Marketing as a discipline is difficult to perfect. In fact, most of it is viewed as an art. Not only are consumer behavior and preferences constantly changing, but results and impact are often impossible to measure with accuracy. Until recently, marketers would simply try various activities and tactics and watch for what sticks or seems to generate interest from the target audience — an imprecise practice that leads to unpredictable results, if any.
And it’s not only marketing’s impact on customers that’s hard to measure; marketing teams also struggle to prove their value to upper management by linking their activities to business outcomes and revenue. Most marketers lack the analytical skills internally that are necessary for measuring marketing’s value in a quantitative way.
Yet budget decisions are increasingly data-driven. Campaigns and advertising are expensive, and without insight into whether they’re generating qualified sales leads or not, knowing if you’re getting a return on the investment is impossible. The ability to attribute qualified and converted sales leads — and even revenue — to specific marketing activities is becoming essential for a marketing team’s continued success and the ability to know which campaigns to repeat.
Artificial Intelligence (AI) has come a long way since John McCarthy coined the term in 1956 and is now being used to predict outcomes and recommend improvements to processes and activities across a number of industries and disciplines. And although AI’s application in the field of marketing may not seem obvious, machine learning and even deep learning techniques are helping marketers save time and target audiences more efficiently and effectively than ever.
With the help of skilled data scientists, AI has great potential to improve a broader spectrum of marketing activities. In fact, AI based advanced analytics can be applied to refine and improve marketing efforts in six specific areas:
Propensity to buy
Branding and CX
In this article, we’ll look at each application in detail and provide concrete examples of how marketers can achieve better outcomes by leveraging AI techniques in each of these areas.
1. Campaign and Advertising Optimization
AI can be used to optimize promotional campaigns, cross-sell and up-sell efforts and eliminating the guesswork in identifying and segmenting prospects:
Visitor Behavior Modeling: ML algorithms analyze the behavior of website visitors, allowing for real-time campaign optimizations that target audiences most likely to convert.
Content Optimization: ML algorithms can help you find new ways to optimize content in terms of both layout and copywriting, to be more relevant to the specific recipients and groups of recipients.
Campaign Effectiveness Analytics: AI can help identify which customer segments you should include or exclude from campaigns and, consequently, which campaigns to invest in more and where to scale back.
Targeted Ad Campaigns: AI tools can analyze and identify what types of consumers follow which social media accounts and display advertisements or promotions to those audience segments on the channels they frequent.
AI in Action A marketing team in a mid-stage B2B startup uses a clustering algorithmto identify certain segments of its audience that are reading blog posts about a specific topic. They can use this information to personalize email communications with content offers that pertain to a prospect’s specific interests, therefore increasing the likelihood of conversion.
2. Campaign Personalization
AI significantly reduces the time to create, deploy, test and optimize personalized campaigns at scale. Using ML algorithms, marketers can analyze millions of data points about the consumer to determine the best times to contact them, how often to contact them, and what content is most relevant, given their interests, preferences and behavior. AI can surface which subject lines are most effective in encouraging click-throughs, how long emails should be for various audience segments, and what types of content work best in different parts of the buyer’s journey.
ML algorithms can help personalize the experience a visitor has on a company’s website and enable push notifications specific to individual users. Using insights provided by machine learning, marketing teams can display the most relevant content to individual visitors, based on how they have interacted with a company in the past.
AI in Action A large retail chain with a strong online presence uses a recommendation engine to surface other products a consumer might be interested in, based on previous purchases or recent searches. Recommendation engines also provide options for upsell and cross-sell purchases, such as pairing light bulbs with lamps, or car chargers and cases with cell phones.
3. Predictions: Propensity to Buy
In B2B domain, we make only dozens or 100s of sales, and they really matter. Being able to better marshal our sales reps to focus on those most likely to buy is critical. But how do you determine which leads in the funnel are ripe for a sale? Most sales organizations rely on intuition and/or cherry-picked hypothesis — which isn’t always that reliable.
Propensity to buy models take into consideration all of the data points and apply ML techniques to the dataset to determine which leads are most likely purchase, helping the sales team focus their time and efforts.
For example, if a company has bought something recently or frequently, they will score higher than a company who has only browsed products but not purchased anything in the last six months. Firmographics and response to marketing campaigns are also key indicators of propensity to buy. In fact, there are hundreds of data points that could help predict a purchase. In the absence of ML models, companies can manually scroll through the data and apply human logic, but this is tedious, time consuming and highly error prone. Applying ML techniques take out the guesswork, producing fast, reliable responses, and becomes even more predictive over time, ultimately resulting in confident decision making.
AI in Action A large networking company uses propensity modeling to understand which leads are the most likely to buy a specific product. Such “lead scoring” helps sales staff target their efforts more efficiently, saving the company money and resources, while increasing conversions.
4. Predictions: Churn Prevention
Imagine you’ve had high engagement on your company’s Facebook page for the past eight months and suddenly, engagement drops. You’ve continued to perform the same activities and post the same types of posts, so it’s not clear why people are less interested. AI-powered churn prediction helps to analyze the reasons customer engagement has dropped across a specific channel or multiple channels, so you can take action to correct the problem.
Once you’ve determined the reason engagement is down and churn is up, AI can be used to trigger relevant offers and drip campaigns to keep prospects and customers engaged. It can identify behavioral trends that eventually may lead to churn, allowing you to take corrective action before losing the customer.
AI in Action A startup ramping up business development efforts uses AI-powered marketing spend analysis to determine the most successful acquisition and retention methods. This type of analysis looks at how current customers were acquired, and what activities on behalf of the sales or customer service team helps retain them over time.
A digital transaction can result from numerous marketing touchpoints. A buyer sees an ad on Facebook and clicks through to read a white paper. After reading the white paper, she might visit the website and download an ebook. She might receive an automated email thanking her for downloading an asset. A sales rep might call to follow up and ask to schedule a demo. The list goes on and could include any number of touchpoints and combinations of touchpoints before the buyer makes a decision to buy. How do you determine which touchpoint had the most influence? How can marketing take credit for a lead, and determine what activities work the best toward conversion?
It’s common for marketers to credit a lead or conversion to the final touchpoint before the sale — last touch attribution. But this isn’t the whole picture. Before that last touch, a customer may have had numerous interactions with your brand and content. Without accurate attribution, you don’t know which marketing activities are most impactful and deserving of additional marketing spend — or which ones are a waste of precious budget.
Statistical methods of lead scoring and attribution don’t model complex consumer behavior accurately, but AI models do. They can be used to analyze the impact of various touchpoints and help you understand which activities — or what combinations of activities — are most likely to result in a conversion. Having this information helps you justify budget requests and more effectively measure and report on KPIs that helps move your business impactfully forward. Additionally, and importantly, knowing that certain marketing activities — not actions of the sales team — resulted in a purchase helps marketing prove its value to upper management.
6. Branding and CX
About 80% of enterprise data is unstructured, text-based information that is not organized in any meaningful way. Much of this comes from external sources, such as reviews, surveys and social media. Information from these sources holds the keys to improving customer service and elevating brand sentiment.
Applying NLP and ML algorithms to unstructured data helps brands understand the strengths and weaknesses of their products and services. It provides the information needed to take action to improve upon shortcomings and standardize on what’s working well. Doing so initiates a virtuous cycle of CX improvement that continually boosts reputation and brand sentiment.
AI in Action A popular consumer electronics vendor uses natural language processing algorithms to assess which terms about its business are appearing in online reviews and social media posts. Analyzing this data enables them to spot service issues at each physical location, so they can work with managers and staff to resolve problems, and train them for greater success.
One AI Solution Doesn’t Fit All
Most marketing organizations use some type of marketing automation tool for specific tasks: campaign management, advertising or lead tracking. These automated plug and play marketing automation platforms are now even offering “AI”, and they can be good for one specific task or application, but you can’t ask questions of the data that are outside the platform’s purpose. And because every business’ data is custom, wouldn’t this require custom analytics?