Why Data Matters for AI-powered Marketing
Artificial intelligence (AI) offers marketers almost an infinite ability to interpret loads of customer data into meaningful and actionable insights, which creates tremendous value for business. But when it comes to the point, fewer CMOs really understand what AI is and how it uses data and why.
This introduction article explains in simple terms what data you need to get started with AI in your executive marketing strategy.
Where to Best Implement AI in Digital Marketing
Where are the biggest opportunities to implement artificial intelligence in digital marketing? First of all, tasks that can be intelligently optimized and automated are best to delegate to an AI-enabled system.
Here are a few more places where artificial intelligence can help CMOs and digital marketers do more with less effort:
- Manage content at most stages stages, for example – from research and drafting to scheduling and publishing.
- Collect, moderate and reply to user comments and feedback
- Test headlines and ads
- Optimize website/app content and social media updates
- Review analytics, provide insights, write reports
- Predict customer behaviors for ad optimization (lower cost per ad, higher conversion rates)
- Create personalized user experiences on landing pages
AI is about Data
Data powers all artificial intelligence solutions. Before you even start to look for possible AI solution providers for your business, first figure out what data you can collect, access and use. Any AI marketing system requires datasets to work with.
For example, recommender engine goes through thousands of data rows and columns to predict what products or content people will buy or like on a fashion website. Image recognition in social apps analyzes picture datasets to identify your face in photos. Natural language processing (NLP) recognizes voice and typing (words, phrases and tone of a conversion) to provide a solution or answer in return (e.g. Apple Siri, Amazon Alexa, Google Assistant). See AI 101 to learn more.
Let’s see what types of data exist:
- In-house
- 3rd party
- Structured
- Unstructured
In-house and 3rd Party
Brands use AI to work with data. If you don’t have data or it’s insufficient, AI-powered marketing solutions can be useless for you unless they use their own datasets. So, start with estimating what kind of data you have – whether it is in-house or third-party generated.
For example, if you’re going to use an AI-enabled CRM system (e.g. Gong.io client onboarding system), keep in mind that this solution works best with in-house datasets, such as customers, contacts, voice/video records, orders, transactions, appointments, tasks etc. Another example is a product analytics tracker for analyzing and predicting user behavior in a mobile app (e.g. Kochava.com in-app tracking) or on a website (e.g. Gamalon.com web analytics).
If you don’t have access to in-house data, you need to rule out solutions that require it, and focus on seeking AI-powered systems that use 3rd party data, e.g. Facebook Ads that offers deep targeting capabilities based on third-party user data.
However, keep your marketing campaigns open to in-house and 3rd-party datasets, because plenty of AI solutions can utilize both data types, helping you to achieve your marketing goals.
Start with estimating what and how much data you have, whether it is generated in-house or by a 3rd party, whether you have full access. Next, discuss with AI solution providers how your datasets can best be used with their solutions, what solution can be considered a good fit for your marketing campaign and why.
Structured and Unstructured
Your data – text and multimedia content – can be structured or unstructured.
Structured data is easy to view and manage as it is displayed in columns and rows, like a Google spreadsheet. Data can be structured with help of a programming language (e.g. SQL) that collects, stores, searches, organizes and presents all various piece of data as a single, tightly organized database. Examples are CRM databases, accounting spreadsheets, smartphone address books, others.
Unstructured data is not organized and can be displayed as multiple different files or a single files with endless lines of numbers, words and special symbols that are often impossible for people to read and understand. Examples of unstructured datasets are emails, customer requests from web contact forms, various reports, order receipts, transactions, blog posts, others.
What Stories can You Tell with Your Data?
If you create content, consider the stories your data might be able to tell.
Natural language generation (NLG) is an artificial intelligence technology that turns structured data into narratives. NLG takes numbers and tells stories about them based on rules that humans create.
For example, Bloomberg, The Guardian, the Washington Post, Reuters and other outlets use AI-powered NLG systems to create pieces of news based on data from various sources (ex.: Google Analytics, CRM). Read Hannah Parkinson’s article about the robot apocalypse.
NLG might be a good fit for you too. Start asking yourself: What kind of content (blog articles, statistics reports) would you delegate to AI? Is there a need to scale content for your brand blog? Does it make sense to show your numbers to audience?
What about Robots?
AI has the potential to not help but absolutely overperform human in marketing and beyond.
Look at Atlas, Spot, Handle and other impressive robots built by Boston Dynamics. They are taking their first steps but once they become super-intelligent, fast and adapted, this kind of AI is going to be hard to put it back in the box. Don’t you think we have come a long way from Robots War?
