You probably have heard of analyzing funnel fall-outs in the website purchase process – analyzing this helps us optimize for better conversions. However there is a lesser known technique called Search Funnel Analysis – where we look at visitor falling off an impressions (not clicking) or not converting after a click (Visits but does not purchase, may or may not have entered the purchase funnel).
This analysis relies on traditional marketing techniques like awareness & reinforcement – contributing to final purchase after many impressions, clicks & visits. The Search Funnel Analysis identifies these Awareness & Reinforcement contribution of these keywords – using terms like Impression Assists & Click Assists. Let us see how we can build a quantitative model to leverage traditional marketing approach in standard adwords optimization.
2.The second step is to click on “Columns” in the Keyword Report and select “Customize Columns”.
3.The third step is to add columns to the report from the Customize Columns section – need to choose the Search Funnel under Select Metrics and then add all the columns to the report.
4.Let’s take a look at the Search Funnel Report now
The parts that we are most interested in are bounded in red. Standard optimization techniques would consider CTR, CPC, Conversion Rate and Conversion Costs.
However we will be looking at the new search funnel metrics along with Average Position & Converetd Clciks – for advanced optimization. The logic used here is from Marketing as opposed to AdWOrds per se – although we will be using AdOWrds & Analytics data to make the optimization choices.
Let me take a simple example here.
One of the metrics we see here is Assist Impressions – let’s define what this is
An impression assist happens
a.when a Search Engine visitor searches with a search term “keyword 1”,
b.the ad impression shows up
c.the SE Visitor does not click on the ad
d.the visitor then returns to the search engine, tries a different search tem “keyword 2”
e.the ad impression shows up
f.SE Visitor clicks on the ad, visits the website and completes a transaction
In the above scenario, keyword 2 will get credit for the transaction. However since keyword 1 brought the brand in front of the customer at an earlier stage, it is considered to have played a role in the conversion as well – called as Impression Assisted Conversion.
The impression from “Keyword 1” was in the path of Search to Visit to Conversion process and is considered to have assisted the final conversion.
An Impression Assisted Conversion is counted only when the impression (without ad click) from Keyword 1 contributed to a conversion – through a click from an Ad impression shown when Keyword 2 was searched for (not “keyword 1” again). This “keyword 2” can be from the same campaign or a different campaign.
The other metric to consider is Assist Clicks
A click assist happens
g.when a Search Engine visitor searches with a search term “keyword 1”,
h.the ad impression shows up
i.the SE Visitor clicks on the ad but does not purchase
j.the visitor then returns to the search engine, tries a different search tem “keyword 2”
k.the ad impression shows up
l.SE Visitor clicks on the ad, visits the website and completes a transaction
In the above scenario, keyword 2 will get credit for the transaction. However since keyword 1 brought the customer to the website at an earlier stage, it is considered to have played a role in the conversion as well – called as Click Assisted Conversion.
A Click Assisted Conversion is counted only when the click (without a purchase) from Keyword 1 contributed to a conversion – througha click from an Ad impression shown when Keyword 2 was searched for (not “keyword 1” again). This keyword 2 can be from the same campaign or a different campaign.
Here is an example for Impressions Assist – I search for an air ticket with the search term “flight ticket from Chennai to Mumbai” and see an ad for Spice Jet. However I do not click on it. Next time I could search using the term Spice Jet, land on the site and purchase a ticket. Obviously the first search term “flight ticket from Chennai to Mumbai” introduced me to the brand and enabled a conversion during the next search.
Similarly here is an example for Click Assist, I could click on the Spice Jet Ad when I used the search term “flight ticket from Chennai to Mumbai” and visit the website the first time – however I did not complete a tranaction. Next time I could search using the term Spice Jet, land on the site and purchase a ticket. Obviously the first search term “flight ticket from Chennai to Mumbai” introduced me to the website, services & pricing and enabled a conversion during the next search.
Given that Impression Assists and Click Assists are common in many medium and large sized accounts, let us see how we optimize using advanced techniques.
Step 1: How do we credit keywords that contribute to impression & click assists? Would you value a click assist more than the impression assist?
One way to do this is to approximate # of impression assists to one conversion – say if it takes seven impressions before a searcher becomes a customer, we can approximate impression assit to 1/7th of a conversion (0.14).
Similarly if it takes 3 clicks on an average before an adwords visitor transacts on the website, we can allocate 1/3rd of a conversion credit to Click Assists.
The metrics required for the above weightage calculations are available in Google Analytics & AdWords.
So in this specific exercise let us assume Impression Assists carry 10% conversion value and Click Assists carry 25% conversion value. I am now creating additional columns in the Search Funnel Reports – to account for these notional conversions.
Adjusted Click Assisted Conversions for Keyword 1 = 0.25*217 = 54
Adjusted Impression Assisted Conversions for Keyword 1 = 0.1*1023 = 102
AdWords Converted Clicks for Keyword 1 = 254
Adjusted Conversions for Keyword 1 = 54+102+254 = 410 (Adjusted Conversions Column)
Actual Cost for Keyword 1 = Rs. 173,004
Adjusted Cost Per Conversion = Rs.173,004 / 410 = Rs 421 (Adjusted Cost per Converted Click Column)
AdWords Cost per Click for Keyword 1 = Rs. 681
Variance after adjustment = 38%
How do we interpret this?
1.Higher the variance (downward trend in costs) – more valuable is that keyword for Assist Conversions. These are the keywords we would like to spend that extra money on
2.The other metrics we should look at are
a. existing Search Impression Share – can I spend for more impressions?
b. existing average ad position (below 2.5 or above 3?) & position (Top or Other) – can I spend for additional clicks?
3.All things above being equal, we can prioritize by comparing engagement metrics imported from Google Analytics – bounce rates, pages per session and micro conversions.
Based on the above rules, we can prioritize keywords that offer the maximum conversion & costs benefits along with the ability to handle additional spends. For example, a keyword with 40% variance in conversion costs is attractive, but if it is already at 95% impression share and ad position of 1.2, there is no leverage gained by additional spend – it is already maxed out.
It is also important to remember that these are iterative. The above report is just a starting point and we need to monitor performance of these keywords on a weekly basis – once the bid changes are applied.
There are other reports that can be reviewed for optimization – by Device & Paid vs Organic reports. Will review them later.