In today’s digital world, filled with tons of data points and metrics, understanding the “why” in data can be very difficult — especially when trying to narrow down where to start. However, you can keep things simple by developing an analytical approach. If you envision two major reasons for change (rate vs. mix), you can use data to inform where you should start by limiting your options as opposed to looking at everything simultaneously.
Change types
A mix change is driven by sheer volume, holding the average of rates as a constant. Potential drivers of a mix change are:
- Increasing/decreasing budgets
- Adding/removing negatives/keywords
- Turning on/off awareness-driving campaigns
- Seasonality (depending on your business/industry, this can also be a rate change driver)
A rate change is driven by a change in customer behavior taking the average of measures as a constant. Potential drivers of a rate change are:
- Ad copy changes
- Better/worse user interaction on the website
- Sales (Depending on your business/industry, these can also be a mix change driver)
- Increasing/decreasing position
These are not the only reasons for mix and rate changes, and in some instances a mix change could have rate change implications and vice versa.
Example 1) 100% mix change
Last month you received 1,000 impressions. This month you received 500 impressions. The click-through rate (CTR) was 10 percent for both months. After calculating it, you get 100 clicks the first month and 50 clicks the second month for a total difference of 50 clicks. To determine which drove the most change, you can use these formulas:
Rate Change
((Measure[old]+Measure[new]))/2 * (Rate[new]-Rate[old])
Mix Change
((Rate[old]+Rate[new]))/2 * (Measure[new]-Measure[old])
Now, plug in the numbers:
Rate Change
((1000+500))/2 * (10%-10%) = ((1500))/2 * (0)
Result: 0 Clicks
Mix Change
((10%+10%))/2 * (500-1000) = ((20%))/2 * (-500)
Result: -50 Clicks
In this example, you see that 100% of the performance change is due to a change in impressions. Knowing this, there are some items we can eliminate as a potential cause for the change since they only affect rate — like ad copy or position changes (in a general sense). We can then dive into what might cause an increase or decrease in impressions, like increasing/decreasing budget, adding keyword negatives, and turning off awareness-driving campaigns.
Example 2) 100% rate change
Over two months, you drove 100 visitors to your website. While you converted 25 last month, you converted 50 this month (a 25% conversion rate vs. 50%):
Rate Change
((100+100))/2 * (50%-25%) = ((200))/2 * (25%)
Result: +25 Conv.
Mix Change
((25%+50%))/2 * (100-100) = ((75%))/2 * (0)
Result: 0 Conv.
What drove the 25 extra conversions? Given this example, there wasn’t a change in mix (i.e., more visitors to the site) but we’d look to see if there was a fix/change made to the site to drive conversions or if a sale was driving conversions.
Example 3) Mix and rate change in action
You ended the previous month with a CTR of 6.7% and the current month with a CTR of 5.03%, which makes for a 25% decrease in CTR. You had 525M impressions in month 1 and 850M impressions in month 2. The calculations show 35,175 for month 1, 42,713 for month 2, and that you actually netted 7,538 more clicks. Let’s work out the numbers and see what drove the change:
Rate Change
((525,000+850,000))/2 * (5.03%-6.70%)
Result: -11,516 Clicks
Mix Change
((6.70%+5.03%))/2 * (850,000-525,000)
Result: +19,053 Clicks
Absolute Change
Abs(19,053) + Abs(-11,516)
Result: +30,568 Clicks
Net Change
19,053 + -11,516
Result: +7,538
These calculations show that the majority of the change was driven by mix, which could be associated with a new branding initiative or by pushing more content type strategies. Despite your CTR dropping by 25%, you see a 21.43% increase in clicks due to the sheer volume of change compensating for the drop in CTR.
In the example above, some may be concerned with the overall drop in CTR. However, it may not be a problem at all but a lack of understanding of what is going on “under the hood.” Therefore, leveraging the data and using this approach can provide insight on where to dig in. When you enter a “fork in the road” during the problem-solving process, being able to remove one of the paths is a big help.