The importance of custom click curves

In this article, I explain how to build custom click curves for better traffic predictions and outlier detection.

The importance of custom click curves

As the design of Google search results pages (SERPs) changes, so does the flow of attention. When Google starts showing a shiny big video carousel for a keyword, for example, fewer clicks go to regular search results.

SERP Features have a big impact on organic traffic and they change all the time, making it harder for SEOs to prioritize keywords and accurately project traffic. It’s vital to consider all variables when comes to click distributions in the search results and build different models for yourself.

You can find CTR underperformers with Ryte or use click curves from AWR. But why should you build your own click curve instead of just taking generic values from a robust study or tool? Even further, why should you build several click curves for your site while one might do it? For three reasons!

One, custom click curves help you find CTR outliers - keywords with a higher or lower click rate than the average. You prioritize improving rank for these keywords or, if they already rank #1, see if you can drive their CTR even higher.

Second, custom click curves help you project traffic more accurately. Opportunity sizes decide what gets prioritized and what gets shipped. The Google SERPs are noisier and messier than ever before and differences in CTR when certain SERP Features are present are massive. It’s easy to make inaccurate predictions for traffic and conversions in this environment but, in order to build trust with clients or other teams, accurate traffic predictions are vital.

Third, custom click curves can help you prioritize work better. If you get to a point at which you can compare curves for different circumstances, you can improve the rate of good decision-making. Say, you compare click curves for SERPs with a featured snippet and SERPs without. You realize the difference in CTR for the top result is huge (no surprise) and prioritize going after keywords with featured snippets. That’s a good decision. No, not every case is as clear as Featured Snippets, plus there are many combinations of SERP features. Custom click curves allow you to compare them.

How to create your own click curve

To create a custom click curve, all you need is Google Sheets (or Excel) and the Search Analytics for Sheets extension.

Step 1: Export rank and CTR for all of your keywords from Google Search Console with the extension.

For the filter:

  1. Take data from a full month without major seasonality (avoid November, December, and January)
  2. Group by Query
  3. Filter for country = your target country, and device = desktop

Step 2: In Google Sheets

  1. Remove all queries that got less than 5 clicks
  2. Remove all queries with position > 10
  3. Remove all queries with impressions < 50
  4. Remove brand queries
  5. Create a new column next to the “position” column, call it “rounded position”, and apply the “=ROUND()” function to the cell next to it in the “position” column. The goal is to round the position so we can create a pivot table.
  6. Create a pivot table (rows = rounded position, value = CTR), make sure to summarize the value as MEDIAN

Step 3: After you filtered and pivoted your data, you get the median CTR for each top 10 position. I prefer to take the median for CTR instead of the average to not give outliers too much weight.

You can now use the curve for traffic projections or outliers detection. For the latter, simply sort your list of keywords by position and compare with your median click curve. Alternatively, create a filter and set a condition that shows you only rows with CTRs higher or lower than your median for the position for which you want to find outliers. Say, for example, your click curve found that the median CTR for position 1 is 17% and you want to find keywords that perform worse. Now, go back to your initial data pull and filter the CTR column by a condition of “less than 17%”.

Custom click curve variations

As an avid reader, you know that it’s important to track SERP Features of all sorts since they distort click rates. Also, mobile and desktop CTRs are different. Query length matters. You need to be on top of these dimensions. Time for some click curve variations!

Desktop vs. mobile

Most SEOs look at combined data from GSC, which is a trap because desktop and mobile ranks and CTRs can vary a lot! To compare your mobile and desktop click curve, all you need to do is also pull mobile GSC data from the Search Analytics for Sheets extension as well and compare it with desktop.

The site in the screenshot below gets fewer clicks for position #1 but more for #2 on desktop. For mobile, it’s much more important to win position one. That’s an important insight, especially since this site gets a ton of mobile traffic!

Not comparing mobile and desktop click curves and matching that with the percentage of traffic from each device increases the chance of missing forecasts and faulty projections.

SERP feature click curves

To build custom click curves that take SERP Features into account, use Semrush (under Organic Research) or Ahrefs (under Organic Keywords) to export SERP features for your keywords.

Before exporting from one of the two tools, make sure to filter out ranks with position > 10 and keywords with less search volume than 100. Select the same time range you picked for your GSC export and filter out brand keywords.

After you imported the data into Sheets, apply “Split Text to columns” (under “Data”) to the column that holds SERP features. Create a VLOOKUP that ports CTR by keyword to the SERP Feature export. Be sure to stay consistent with desktop and mobile data, since Semrush has SERP Features by both.

You can now rebuild your click curve with SERP Features in mind by creating a pivot table and using position for rows, primary SERP feature for columns, and CTR for values. Since there are many combinations, I suggest building click curves by primary SERP Feature (usually shown in the first column of SERP Features).

You might not get enough data for every position and that’s okay. You can still look at comparisons. If you have a lot of data, on the other hand, you might be able to split SERP Feature click curves into desktop and mobile.

The challenge: data

You can build click curves for many variations: by page type, subdirectory, or user intent. The biggest challenge here is data since you need enough clicks and stable ranks to get a reliable CTR. You get the best SERP Feature tracking data when you create a specific project for your keywords in your 3rd party rank tracker of choice. Many sites are too small and have too little traffic to build robust custom click curves compared to larger sites (anecdotally, I’d say the site needs at least 100 keywords ranking on position 1).

The value of custom click curves is undisputed, especially when repeating the exercise regularly or automating the process. You could build a new click curve every day when judging by the change rate in the SERPs but once a quarter or every 6 months seems like a realistic cadence.