I had the good fortune to deliver a keynote address at the Search Insider Summit (SIS) in Key Largo, Florida on May 6th. I love this conference because it is composed of a small group of experts who share their best practices and other lessons learned freely. As such, I always feel like I learn more than I share. I highly recommend attending a future SIS conference. They are held in May in Key Largo and in December in Deer Valley, Utah. The venues can’t be beat and your access to experts is unmatched in the world of search marketing conferences.
If you’ve picked up mine and Mike Moran’s new book Outside-In Marketing: Using Big Data to Guide Your Content Marketing, the theme of my talk might be familiar to you. It’s all about learning how to build and optimize owned content properties to intercept more of the audience in their information journeys, and ultimately help them become loyal clients.
The talk fit into several conference agenda themes quite well. I want to focus on three of them in this blog post because it seemed like most of the conversations and questions centered on these three themes:
- Voice search: Queries are becoming more conversational as more and more users as Google, Watson, Siri, Alexa, and Cortana questions. This has huge implications for keyword research and audience targeting.
- Audience targeting: Most of the experts in the room had war stories to tell about integrating client data from their CRM systems with their web analytics data to target the audience with greater and greater precision.
- Attribution modeling: The more touch points the buyer has with your content, the more complicated it gets to give the proper credit to a piece of content and its related tactic. This leads to impoverished investment in content marketing, and other problems.
These three trends are changing the way marketers build systems to engage with clients and prospects. Here I want to briefly lay out the problems and questions related to each. In later posts, I will show how to solve the problems and answer the questions.
Voice search
When we analyze queries, we try to unpack query grammar in addition to the meanings of the words. We do this because queries are ambiguous. For example, try Googling “Cloud Mobile” and “Mobile Cloud.” When you do, you will find that the two queries are quite different, though the words are the same. If you just focus on the words and not the queries, you would create a lot of waste for your content marketing activities, and you would annoy a lot of your audience.
When search was exclusively done on keyboards, searchers made a lot of refinements in their queries before clicking anything in the results. So, word order and proximity (how far apart the nouns in a noun phrase are) were key variables to test in keyword research tools. Now, more than half of all queries are conversational, according to the Bing representatives sponsoring the event. Most of these queries are actual questions asked of the digital assistants, which are powered by search engines. (Hint: Bing powers Siri, Alexa, and Cortana, so start including Bing in your search marketing plans, if you don’t already.) Digital assistants are training searchers to use search differently even when they type their queries.
The good news is, it is easier to parse a query stated in the form of a question (“what is…?” “why should I…?” “how do I…?”). A good way to build a content strategy is to answer all the questions your target audiences ask often enough that it’s worth the effort. The challenge is, voice search places all kinds of additional burdens on content strategists. To answer your audience’s question, you often need to know a lot about the context in which they asked it. For example, the answer to “how do I get to the nearest…?” is dependent on the location of the searcher. How do you answer those questions in real time? I will tackle that question in my next post.
Audience targeting
A lot of queries are generic questions like “what is mobile cloud?” which are relatively easy to answer, because they don’t depend on the context of the user. Because they are conversational, it is also easier to understand audience intent. When someone uses “what is” in a query, they are typically early in their buyer journey, for example.
Suppose you build content of all of the top-of-funnel questions your target audience asks (again, often enough that it’s worth your effort). What do their interactions with this content say about their client journey? How do you build a system that is responsive to their needs in a way that helps to progress them through their client journey on and in their terms? That is the central challenge of modern digital marketing.
It is a question I will leave to future posts (a lot of the answers are in our book). But I will refine it here in a way that only occurred to me during a breakout session in the conference. The main challenge is not strategic. The main challenge is technological, and some of it is legal. To do this right, you need to tie all kinds of data together from all kinds of systems in your company. You need your keyword research platform, your web analytics platform, your CRM platform, your product portfolio database, and some sort of audience persona database all speaking the same language. This is the challenge of big data. It’s as though each data source is in a different language and you need a universal translator.
Most of the participants also had huge issues with their legal departments. Without belaboring the point, retargeting known audience members while not violating any laws (especially in Europe) gives lawyers ulcers. Some of the companies in the round table on this topic spent years convincing their legal departments that you can do this without storing or using any personally identifiable information. I will discuss this challenge in another future post.
Attribution modeling
The other challenge every participant at the conference talked about is how do I attribute the discrete contribution pieces of content had to the revenue they generated. My company (IBM) has the longest and most complicated sales cycle for a lot of its products. But even some of the companies with less than an hour between initial contact and sale had trouble with attribution. One speaker had seven different models, which are like lenses through which to view your campaigns. There is no consensus on what the right model is.
What I promise to write in a future post is a POV on how to think about attribution, so that you can form the right model for your buyer journey, which also passes muster for your executives, especially in finance. Briefly, the right model is based on performance. If an asset in a campaign is always downloaded in a client journey leading to purchase, it should be weighted higher than the others. If an asset is very rarely downloaded, the opposite is true. In short, the model you use should be based on the strength of your assets. Again, I will elaborate on this in a future post.
Obviously, I have my work cut out for me. That’s a good thing. Conferences like this get me out of my comfort zone and into the mix of solving the problems preventing my company from optimizing its digital marketing performance. That can only be good.
