There’s a famous adage that says, “You can’t manage what you can’t measure”.
In simple words, it’s impossible to know what’s working and what’s not, if there are no defined metrics to measure success.
On the other hand, the vague “catchall” tactic to marketing used to deliver results for some companies, and the push for data-driven marketing initiatives comes as a response to the customer’s demand for more personalized experiences.
Based on a study by Janrain, % of online consumers said that they get frustrated with websites when content appears that has nothing to do with their interests.
So, to compete in this new highly competitive marketplace, brands and agencies need to put a greater focus on analyzing, generating, and applying data to their client’s campaigns.
Not sure where exactly to focus your efforts? Don’t know what data you should be gathering in the first place?
Data-driven marketing is the new hot topic today and is exploding the marketing landscape. With easy access to huge new data sources, and the new tools like artificial intelligence (AI), and the insatiable demand for data to support digital marketing communications, data has turned into an essential resource for savvy marketers.
In this blog, we are discussing the new developments in data-driven marketing to customers, to bring you up to speed with what’s new and what’s working. So, let’s look at each topic in turn. While some are applications, and some are tools, but all are worth knowing about.
Prospecting
Finding new customers is a never-ending objective for marketers. Generally, mass marketers used to target their audiences based on demographic factors like income, age, geography, and gender. But direct marketers earlier found an even more effective method, looking at past buying behaviour. Recognizing the predictive power of past buying behaviour in the 1990s, the mail order catalog industry developed cooperative databases and shared customer transaction data among themselves.
Now, transactional data is no longer the exclusive province of insiders. Customer purchase history, by product or service category, is accessible to all through commercial databases. It’s been proven several times that behaviour data is much more effectively predictive of future purchase behaviour than compared to demographic data. The marketers can now easily access behavioural data on millions of customers to boost their prospecting via countless channels like direct email, email, display advertising, phone, and social media.
Predictive modelling, a long-time backbone of data-driven marketing, is usually structured as follows: A big set of customers is divided into two halves. For one half, a regression model is built to identify the characteristics of the population that took a particular action such as downloaded, clicked, bought, subscribed, defected, or returned.
The other half is used for validating the model: Did customers in the first half, with the same characteristics also show a higher propensity than average to take that action? If so, the model is sound.
Predictive modelling has helped marketers for decades to select the most likely targets for their campaigns. And with the new advances in statistics and new sources of data, marketers are now leveraging AI and machine learning to enhance their modelling capabilities. Even the unstructured data like and social media posts and text from call centre notes can be included in the combination of additional predictive power.
Machine learning is probably the most exciting recent development in predictive analytics. In machine learning, the desired result, like the purchase or the click, becomes the focus. The computer is requested to identify all the characteristics or actions in the past that led up to that action. With that information in mind, your predictive model can be more precise, and also be set to improve automatically over time.
The next frontier of predictive analytics is AI. For example; a company XYX working to help another publisher company ABC, increase its webpage engagement levels by the most relevant articles to cater. The AI engine observes the visitor’s reading behaviour on the website, identifies their interests, and then suggests additional articles or other resources to read or explore. By mixing offline behavioural data with classic predictive analytics techniques, real-time AI is now converting readers to paid buyers, generating more opportunities and maintaining stories of interest for advertisers to sell affinity products and services by pairing online with offline interests. The AI models are already generating a 70% attention rate for click through to articles of interest.
Lookalike Modelling
One of the most commonly used modelling strategies; lookalike modelling applies linear regression to analyze customer records and identify the characteristics that mold the customer’s behaviour. Then marketers can go to the bigger audience and find similar customers-lookalikes-who share those same characteristics. This technique works well in prospecting, where the characteristics of your best customers can help identify your best potential prospects; and those of your weakest customers can be used to suppress marketing investments in less productive areas.
Example; a non-profit organization asked the Infotanks Media team to model their higher-dollar donors, in this case meaning above $10. The entire Infotanks Media database was then scored for similarity to these top donors and then divided them into 20 individual groups, ranking them from highest to lowest. The marketer was then able to send campaigns to the groups with considerable improvement in the response rates.
Cross-Sell and Up-sell
A standard application of data-driven marketing is finding new ways to sell more products or services and higher-margin products to existing customers. One classic tactic is looking at past purchases to identify up-sell behaviours or the next best product, and then offer existing customers the products their lookalike counterparts purchased in the past. This strategy has been effective for decades.
But what’s new is the capability to broaden the true power of your model beyond just your existing customers. The Infotanks Media team can add more useful data points to the mix to improve the model’s effectiveness. Their analytics group will also create the model for you.
Customer Insight
The prospect and customer data can be a treasure trove of insight into customer behaviour and needs as part of your market research initiatives. Many researchers do primary or original research via interviews, surveys, and focus groups, they should also see at their customer data, especially the behavioural data, which can give better insights and the answers for market research questions.
Answers to such questions can be easily found as: What are my customers buying behaviours and frequency? What are their demographics? Which channels do they use to make a purchase? What are their lifestyle interests and key product(S)? What are their existing spending levels? How do their purchasing habits compare to typical buying households?
Example; publishers use MRI data that is based on surveys to help describe their readership to advertisers and for circulation management. But when the same publisher runs their subscriber files through Infotanks Media database, they usually come up with a very different and accurate view of their reader profile.
Data Append
Every marketer has holes in their database- it’s inevitable. But they can both fill in the voids and enhance existing customer information by appending certain data elements bought from third-party data vendors. The owners of large databases usually make their data available for appending to your in-house database file. You can overlay your file including such critical data fields as income, gender, education, and a most critical one- buying behaviour.
Compared to the data collected directly from prospects and customers, data appending is an inexpensive, fast, and convenient way to enrich your database for purposes like analysis, research, campaign, and modelling selection.
Appended data don’t just fill in the blanks, but it actually can also provide information, additional insights that will help you better select, understand, and target your prospects and customers effectively. Example; an insurance company could easily append data to its customer file, providing them with opportunities to provide insurance at particular times like buying a house, graduation, birth of a new baby or retirement. Or, a jewellery retailer could easily append purchase history like watches and rings, to identify new cross-selling opportunities.
Before you start, you must decide on the fields you want to append. While appended data is much cheaper than proactively collected data, there is still a cost associated with it. So only purchase the data elements that will drive measurable value.
Segmentation
To build segments representing customers who are externally different and internally alike, a mix of demographic and behavioural data come into play.
Example; According to Jessica Best, director of data-driven marketing at Spirit’s Kansas City agency, Barkley- Spirit Airlines has millions of prospects and customers in their email database. And to help Spirit optimize the customer relationship through relevant email communications, Barkley’s data science team identified five targetable segments in the database: infrequent flyers that pay extra for upgrades, infrequent flyers who take advantage of the cheapest flights, super fans- those who could become super fans, and detractors or unhappy fliers.
For Barkley to apply these segments to marketing, they are working with Spirit to use a real-time algorithm to type their database. And as the new customers come on the scene, Barkley’s algorithm would be able to match 95% of the records to this segmentation model, enabling Spirit to send relevant offers and messages to enhance customer value throughout the customer life cycle.
Profiling
Data is a huge game-changer for customer profiling that traditionally was done through demographic segmentation and a lot of guesswork. But now you can easily create a customer profile based on facts and actual behaviours.
Example; a home improvement brand recently asked for help to understand how to move a prospect along a multi-step sales process. The steps involved developing an inquiry, following up with a phone call to schedule a sales appointment, and finally closing the sales. The team is generating a series of data-driven profiles that will indicate the most likely prospects to move from stage to stage.
Marketers apply customer profiles across the entire go-to-market process. Profile them by segments to achieve rich insights into customer behaviours and needs, and to develop product or service offerings that resonate better. Use the gained insights to develop more persuasive and precise marketing communications. Apply the profile of an ideal customer to single out the top prospects for extra attention and improve the efficiency of your marketing investments.
Defection Prevention
Marketing initiatives to retain a customer produces the highest ROI among all marketing investments. And a 5-point decrease in defections can lift per-customer profit by 25% to 85%. Analytics and data give an essential tool in the effort to avoid detection, enabling you to accurately identify with amazing accuracy a customer who is on the way out the door.
Again, a predictive model is developed by analyzing the characteristics and behaviour of past defectors, and then scoring current customers on their similarity. An ounce of preventive marketing like a personalized message, a special offer, to these at-risk customers can return a pound of profit.
Customer Win back
Despite your best efforts for prevention, a tragedy may eventually happen; your customer lapses. The marketers now know that quick action is critical; trying to win them back. And in this case of defection, a reactivation model can be your best friend.
Example; for a publisher of a health newsletter, the analytics team developed an “expires” reactivation model. And to do so, they looked at the expired subscribers who had responded positively to prior reactivation campaigns. These names were matched against the Infotanks Media’s huge database, adding more data variables to each of the records to feed the model.
These enriched data records were modelled, and the publisher instantly gained additional insight that the most likely reactivation candidates were over 50, showed a recent mail order purchase activity were recent movers and/or book buyers- data they could easily use for multiple purposes including product development and marketing messages. Then, the publisher’s recent expires were scored so the publisher could choose the groups and suppress the bottom groups to make their win-back program much more efficient. And the proof of the model’s success came with higher reorder rates.
Mobile Marketing
When it comes to B2B, mobile marketing is still in its infancy, but, its being enthusiastically embraced everywhere by marketing communicators. And compared to the other channels, mobile is especially data-intensive. Mobile produces important data elements which are not available elsewhere like geo-targeting and in-app behaviour.
It is quite common knowledge for the marketers that they have to optimize their landing pages, websites, and email messages to the mobile device platform. This is the first step. Then the data part comes in. The marketers eventually will want to connect their customer’s mobile behaviour data to their behaviour via other channels.
This is a critical challenge, but tools to make mobile data integration easier are slowly emerging. And in the meantime, workarounds are available like persuading customers and creating your app to opt into SMS messaging.
Social Media
The social media behaviour is delivering a world of fresh data with huge marketing value. The data-driven applications seem endless. One of the basics is social listening that involves analysis of text comments about brands and products. Then there’s data-driven marketing and advertising through social websites like LinkedIn, Facebook, Twitter, etc. Facebook has taken a strong lead with their lookalike modelling that Facebook’s statisticians provide. Ads that serves highly targeted prospects can be delivered in several formats, the most popular being messages in the news feed of the prospects and display ads along the edges.
But there are some other exciting data-driven strategies which are emerging these days. Some companies provide social media appending services (like Infotanks Media’s Social Media Appending), by which you can easily add social handles to your existing prospect and customer file, to effectively communicate with them through new channels.
Other offers to reverse-append are your customer records to your existing Facebook and Twitter followers, enabling you to further expand your sales and marketing database. There’s no end in sight of the new opportunities to grab leveraging from data generated from social media.
Targeted Display Advertising
As the advertisers running display ads shift from print media to digital media, they have higher opportunities to both track and target results, thanks to the new data-driven strategies. The targeting is executed primarily by assigning a cookie to a web browser, and then interpreting that browser’s behaviour as the person using it moves around the Internet. This person, otherwise anonymous, person’s online behaviour can also be improved with offline data about the individual through a process commonly known as data on-boarding.
Data on-boarding enables marketers to effectively target their display ads based on both offline and online behaviour like or the marketer’s own CRM systems, supermarket scanner data, as well as the demographic characteristics. They can also develop lookalike models for broader digital display ads reach. Conveniently, on-boarded data also enables you to track multi-channel campaigns effectively.
As digital marketing grows, the data providers are continuing to upload new data about individual customers, stripped of the personally identifiable information (PII), to effectively enhance the value of the medium to marketers.
One of the best examples of data-driven digital advertising is retargeting, where customers who have shown some level of interest in their brand and product or service are delivered follow-up display ads to encourage them to complete registration or purchase. Conversion and click-through rates are dramatically higher for retargeting than those from general digital targeting.
Where is data-driven marketing headed?
Better and more is the likely scenario. Data sources will continue to expand as the organizations find that their data exhaust can be put to good use in marketing applications. And the marketers will continue to find new means to improve their customer data and also to keep them clean.
Meanwhile, companies are also increasing their understanding of the value of prospect and customer data, investing more resources and time in managing and assembling the data, and also finding data scientists to extract the most value from it.
But the challenges remain. One of them is creating a coherent “single view” of the customer. And despite all the challenges, it’s certainly encouraging that the era of digital marketing, big data, and AI has ushered a new appreciation among organizations about the strategic value of customer data.
