Keywords and Keyword Analysis

Is the traffic share taken out of the total search volume?

No. Volume is the number of times someone searched for a keyword; bear in mind, however, that not all searches lead to a click of a result: users might either refine their search or just find the information they were looking for within a SERP feature. Traffic Share is taken as a % of all traffic, i.e., every time someone clicked on a result following a search. Also, note that Google provides us volume figures, whereas we calculate visits based on our own data.

What is the difference between Traffic Share and Traffic Trends?

  • Traffic Share: Contains a distribution of the traffic share for up to 10 chosen domains within the selected time frame and country. It represents 100% per month, so what you see in the graph is a monthly traffic share distribution.

  • Traffic Trend: This new view scales the whole graph to the maximum traffic point during the analyzed period. This view gives a real sense of the seasonality of keywords and what may have been trending more recently.

Do you capture paid search (PPC) for all types of Google searches? For example, Google Shopping is probably mostly paid search, but on Similarweb, if you filter by Google Shopping, it says it's 100% organic.

When we see traffic from specific types of Google Search, other than regular Google Search (such as Shopping, Image, News, Maps, Video, etc.), we cannot differentiate between paid vs. organic. So while we do see all the traffic, it's currently all shown as "organic."

When do keyword volume and cost per click (CPC) get updated?

Volume and CPC come from Google's Adwords API. Google usually releases the data for the previous month on a two-week delay. So you'll usually see the data refresh around the 14th of the following month.

Why does a keyword’s destination URL change month to month? Is this URL the result of the most clicked on a keyword?

We collect SERP data frequently, so this URL is the most recent landing page we've seen with the highest position, not necessarily the most clicked on.

Why do similar keywords have the exact same CPC?

You might find that similar keywords (plurals, for example, like nike vs. nikes) have the exact CPC. This is in line with how Google shares the data with us. They are known to sometimes aggregate certain keywords, but nobody knows the exact algorithm they use.

How is keyword volume calculated?

We calculate keyword volume by combining data from the Google Ads API with our own datasets, including anonymous searches from our contributors' network. We call this a “hybrid” calculation, as the calculation is now based on several distinct data sources, leading to greater confidence in the result. Check out some of the results and how they address these main problems with Google’s volume metric:

Grouping: In this ‘before’ example, Google creates two groups of similar keywords, and assigns them all the same volume. But with the new hybrid keyword volume calculation, volumes are unique:

Grouping_10-May-2021.png

Coverage: In this example, Similarweb can now detect 70% more low volume keywords driving traffic to cnn.com in the US over the last 28 days:

Coverage_10-May-2021.png

We hope this update makes it even easier for you to identify valuable keywords using Similarweb!

How is it possible that there are more keywords results when filtering by a specific country than by Worldwide?

We have data for the top 10,000 keyword results per month for each in our database (paid & organic combined) for both worldwide and each individual country. In some cases, the keywords that we show for a specific country that we don't show for worldwide had a lower traffic share, and therefore, they made the cut to be displayed for that country, but not worldwide.

Note: If you take away all filters (except for country) and filter by one month, you can see more than 10,000 results. Here's an explanation of why:

  • We show keywords (in addition to the original 10,000 from our database) for which we didn't actually see any traffic to the website. You can recognize these keywords because they will have a 0% traffic share.

  • Also, note that for the 10,000 keywords we do have data for (meaning we detected traffic from them to the website), this will not necessarily translate to exactly 10,000 keywords (with traffic share over 0%). This is because in our database, out of the 10,000 keywords we have data for, some may be duplicates. For example, if the keyword "books" sends traffic to amazon.com, it would appear once in the results. But in our database, it could be listed multiple times because it would be listed once as organic Google, once as PPC Google, once as organic Yahoo, etc. Meaning the keyword "books" would account for 3 out of the 10,000 results in our database but only be shown once.

How is traffic share calculated?

We keep the top 10,000 keywords from the contributor network and 10,000 keywords from scraping for each site and month. The contributor network data has visits per keyword.

When a user queries one month of keyword data, and the site has more than 10,000 keywords from the contributor network for that month, then the keyword data we have is not 100%.

Why am I seeing keywords if the traffic share is 0%?

We complement search traffic per domain with keywords we crawl on the Google Search Engine Results Page (SERP). Even if we don't see traffic from a particular keyword to a domain in our anonymous behavioral data, if the site appears in Google results, we show this data with a traffic share of 0%. This provides our users with more keywords to discover per domain and provides new ideas to optimize their own search strategies.

Why is there a higher percentage of traffic share when summing up the total traffic share when filtering by a word in ‘Keywords’ than for that same word in Keyword Phrases?

In general, based on the methodologies, we would expect the two numbers to be quite close but not perfectly aligned. This is because there's something that can cause the number of Keywords to be higher than the one in Keyword Phrases.

For example, let's say we're analyzing uniqlo.com and looking at:

  • Filtering the keyword results to include all KWs that contain "shirt."

  • Looking at the Keyword Phrase "shirt."

Let's say these are all the keywords that we see for “shirt”:

  • linen shirt

  • uniqlo shirt

  • polo shirt

  • sweatshirt

  • overshirt

Only the first 3 keywords will be part of the Keyword Phrase "shirt" because the logic which creates a Keyword Phrase looks for individual words; it doesn't look for the string of letters within other words.

Was this article helpful?
1 out of 1 found this helpful