Thursday, March 9, 2017

Shaping Your Message Through Keyword Analytics and Data-Driven Content


As promised in the last post, the discussion of Google Analytics continues with discussion of how to acquire and apply data to make your product more successful. As with the previous post, we will look at the obvious bits and then dig a bit deeper to identify a not-so-obvious trick.

In the early days of web analytics, people seized on keywords for search engine optimization, to try and drive more traffic to their web site or application. The value of that is declining for various reasons, but there is a greater value you can obtain from keyword data - understanding your customers better and shaping your message to deliver what they want through data-driven content.

And it's from a place you might not think to look.

In the "Old" Days


When keywords first became a thing on the web, it was primarily with Search Engine Optimization, or SEO. At that time, search engines were simple, compared to today. So, if you could figure out the magic mix of keywords, you could get a lot of traffic.

Then people got wiser and realized that lots of traffic is not good enough. You want to attract the people that actually want your product. Bounces are not valuable. [For more about "bounces" and traffic, see the Google Analytics and Product Intelligence - Web Site Traffic & Page Views]

When "cost per click" advertising really hit its stride, so did a host of ways to track the keywords associated with the clicks. If you can't capture that information, then you can't have paid campaigns for those keywords. Google Analytics did a great job of capturing that data and providing it to would-be advertisers.

Then encryption came along...

Keywords and the "Not Provided" Monster

As we discussed in a previous article (Acquisition Channel Data for Better Targeting and Product Improvement), there is an acquisition channel, the organic channel, that reflects people arriving at your site through a regular search, i.e., not clicking on a paid/sponsored "cost per click" search result. However, most organic search data will show something like the following plot in which the vast majority of traffic comes from keywords that are "not provided".


Why is that? Companies have been increasingly protecting consumer privacy, which is a good thing. Back in 2011, Google made SSL (secure socket layer, a secure transmission protocol) the default for users that are signed into their Google Account and even when they search from the chrome "omnibox"*. This causes the search terms to be passed in an encrypted fashion and not captured, resulting in the keywords not being provided to the analytics tool, hence the "not provided" listing.

* - "omnibox" is the name for the address bar in the Chrome browser, where you would type in a URL or keywords for a search

Trivia: Someone was so incensed by this, that they created a page to track the occurrence over a number of sites.

If the vast majority of keywords are hidden to us, how can we use a data-driven approach to identify the language we should include?

Enter Google's Keyword Planner


Google has a nifty tool called Keyword Planner as part of the Google AdWords offering.

The bad news: Unfortunately, no longer you can you just fire up Keyword Planner and play with it anymore. They force you to create a "campaign" first and since you may have never created a campaign with them, you are likely to create a suboptimal one. If your company uses Google Analytics, you may be able to gain access to that data and it's worth checking because you will get a lot more out of it.

The good news (i.e., the workaround): You can set a tiny budget (say $0.02) when you create that campaign and then immediately pause it. This will allow you to explore the features so that you can make smarter decisions.

Luckily for you, I have done this already!  Here are the links if you want to set up your account and play along at home, but my recommendation is to finish this article first.


To make this more tangible, let's create a scenario.

Scenario: Super Duper Weed Killer


Let's say you have a great new weed killer, Super Duper Weed Killer, that is more effective than the current products, but you need to get the message out there. Let's see what we can learn from keywords to help guide the content you would create.

Note: For accessibility, we will limit ourselves to the tools and data that any Google AdWords user could access without running a paid campaign. Also, this is just a demonstration of concept; we will not explore every nuance (and there are many), in order to keep this article concise.

The first stop is the simplest and perhaps most obvious - a simple search with Google. We fire up Chrome and enter effective weed. We get the following autocomplete and autosuggest results:


So, right off the bat, we have some new ideas for words that people might be using to find this type of product:

  • barrier
  • control

If we add killer we get:



And, not surprisingly, the more words we add, the more specific the results become. Certainly, "lawn" is something to add to the list and maybe ask if we have plans for the markets in the United Kingdom and Canada.

But you get the idea. This is a very simple trick that anyone can do, but it's still a good start. Simple searches can provide good information about your competitors and the market space for your product or service.

Let's Use the Keyword Planner


Now, we break out the keyword planner. There are multiple ways to use this tool, but for the sake of brevity, we will use two - "search for new keywords" and "multiply keyword lists".



When we search for new keywords using weed killer as our base, we get the following:


Note: I did not set up a campaign for this, so the data is fuzzier and only shows ranges, but that is good enough for this discussion.

By scrolling through the various suggestions, we can see the searches that are most numerous and most expensive. The first important thing to understand is that the searches with the most hits are not necessarily the most valuable searches. For example, in the short list shown above, the most expensive search, weed control, is four times as expensive and has an order of magnitude less hits than the two searches with the most hits, crabgrass killer and weed and feed.

That said, we are basically in a brainstorming phase of content creation, so we would be wise to document both the highest value and highest traffic keywords. We see a lot of mention of "crabgrass" and our development team tells us that the product is effective against crabgrass, we add that to our list.

The multiply keyword lists feature "multiplies" the keywords you provide in two lists. So, if you were to provide list 1 with weed and crabgrass and list 2 with killer and control, it would consider all four permutations:

  • weed killer
  • weed control
  • crabgrass killer
  • crabgrass control

The multiply keywords tool is best applied after you have identified lists of promising keywords, as it only considers the permutations of the words you provide. Let's try this out with a few of the words we have identified so far:


If we click on Get search volume, we get the following:




This helps us narrow down which keyword combinations are valuable and which may be non-starters (e.g., crabgrass barrier has no searches nor suggested bid). In this case, the top three phrases that rise to the top are:

  • weed killer
  • crabgrass killer
  • weed control

By iterating through these steps, we can quickly start to build a content workbook with the most valuable keywords for our product. We can then inject those keywords into our content. By using this analytical data to shape our content, we can create:

  • Content our customers are more likely to find
  • A message that resonates more effectively with their priorities

Of course, this is just one piece of the puzzle. It certainly does not replace traditional market research, focus groups, customer interviews, etc., but it's a quick and effective little trick to help you find the best language for your content.

No comments:

Post a Comment