Using Social Data to Power Real-Time CRM Optimization
The rise of programmatic digital media has dramatically improved customer relationship management. No longer limited to email or direct mail, marketers can now reach customers within a variety of ad formats across a broad range of digital environments. Access to a broad range of sales data sets and lookalike modeling from digital publishers such as Google, Facebook, and Amazon allow for exponential expansion of the standard audience profile. In addition, DSPs have made it possible to reach a targeted audience in real time wherever they are consuming content online, bringing marketers one step closer to achieving the dream: right person, right message, right place, and right time.
While CRM has come a long way, optimizing for the right message is still a considerable challenge. For example, it is undoubtedly useful to learn that response rates on a targeted campaign have suddenly and unexpectedly declined, but knowing exactly why could be a game-changer. The question is: how does one find this information? Surveys are costly, time-consuming, and limited on discovery (you can’t discover what you don’t ask), and data received via DSPs and click stream isn’t descriptive enough for marketers to make substantive creative changes.
Introducing Social Media Data: the White Knight of Analytics
When properly curated and analyzed, social media data offers a viable solution to marketers’ need for real-time creative optimization against specific audience segments. This inexpensive and longitudinal data source marries the strengths of behavioral data (large, organically generated, statistically relevant data sets that can be easily benchmarked against past results) with the strengths of survey data (descriptive and open-ended data sets that allow for qualitative analysis). This unparalleled discovery and insight, anchored in quantitative analysis, gives stakeholders confidence in the results and empowers them to make informed decisions.
Like all data sets, social media requires some polishing before it’s usable. Here is how you can optimize social media data to get the best results:
Establish User Social Handles to Normalize and Properly Classify CRM Audience Segments
One of the biggest challenges with social media data sets is the integrity of the data. As the famous New Yorker cartoon jokes, a user categorized as a cat or goat could actually be a dog (or, more likely, a spam bot). Because digital behaviors change quickly, it’s also very difficult to normalize comparison data sets; for example, Facebook today looks very different than it did five years ago, as younger users have fled to Instagram and Snapchat. The solution? Match the rich, validated data of your CRM database with users’ social handles. Rather than mistakenly measuring dogs who behave like cats as cats, you can know they’re dogs and marvel at their catlike behavior.
IBM’s product Big Match for Hadoop is a useful tool for matching owned customer data to public social handles. The product was designed to take large volumes of structured and unstructured data and enable fast, efficient linking of data sources, providing marketers with complete and accurate customer information.
Use Machine Learning to Expand Each Audience Sample
If the marketer’s CRM database is large enough and the audience segment is broad enough, there may be enough handles to constitute a viable sample from Big Match alone. More often than not, this first round of handles needs to be expanded. This can be done using a tool such as IBM’s Watson technology, which can apply supervised machine learning techniques on the first batch of handles’ social data to discover behavioral patterns and find lookalikes.
Monitor Your Audience’s Social Behaviors to Deliver a Broad Range of Insights
Segmented social account lists give marketers the capability to monitor thousands of live user conversations across the social web, with customer data that goes back for years. By monitoring social behaviors of key audience segments, you can deliver daily insights on trending topics throughout the campaign to aide in creative planning, media campaign effectiveness, and competitor benchmarking.
Here are a few examples of insights you can gain using this new behavioral data:
• Instead of blindly guessing which post topic will resonate with your Facebook audience most during this year’s Super Bowl, reference a list of topics they’ve discussed recently and which posts they engaged with most during last year’s game to make a more educated decision
• To determine the impact of a new spot used in a video pre-roll campaign, analyze lift in social volume around your product or brand’s key terms, and compare themes and topics surrounding these key terms with those of a previous spot
• Compare how your spot fared against a competitor’s across your priority audiences by benchmarking to your competitor’s key product and brand terms
This highly effective strategy for social media data analysis can be summarized in four easy and straightforward steps:
1. A marketer identifies target audience segments from the CRM database (ex “Past Purchasers”)
2. Social handles are identified and appended to a representative sample of each audience
3. Audience segments can be expanded to increase sample size using machine learning capabilities
4. Social media behaviors of audience segments are monitored to deliver useful insights
The expanded opportunities of real-time marketing lead to expanded expectations for real-time insight. Social media data, combined with the advanced analytics toolsets offered by companies like IBM, gives you the ability to deliver against, and exceed, expectations.