High-Level Strategy
Finding the right starting point
In understanding why I took the approach that I did with Causal, there are two main concepts to grasp.
The concept of a target audience is likely very familiar to anyone with a marketing background, but it's worth reiterating. A big part of executing successful AI content campaigns is understanding what exactly your target audience is searching.
Keyword research tools (ahrefs, Semrush, and the like) will do some of this work for you. They'll give you a broad idea of how people search for things, and how many of them are searching for those things.
That said, keyword research tools will tend to direct you towards the highest volume search terms, rather than the ones that are most valuable to your brand or site. The trick here is to understand at a deeper level what your ideal customer is searching, in the moment where they need your product or brand.
Let's consider an example of how this applies to Causal. A big USP of Causal is that the product is able to integrate with a range of common accounting platforms, allowing users to calculate complex financial metrics on top of their accounting data. We knew that someone looking to calculate specific metrics from their accounting data would therefore be a good fit for Causal, and be likely to convert.
I therefore came up with the idea of building a page structure of:
How to Calculate {metric} in {platform}
Where {metric} is a variable that stands for a list of possible financial metrics (MRR, EBITDA, etc.) and {platform} is a variable that stands for all the accounting platforms that Causal integrates with (Quickbooks, Xero, etc.). We then used Byword to build out pages around each possible combination of metric and platform, and layered some specific CTAs on top to convert users to Causal.
I go into this campaign in more detail later on (you can skip ahead here if you like!), I just wanted to flag it early on as an example of building page structures around your ideal customer's searches, or the specific problem that your site/product solves.
This is a particularly interesting example in my eyes, because the large majority of search terms that we were targeting didn't even appear in keyword research tools, as they're too low-volume. This might put you off building these pages by hand, but this isn't a problem when you're generating thousands of these pages in an automated, cost-effective way. By doing this, you're able to target a huge number of high-intent, low-competition search terms, with minimal human resource required.
There are two broad ways of building out an AI-generated campaign: search-based campaigns, and structured campaigns.
A structured campaign is any sort of campaign where we decide on a specific keyword or page title structure, containing one or more variables, and then build out all possible iterations of that structure. If you've read the page on N-Dimensional SEO, you'll recognise this as corresponding to 1- and 2-dimensional SEO.
The example we looked at in the section above is a Structured campaign. We essentially compiled a big list of metrics, a big list of accounting platforms, and generated all possible combinations of each list in a specific structure (How to Calculate {metric} in {platform}).
| {platform = quickbooks} | {platform = xero} |
{metric = MRR} | How To Calculate MRR in Quickbooks | How To Calculate MRR in Xero |
{metric = EBITDA} | How To Calculate EBITDA in Quickbooks | How To Calculate EBITDA in Xero |
Structured campaigns are generally good when we want to build out content around some finite set of things (like metrics, and platforms). It lets us build out one page for each possible search term that we want to go after, and completely cover a topic in our SEO.
Most of Causal's SEO campaigns were structured, as we'll see in the next section, but some of them were slightly different.
The other main type of campaign that I'd recommend considering when it comes to AI-generated content is a search-based campaign.
These campaigns don't take some finite list of objects as their inputs, instead they take search terms. In Causal's case for example, we executed two main search-based campaigns:
- /how-to/, covered here. In this campaign, we used ahrefs to find the highest-volume search terms that contained the phrase 'how to' and then either 'excel' or 'google sheets'. We then produced content on over a thousand of these, answering the question posed by the search term.
- /excel-shortcuts/, where we again used ahrefs to find the highest-volume search terms containin the words 'excel' and 'shortcut', and used Byword to produce content around the search term.
This type of campaign is actually the motivation behind Byword's Keyword Mode. Keyword mode takes a keyword (/search term) as input, and generates its own title based on that keyword, making it quick and easy to generate search-based campaigns.
The benefits of search-based campaigns are that:
- Once you have a rough idea of what sort of search terms you want to go after, research tools like ahrefs make it quick to find all the search terms you want to go after.
- They're pre-validated, in the sense that you're going after search terms precisely because they have volume. The downside of this, of course, is that they typically go after more difficult/higher competition search terms.
Now that we've understood these concepts, you'll be able to understand why we built out the AI content campaigns that we did, and the thinking behind them.
Last modified 1mo ago