Best Practices for Selecting and Adding Keywords

To get the best results from 6sense, you need to select the right keywords and add them to 6sense. (Keywords are often phrases, not singular words.) It’s critical, but it may seem daunting to sift through all the potential details that might give you the best results. Here are some tips and guidelines about how to get started, how to refine your keywords, what to avoid, and understanding how our data matching and normalization provides results.

Start From Your Existing SEM/SEO Keywords

SEO/SEM keywords are different from intent data, but you can use them as an excellent baseline for a starting intent strategy. You want to avoid top-of-funnel keywords and focus on mid-funnel and lower keywords.

Start with a list of top-performing thematic, product-specific, and competitive keywords related to a specific product category. These are the keywords that drive the highest conversion rates and traffic to your site. They are often your top-performing SEM keywords. Copy and paste all the SEO/SEM keywords that fit this criterion into 6sense.

Add Keywords Using the 6sense Recommended Keywords

Recommended keywords can help you expand your reach. Our keyword recommendations are based on words that have a high degree of similarity to the ones you have already selected. They help you find terms that you might have overlooked before.

When evaluating keyword recommendations, use your discretion and knowledge of your own products to decide which recommendations to include in your keywords. Pay careful attention to words with multiple meanings. You may want to think twice about using them because they can add noise to your data.

Favor Product-Specific and Niche Keywords

Product-specific and niche keywords are stronger than more general high traffic search terms. The more specific a keyword is to your product or business, the more likely that a customer on a webpage related to that word is interested in your solution. A keyword like “artificial intelligence selling tools” is a much better indicator of intent for 6sense than something like “AI.” The more specific a keyword is the lower the volume of matches you will see for that keyword. However, those matches may be better matches for intent. You need to walk a careful line to get good volume without introducing too much noise.

Add Competitor Names and Products

Competitor names and products make excellent keywords. Even better, try combining a competitor name with its product—for example, “6sense Account-Based Marketing Platform.”

Be careful about adding competitor name keywords that are general in meaning or have lots of meanings. For example, a name like “Lush” could be the name of a store or a word used in descriptions of pictures of beautiful flowers.

Update Your Keywords Regularly

In first phase your rollout of 6sense, you do not need to identify and include every single keyword that is important. However, you don’t want your keywords to go stale, either. Review and tweak them regularly. The market and your products are constantly changing. If you do not update your keywords as they do, then your results will suffer.

What to Avoid When Adding Keywords

Here are some tips to help you avoid common pitfalls:

• Avoid using overly generic or very commonly used words. These types of keywords add noise to your keyword data. A word like “password” appears in a lot of contexts that may be irrelevant to your use case. “Password manager” is more specific and has significantly less noise. Using a keyword like “password management software” identifies the solution you provide and is a much better intent signal.
• Avoid words with multiple meanings. These types of words also add keyword data noise. A word like “Apple” could refer to Apple the company or apple the fruit. Using instead a keyword like “Apple MacBook” or “Apple iPhone” better indicates intent for a specific product.
• Don’t use words that only serve a grammatical function. For instance, the English articles “the”, “a”, and “an” should be omitted in keywords. Also, avoid phrases like “my server rack overheats” and instead favor keywords like “server rack overheats.”
• Try to use words that are longer than five-letters long. Short keywords (under five characters) have more noise than longer words. A two-character word like “Q2” not only has a wide variety of meanings in several contexts, but also comes up randomly in things like URLs. We have several filtering steps that attempt to sort out this noise, but the shorter the word, the more likely that it increases data noise.
• Don’t use acronyms and initialisms. Acronyms are one of the worst offenders. They are almost always poor intent signals, can have multiple meanings, and are frequently under five characters in length. Even a common acronym like “CIA” that most people associate with the American Central Intelligence Agency is also used to represent several other organizations, like the famous Culinary Institute of America.
• Don’t use special characters like #, @, and $. Many of the natural language processing algorithms that we use do not support special characters. Because of that, they don’t produce matches, so we strip them out during normalization. Note: We do process accented characters (é, á, ñ) and two-byte characters (語,国,언).
• Don’t use high volume as your main indicator of a good keyword. Higher volumes are more likely an indicator of noise in a keyword rather than an indicator of strong intent. The best indicator of success is the opportunities generated for your business. That is the metric you should follow when trying to understand keyword performance.

Matching and Normalization of Keywords

Matching: For multiple-word keywords, our matching logic uses arrays to match back intent signals allowing for much more comprehensive matching. With this type of matching logic, word order does not matter. We chose this method because the methods we use to extract keywords from events don’t necessarily extract keywords in the most human-readable order. When extracting keywords from a webpage, what we care about is the likelihood that the page is relevant to that word, not the order in which those words appear.

For example:

• 6sense ABM > ["6sense", "abm"]

This matches both “6sense ABm” and “ABM 6sense.”

• Artificial?Intelligence > ["artificial", "intelligence"]

This matches both “artificial intelligence” and “intelligence artificial.”

• 6sense Revenue AI! > ["6sense", "Revenue", "ai"]

This matches both any combination and any order, so “AI drives Revenue at 6sense” would match.

Normalization: During normalization, we strip all special characters, spaces, and capitalization from a keyword. We do this because these factors cause problems with many machine learning algorithms. By eliminating them, we can use several extraction and natural language processing techniques that we would not be able to otherwise.

The keywords in the following example all normalize to the same thing and return the exact same number of events and accounts:

• 6-sense -> 6sense
• 6sense -> 6sense
• 6sense! -> 6sense
• 6 sense -> 6sense
• 6SeNSe -> 6sense

These keywords all become “6sense” after normalization.

We do not capture misspellings in normalization. If your business name, product name, or key concepts are commonly misspelled, list out those misspellings as separate keywords.

• 6cents -> 6cents
• 6-senses -> 6senses
• Six-sense -> sixsense

Historical Activity and Adding or Deactivating Keywords

When you first add keywords, the first batch of keywords show the data from seven days of historical activity, and each time you add a new keyword, the 6sense platform provides a seven-day history of that keyword.

If you deactivate a keyword, data stops being collected for that keyword. The impact of keyword deactivation varies for different features that rely on that keyword data. For details of these implications, review Keyword Administration.