The Attribute Optimizer helps you produce accurate studies, whether you are a Boolean expert or not. Affinio uses Boolean logic to communicate your search criteria (keywords or phrases) with the data source.
For this walk-through, let’s say you want to study the topic of trout fishing.
1. Start by selecting a language.
Choose which language you would like sample posts to be in. Here, we'll stick with the default.
2. Enter a word or phrase.
If there are two (or more) keywords or phrases that you’d like to base your query on, you can include multiple words in quotations separated by OR. If you are familiar with Boolean, other commands will not work in this step. Let’s consider a few different approaches.
What if I input the phrase “trout fishing”? This phrase may be too specific to render the proper search results. For example, you may want your study to include people who say things like “caught a trout” or “fishing for trout” and so on.
What if I input the words “trout” OR “fishing”? This entry would produce a study of people who are talking about either trout or fishing. If you really want to drill down on trout fishing, this input may render broader results than you’re looking for.
Let’s start with just “trout.” Your study will include anyone who uses the word “trout” in a post.
3. Scan a series of anonymous, sample posts.
This process will help you find terms related to your query that may be out of context (e.g. homonyms, profanity, etc.). Keep in mind, these sample posts are not reflective of any filters you may apply to your audience, such as location.
From the screenshot below, you’ll see the professional baseball player’s name, Mike Trout, appear in the sample posts. Next, we’ll show you how to filter this type of result out of your query.
4. For ambiguous topics, upvote and downvote keywords. (optional)
When you upvote a word, it means you want it to be included in the query along with your primary topic (i.e. like the Boolean AND command). When you downvote a word, it means you want that word to be excluded from your query (i.e. like the Boolean NOT command). As pictured below, there are three ways you can do this: highlight a word from a sample post, enter a keyword (or phrase) using quotations, or choose from the list of common, related terms.
Keep in mind that this step is optional. You don't need to upvote/downvote anything for a search to work. You could run the report on only the initial "trout," and bypass this 'optimization' step altogether. We intentionally chose an ambiguous topic for the walkthrough to show you how this tool can be used. Next, we’ll discuss how you can refine your query using the slider scale. Again, follow this step only if you do want to add/remove terms.
For this search, we will downvote “Mike” and upvote “fish” and “caught.”
5. Fine tune your query. (optional)
As you’ll see from the screenshot below, when you upvote and downvote words, more related terms will autopopulate in these categories. For inclusion in the study, a post will not need to include all upvoted words simultaneously. But it will need to include your original topic plus at least one of those words at the same time. You’ll want to be careful not to overcomplicate some reports (i.e. narrow them down too much). If a search doesn’t find enough authors, it will fail.
You can drag and drop any unwanted suggested words out of these categories with your cursor, or adjust the topic similarity with the slider scale. If you want to include (or exclude) only the terms you have specified, adjust this scale to 100% by sliding it to the far right. If we adjust to 100% in this scenario, it means that posts that contain “trout AND fish,” “trout AND caught,” and “NOT Mike” will be included in your query.
If you are comfortable with Boolean logic, you can also double-check the Audience Summary as it appears within the final steps of report creation.
6. Click Continue when you are happy with what you see in your Audience Summary.
