Updated: Sep 15
Why should you care? While Google Adwords is an amazing tool for creating ads, advanced targeting and reaching a very wide audience but when it comes to optimizing your Adwords Campaigns, keywords or ad groups, it misses one of the most important pieces of information about your customers - The Behavioral data of your customers.
The data that helps you to understand why some of your Campaigns have an amazing CTR but customers aren't converting at a rate you expected it to be. It doesn't have the data that help you understand if there is some problems with the landing page or maybe the checkout page on your website.
It doesn't help you to understand , How many visits and days it took a customer to finally convert on your website? and Which other channels in conjunction with your PPC campaigns, your customers are interacting before making the final purchase?
There are lots of amazing data about your customers that is practically sitting in your web analytics tool and not utilizing that data is a big mistake.
In this post, With step by step process, I will show you - How can you to do it by yourself and optimize your ppc campaigns fastly and efficiently.
To download the custom report bundle which we are going to use in this post, just click here and if you can't find any custom report I am using in this post, don't worry I will link them as we move along with the post.
1. Getting some context to your current performance -
Most of the time, when we start optimizing our Adwords campaigns, we jump directly into campaigns, keywords, and ad groups. But it is always a good practice to first understand our current performance then dive deeper into optimization. Why? Because it helps us to see the bigger picture and understand what is good and bad, helps us to find things that are broken or help us to understand if we are doing good or sucking over time. Let me show you with some examples, How Context can bring hidden insights from the data which wasn't obvious before. Let's look at an example -
Here, I am comparing two relevant metrics to each other to get some context. You can see that I have changed the metric Session from the drop-down menu in the left to impressions and compared it with CTR to give some context.
You can see above that the store has a huge spike in impressions on 18th June but on the same day, it has the lowest CTR. And you can also see that as the impressions were getting lower and lower after 26th June, the CTR was started to go up, in fact, it reached the highest point during the entire month.
Life Lesson - you can have millions and millions of impressions for your ads but if it is not relevant and helpful to the customers, then you can't drive them to your store and get them converted. That's why I hate measuring performance based on impressions.
Ok, let's look at one more example.
It is good to understand your own performance but what about your competitors? Are you crushing them or is getting crushed by them? Super important to know. Use the competitive intelligence data to gain more insights.
This is the trend line of total visits for the past 6 months of Flipkart, one of India's biggest e-commerce company. They got the highest traffic in may because they ran a huge discount offer at this time for their 10th year anniversary, then the traffic went down in June. Ok not too much to freak about, it is obvious, customers like discounts.
Now let's give some context with CI data.
Now, look at the above pic, Did you find something interesting? I bet you did. Amazon started their operation in 2013 in India, way late than Flipkart, but they are totally crushing them. The difference isn't one or two million in visits, it's astonishingly double the number of visits Flipkart is getting.
I think, now you understand, why did I told you to look at the bigger picture first, then dive deeper into the data to optimize or creating a marketing strategy. It will force you to become more aggressive and kill the campaigns, keywords, traffic sources that aren't delivering good results and focus on the ones that helping your business to grow.
Btw, I am using SimilarWeb for the above competitor analysis.
If you want to learn more about competitor analysis, please read this post - Competitor analysis of Zara, H&M and Topshop with Google trends and SimilarWeb.
2. Where do your most valuable Customers come from? -
One of the biggest wish marketers or business owners have is that they want to spread their message to the whole world.
But money always becomes a big issue for them. So to target as wide as possible with less money, we need to think more creatively. That's Where the Map Overlay report in Google analytics comes in.
Map overlay report gives you the breakdown of important metrics by Country, Region, City etc. If you want to follow along with the analysis I am doing, then Download this custom report - Map overlay visualization report.
We can see that the United States have the highest number of users. Also, it has the low bounce rate, High Average session duration, revenue and Per session Value.
The second good performing country is Canada with good revenue and Per session value. Australia is good, but not as good as it should be, so does the United Kingdom. We need to do lots of optimizations to improve their performance.
But what's up with India, Japan, Germany, and Spain. Japan has $0.00 revenue from all the users who visited the website from there. OMG!
Quick cheap tip - Click on the comparison tab at the top right corner and change it from user to bounce rate from the drop-down menu of compared to site average.
What is Bounce Rate?
Percentage of all sessions on your site in which users looked at only a single page and leave it right away without taking any action.
In Plain English, "I came, I puke and left".
Bounce rate is a great key performance indicator(KPI), though it will not tell you everything, but it will give you a great starting point in your analysis then you can dive deeper to figure out the real cause of it.
In the above pic, it is totally clear that the United and Canada is doing good and others are doing bad.
Low Bounce Rate - Good
High Bounce rate - Bad
Now we also know why does Australia and United Kingdom have so less per session value because most of the traffic is bouncing at a higher rate.
Now Click on the Per session value column to sort the report from highest to lowest, previously it was sorted by users.
Look at the countries in the green box. These countries were totally out of the picture before. Low traffic but a higher Per session value.
Also, focus on the per session value of Panama, the only sad thing is that the traffic from Panama is very low. I wish we had more users from Panama. Now we know where do we need to focus and take the advantage of the opportunities that lie. Later in the post, I will describe some more amazing techniques that will help you to do it. So keep reading, don't stop :)
Let's focus on the United States because 47% of the users and 95% of the revenue of this website comes from the US.
Just click on the United States to drill-down to the regional level. Now, Look at both the maps.
If you look closely, you can see that although we have the highest number of sessions from California but Per session value is highest in Massachusetts. look at the map to compare the performance of different regions in terms of sessions and per session value.
You can hover your mouse anywhere on the map to see respective values and you can also further drill-down just by clicking anywhere on the map itself.
And when you do the analysis, don't forget to also use some behavior metrics like bounce rate to get the complete picture of performance. Always use Acquisition, Behavior and outcome framework in your analysis.
3. Campaign optimization -
The above report is sorted by RPC. The goal here is to understand where are we making the most amount of money for every click compared to how much each of those click costing us.
Look at the "Electronics" campaign, OMG it has the highest number of impressions among all campaigns but $0.00 in revenue, So pathetic. No wonder, why the impressions were so high but CTR was so low, as we seen earlier in this post. The culprit is found :)
For the Dynamic search ads, the CPC is $0.40 and RPC is $0.50. Are we making any money from this campaign? The total cost of this campaign is $407 and revenue generated is $502 but don't forget that the cost of goods sold isn't added to the cost yet. I don't think we are making any profit from this campaign.
We can't measure the (not-set) in Campaigns because Google analytics didn't receive any information about the visits associated with it.
The only campaign that is performing well is "Accessories".
Let's dig deeper to understand why the Electronics campaign is performing so bad and try to optimize it.
The match type selected for this campaign's keyword is Broad match to show the ads to a wider audience. If I click on the electronics campaign and then the sunglasses brand group, it will show you the keyword report.
Now, look at the above picture. Actually, this website bids on Keyword "women sunglasses" but due to broad match, the keyword is matching with lots of different search queries.
Search queries - are the actual queries that people typed into the search engine.
Almost 8 out of 10 search queries are totally different from the keyword they are bidding on.
Broad match helped them to target wider audiences but customers intent wasn't matching with the keyword
They have nothing to do with "Jimmy Choo" or "polaryte HD sunglasses men and women the dealer.
Use the inline filter to only select the "Aw-Electronics" campaign then drill-down to keywords and then add a secondary dimension "search query" from the drop-down list to get results like I did above.
4. Keyword Optimization -
We gonna apply the same technique that we used before to find out which keywords are bouncing at a higher rate compared to the site average.
It turns out that for the top 20 keywords, the keywords related to Google store, Youtube merchandise, Google products and Google backpacks are bouncing at a higher rate
To optimize the keywords, we are going to use a custom report - PPC keyword/matched query report.
As we saw earlier that the keywords we bid on and the actual search queries can be very different from each other which can lead to higher bounce rate and lower conversions.
Let's see if this is the case-
The match type used for "google store" is an exact match and if you click again you can see the actual query.
As it turns out there isn't much difference.
so what do we now?
Let's dig a little bit deeper to see if the users are landing on the right page or not?
Hmm, it is also not the case. The visitors are landing on the right page, the Google store homepage.
Let's not lose hope, try one more time. Apply the secondary dimension "device category" to the report. Now, look at the below pic.
It looks like people are having some difficulties using the website in mobile and tablet compared to desktop. They are not converting at all in this case.
Let see the next keyword - Youtube merchandise
Same story here. But if you look at the below pic, you can see that people are landing on a wrong page.
Rather than sending the traffic to the youtube product page, GMS is sending the traffic to the homepage.
By looking at some keywords, we find out that people are landing on a wrong page and people are also might have some issues in using the website on mobile devices.
To be more confident in our hypothesis that people are really finding difficulties to use the website on mobile and tablet devices, let's use Advanced segmentation.
As we can see that the visitors are bouncing at a higher rate on mobile and tablet compared to desktop. E-commerce conversion rate and per session value is also low for mobile and tablet traffic.
And if we look at the segmented keyword report by device type, here also you can see that lots of keywords have zero or less conversion and per session value compared to desktop
There are lots of other ways to find the real cause, one way is to look at the shopping and checkout behavior report in the Google Analytic's e-commerce section under conversions. one of the most efficient ways of all is to collect qualitative data. Rather than inferring what is the cause, why not directly ask the customers?
Click here to learn - How to apply and remove segments in Google analytics.
Now the next question is - What to do if you have a large volume of keywords?
It really becomes a big problem when we have a very large volume of keywords, pages etc. Most of the time we only look at the top 10 or 20 rows of data and move on with our life. But the top 10 or 20 rows rarely changes too much. The real opportunities lie when we move beyond the top 20.
So let's see how to actually do it.
A. Use inline filter -
Click on Advanced.
Now select Bounce rate and per session value. In this example, we want to find out keywords which have bounce rate lower than 30% and has per session value greater than $5. Hit apply and boom!
Now we have a list of keywords that are performing very well. Now, we can invest more money on these keywords and find out low performing keywords with the filter and either optimize or completely kill them. Sadly, in the above case, we have only 5 keywords that match our criteria. I wish there is more.
But you see, how much powerful this feature can be to cut through a huge amount of data and get the ones which we are most interested in. You can change or apply more conditions, whatever make sense to you depending on what are you trying to accomplish.
B. Term Cloud -
First, go to Acquisition > Adwords > Keywords
want to see which keywords are generating more revenue?
Here it is. One thing to remember when using the term cloud. It will only show you the total rows of data that is selected at the bottom. So if you want to see more rows of data just change the number of rows from the drop-down menu.
Let's see, the visitors from which keywords actually added products to the cart and started the checkout process?
Now, you have the answer.
Ok, now let's checkout the visitors from which keywords actually completed the purchase?
Now compare both the visualization and find out visitors from which keywords started the checkout process but didn't complete the purchase. Find out what caused them to abandon the purchase? Look at where these people are coming from? what can we do so that this doesn't happen again? Real analysis with amazing insights.
C. Weighted Sort -
What is weighted sort?
Weighted sort applies a smart algorithm to your data to bring the highest and most statistically significant metrics to rise up to the top.
To use weighted sort, click on the bounce rate ( or any other percentage based metrics) column header in the table. Then above the table, use the sort type selection menu to select weighted.
If you do that you will see a list of keywords that Google analytics algorithm think you should focus on. In this case, if you improve the bounce rate of these keywords you will get the most value out of your investment.
wait there is more.
Now reverse sort the table.
Boom! Now you have a list of keywords which already have lower bounce rate and if you focus on these keywords and improve their performance, you can extract as much value as possible. No more beating your head to understand what to focus or not to focus in the ocean of keywords. Google analytics do this heavy lifting for us.
Isn't that great?
One more tip - you can apply weighted sort on almost any report in google analytics like traffic source/medium, landing pages, internal site search etc. want more magic? Keep reading the post :)
5. Ad group optimization -
we learned a lot of ways to optimize our Adwords ads. In this section, we are going to use another great visualization technique to optimize the performance of our Ad group as well as campaigns and keywords. Let's get started.
First, Go to the Acquisition > Adwords > Treemaps
Here we are comparing the number of users (Acquisition metric) of our ad groups to the bounce rate ( Behavior metric) of those ad groups.
Just by looking at the treemap, we can now understand that Merchandise brand[E] has received more users and lower bounce rate compared to other high traffic ad groups like Merchandise-brand [BMM] and store-brand[E]. And if we look at the right bottom of the pic, we can see that some of the ad groups like Swag-brand[BMM] and gear-brand[BMM], though has low traffic but has even lower bounce rate than other ad groups.
Now, let's compare an Acquisition metric to an outcome metric to understand the complete performance of these Ad groups.
In the previous analysis, the bounce rate was very high for the Store-Brand[E] ad group but after applying an outcome metric "Avg. order value", we can see that it has a very good avg. order value. If we can reduce the bounce rate of this ad group, we can make more money and can further increase the Avg. order value of it.
If we have only applied the Acquisition and behavior metrics in our analysis, we would have made a wrong assumption about this ad group. But after applying an outcome metric, we are now able to see the complete performance of this ad group.
As I said before always use the ABO framework in your analysis to get the full picture.
Want to know which keywords are performing better in a particular Ad group?
Ok Google, show us :)
Boom! Now, just by looking at the treemap, we know what is good and what is bad?
Isn't analytics so cool?
You can do the same for Campaigns also.
6. Bidding strategies to maximize the ROI -
Rather than spending all your money equally during the whole day and week, you can get more out of less by spending more when your customers are more likely to convert and less when they aren't. Hour of days report in GA under AdWords section exactly help you do that.
You can see that though people are coming throughout the day, sometimes more or less but most of the purchase is happening during 10 & 11 am, then again during 2 and 3 pm, at 5 pm in the evening and 7 and 8 pm in the night.
And from the day of the week report, it is clear that most customers are buying on Wednesday, Friday, and Saturday.
So why to waste your money when the visitors of your website aren't most likely to convert. Use these two reports to do the smart bidding and save your precious money. But don't forget to check these reports periodically and make changes accordingly.
7. Multi-Channel Funnels for Adwords -
In Google Analytics, conversions and E-commerce transactions are credited to the last campaign, source/medium that referred the user when he or she converted.
But in reality, customers do lots of thing before making a final purchase like researching for a product, price comparison to find products and lower costs, searching for coupons and deals etc. via multiple touch points across multiple channels. By Default, Google analytics will attribute all of the credit or e-commerce revenue to the last marketing channels or source etc. which leads to conversion. This is called Last-Click attribution model. Which is not fair at all. So the question is - How will you decide to give which channels will you give the credits and how much? In the above report, you can see that direct traffic is closing most of the sales and conversions but referral and organic are also contributing to close the sales by introducing new customers to the brand. Multi-channel funnels help you to understand how to better allocate our marketing budget and time wisely.
A. Assisted Conversion report -
Assisted conversion and assisted conversion value - tells you the number of assisted conversions and their monetary value of sales and conversions the channel assisted. The higher these number, the more important the assist role of the channel.
Last Click or Direct conversions and Last click or Direct conversion value - tells you the number of last click or direct conversions and their monetary value of sales and conversions the channel closed or completed. The higher these number, the more important the channel's role in closing sales and conversions.
Assisted/Last Click or Direct Conversion - is the ratio which summarizes a channel's overall role.
A value close to 0, indicates that the channel completed more sales and conversions in the last stage of sales funnel than it assisted.
A value close to 1, indicates that the channel equally assisted and completed sales and conversions.
A value greater than 1, indicates that the channel has a bigger role in assisted sales and conversions.
In the above pic, look at the last column. We can see that the value of Display is 5.50, which is totally amazing. None of the other channels are even close to it. Which means that Display is really really good at assisting sales. You should definitely love and adore this channel.
In Plain English, this is your secret lover, who loves you and care about you but didn't ask for anything in return. Show some love!
The value of Paid search is 0.59 which indicates that it mostly helps in closing more sales than assisting.
So, if you only focus on the default Last click attribution model then you are making a big mistake. You are not giving enough credit to Display for introducing your brand to new people, working hard in the early stage of the conversion process to convince your customers, which finally leads to sales later mostly by some other channels.
Let's understand the multi-channel funnel for our Adwords Ads.
Remember, In our analysis, we made an assumption about the Dynamic search ads Campaigns that though it was generating some revenue but we are worried that we might be losing money because the difference between the cost and the revenue was very low. But now if we look at the assisted conversion report, we can see that this campaign also assisted 10 conversions and added an additional $414 in assisted value to the site. If we only considered the default last click attribution model, we would have been made an error in our analysis.
We can also see which Ad groups and keywords are driving more assisted conversions. We can see that All web pages and Google ad groups have the highest value, driving more assisted conversions. while Store Brand[E], Drinkware brand[BMM] and Gear brand[BMM] all have the ratio of 1 means these ad groups are equally helping in assisted and final conversions.
If you click on the other drop-down menu at the top, you can apply lots of other dimensions to this report like Source/medium, country, device category etc.
B. Model Comparison -
Go to the Conversions > Attribution> Model comparison tool
Select the Time decay model against Last interaction.
Now, look at the last column % change in conversions, as we can see Dynamic search ads have a positive 37% shift from the reference model in this case Last interaction. Again we can see that Dynamic search ads are performing better than we had given credit to it previously.
When the results are significant a green and red arrows are shown and when it is directional, up or down gray arrows shown.
Also look at the CPA column and compare between two model. We can see that the CPA for Dynamic search ads is not $50, it is actually $37 dollar.
Much better than the previous result.
If you want to understand and measure your campaigns more effectively then Please, Upload the cost data of your all marketing Channels and campaigns.
And Use Assisted conversion and model Comparison tool to better allocate your marketing budget/efforts across different marketing channels and campaigns.
I hope you like this post.