetites for your credit card solicitations and that appetite is much different from your best customer sales messages.
3. Implementing your strategy may be harder than developing it - So now you know what you want to implement. Customer segment X gets Y transactional messages a month and Z surveys a month. But how do you insure that they only get what you've prescribed? There are a variety of ways to accomplish this with either process changes and / or technology solutions (One of our favorites is Unica's Optimize.).
If you are considering a contact strategy for your organization, we would love to help you understand your customers and improve your communications.