What if I told you that you're paying too much for your Google AdWords clicks? What if I told you that all that time you've invested into building highly-targeted, long-tail Google AdWords campaign was in vain? What if I told you that only a small fraction of your thousands of Google AdWords keywords and ad messages will ever receive any impressions or clicks?
Well, very likely some or all of these are true.
To see why, let's suppose you have four Google AdWords keywords:
- red bikes
- red bikes sydney
- men's red bikes
- men's red bikes sydney
All four of your keywords are set to phrase match, meaning that they can only be triggered on Google if the searcher types a search query which includes that keyword as a phrase (i.e. those words in that order, plus anything before or after the phrase).
Each of your 4 keywords are in their own ads groups, and have their own tailored ad messages. The ad messages for your keyword 'men's red bikes', for example, mention 'Men's Red Bikes', and takes visitors to a page on your website showing your range of men's red bikes.
Now suppose someone searches Google for 'men's red bikes sydney'. Which of your 4 keywords will be triggered?
The answer is any of them.
Even though your keyword 'men's red bikes sydney' very closely matches the search 'men's red bikes sydney', your more generic keyword 'red bikes' could also be triggered, as this also satisfies the criteria of phrase matching. You could therefore be showing your more generic and less relevant 'red bikes' ad message, and taking visitors to a generic red bikes page on your website, even you have a much better ad message tailored to 'men's red bikes Sydney', which instead takes visitors to a page on your website showing your men's red bikes in Sydney.
Google's Help Centre points out that when multiple different keywords are eligible to show for a certain search query, Google will choose to show the keyword which has the highest Ad Rank. Ad Rank is made up based on a combination of Quality Score and CPC bid, meaning that all other things equal, if you have a higher bid for one particular keyword, that keyword will have a higher Ad Rank and therefore be more likely to show:
"When several ad groups contain keywords that match a search term, the system will prefer to use the keyword with the highest combined Quality Score and cost-per-click (CPC) bid. We call this combination Ad Rank."
What this means is that with a higher CPC bid, your more generic keywords which have a relatively low CTR and Quality Score, are more likely to show compared to your keywords which are more targeted and have a high CTR and Quality Score.
In other words, with a higher CPC bid, your generic keyword 'red bikes', which has a poor Quality Score of 3/10 and a poor CTR of 0.1%, can be more likely to show for specific searches such as 'men's red bikes sydney' than your more targeted and closely-matching long-tail keyword 'men's red bikes sydney', which has a good Quality Score of 9/10 and a good CTR of 5.0%. If your CPC bid for your generic keyword 'red bikes' is high enough relative to your specific keyword 'men's red bikes sydney' to more than compensate for the lower Quality Score and achieve a higher Ad Rank, then your more generic keyword is likely to show.
This is something which has been illustrated by Paul Downs in his article for The New York Times:
The daily budget hadn't been exhausted, but Google was only showing the ads that got the most impressions and the most clicks " even though their click-through rate was much lower than some of my other ad groups. All other ads were being shut off, so that more money could be devoted to the ads with the most traffic. This approach maximizes Google's revenue. If the impressions are huge, it doesn't matter to Google if the click-through rate is low " even though the minuscule rate is an indication that there is little real interest in the product.
In other words, my budget was being wasted. It was putting ads in front of people who weren't as interested in our products and couldn't really afford them. But because the number of people searching was so high, the total number of clicks generated far outstripped the traffic from my more focused ad groups. Google's algorithm saw the total number of clicks generated as evidence of success, regardless of whether we closed any business. By all of its own metrics, the AdWords campaign was a home run. I had received lots of impressions and bought lots of clicks. The only problem was that these apparently were the wrong clicks.
What Paul is describing here is Google diverting traffic to generic keywords with a low CTR, even though there are better keywords in the account which have a higher CTR and more targeted ad messages. Paul may also be right in pointing out that Google's technique to divert traffic to keywords which are more generic, more competitive, and therefore have higher CPCs, may in fact be in the best interests of Google to increase their revenue, in expense of its advertisers. This process of Google diverting clicks to the same (more generic) keywords is something I have previously proved can lead to higher CPCs and higher revenues for Google, at the expense of reduced profitability for advertisers.
In short, what I suggested in my analysis is that merging demand for two separate keywords (keyword 1 and keyword 2) to the same generic keyword, will generally lead to higher CPCs and reduced profits for advertisers, but higher revenues for Google.
Lower profits for advertisers, higher profits for Google
To illustrate how this is true, I will quickly demonstrate using marginal analysis and profit maximization theory (if you'd like to follow the logic more closely, and better understand what each graph shows, you might like to check out my full analysis here).
Now suppose you were bidding on just two keywords, keyword 1 (men's red bikes) and keyword 2 (red bikes sydney). As you can see with the grey shaded area below, keyword 1 (men's red bikes) generated a profit for the advertiser of ($0.75 - $0.30) x 2,000 clicks = $900.
Similarly, as you can see by shaded area below, keyword 2 (red bikes sydney) generated a profit for the advertiser of ($1.50 - $0.70) x 1,000 clicks = $800.
So adding these up, the total profit for keyword 1 (men's red bikes) and keyword 2 (red bikes sydney) is therefore $900 + $800 = $1,700.
However, if Google chooses to divert the traffic for both keyword 1 (men's red bikes) and keyword 2 (red bikes sydney) to the same generic keyword (red bikes), the profit for the advertiser, again denoted by the shaded are below, is now ($1.00 - $0.60) x 3,000 clicks = $1,200.
In other words, even though the advertiser is still receiving 3,000 clicks from exactly the same searches, due to Google diverting those searches to the same generic keyword, the advertiser's profit has fallen from $1,700 to $1,200.
The interesting part, however, is what happens to Google's revenue. For keyword 1, Google's revenue is $0.30 x 2,000 clicks = $600. For keyword 2, Google's revenue is $0.70 x 1,000 clicks = $700. Adding these up, Google's revenue is $600 + $700 = $1,300. So in the original scenario, where searches are directed to the more targeted keywords 'men's red bikes' and 'red bikes sydney', Google's revenue is $1,300.
However, when Google diverts the same searches to the generic keyword 'red bikes', Google's revenue now increases to $0.60 x 3,000 = $1,800.
For exactly the same searches, the advertiser's profit has fallen from $1,700 to $1,200, while Google's profits have increased from $1,300 to $1,800.
This process of diversion of searches towards less relevant generic searches is what Paul is describing in his article for The New York Times. Even though Paul has more relevant keywords in his Google AdWords account, which are able to show more targeted ad messages and probably generate a higher CTR and a higher Quality Score, it is generally in Google's interest to divert searches to broader, more generic, and more competitive keywords, which are likely raise Google's overall revenues.
4 strategies to regain control
Fortunately, there are a few techniques PPC advertisers can adopt to prevent or limit more specific and targeted search queries triggering your generic, shorter keywords, an instead push them towards your more specific and targeted keywords and ads:
1. Opt out of 'include plurals and close variants' in the campaign settings
If you have exact, phrase, and modified broad match variations of all your keywords, then there is no need to let Google match plurals and close variants to your exact and phrase match keywords. Any plurals and close variants will naturally match to your modified broad match keywords, allowing you to bid much higher on your exact and phrase match keywords. This will increase the Ad Rank of your exact and phrase match keywords relative to your modified broad match keywords, meaning that your more targeted, relevant, and long-tail keywords are more likely to show for more relevant, targeted, and long-tail searches, compared to your generic broad-matches keywords.
2. Use strategic bidding techniques
Structure your keywords based on the amount of keyword qualification they contain. This will allow you to set much higher bids for your more-qualified keywords such as 'men's red bikes sydney' than your less-qualified keywords such as 'red bikes'. By having progressively higher bids for your progressively more qualified keywords, you are likely to increase your Ad Rank for your more qualified keywords, making them more likely to show for eligible searches. So when someone searches for 'men's red bikes sydney', your more qualified keyword 'men's red bikes sydney' will be more likely to show for the search compared to your less qualified keyword 'red bikes'. You are strategically using your bids as bait for Google.
3. Funnel your keywords
If you have a campaign which contains keywords based around the different colours of your bikes, then the colours of your bikes should be placed as negative keywords in your other campaigns. This will mean that if someone searches for 'blue ladies bikes', you can be sure that your keyword 'blue ladies bikes' will be triggered, and not your generic keywords such as 'ladies bikes', which are located in your generic campaign. If you plan your campaign structure and keyword targeting in advance, you can develop a strategy to keep different types of keywords separate, and determine what types of negative keywords are needed to funnel the right searches to the right keywords and ad messages.
4. Use The Broad Match Generator
The Broad Match Generator is a technique I have developed to use your broad match keywords as a tool to generate new search queries. Broad match keywords are kept in a separate campaign, and once some search query data is collected, the search queries for this broad match campaign are analysed. If the search query is relevant, it is added as its own exact and phrase match keywords in a separate 'exact and phrase match' campaign, and given its own targeted ad messages. The new keyword is also added as exact and phrase match negative keywords to the broad match campaign, preventing any exact and phrase searches for that keyword matching to the broad match campaign again in the future. Alternatively, if a search query is not relevant, it is simply added as a negative keyword to all your campaigns. Over time, through expansion of exact and phrase match keywords, and expansion of exact and phrase match negative keywords, you can be sure that the only searches which match to your broad match campaign are those which cannot match to your exact and phrase keywords. It is a form of funnelling which is extremely effective at increasing your control over which searches get matched to which keywords.
Conclusion
Google will naturally take advantage of many a PPC advertiser's lack of knowledge, lack of expertise, and lack of foresight. Businesses who therefore fail to develop a highly-organised, strategic, and efficient Google AdWords campaign structure, will likely receive poor results from their campaigns, and pay excessively high CPCs for the privilege. However, businesses who invest in highly-targeted, tailored, and relevant long-tail campaigns, with strategic use of match types, negative keywords, CPC bids, and keyword qualification, are likely to develop a Google AdWords account in which Google are unable to exploit. CPCs will be lower, long-tail searchers will match to your long-tail ads, and overall results will improve.
The choice is yours.
If you liked this post, you might also enjoy 5 Steps to Improve your AdWords Account in 5 Minutes
Great post. I have yet to set up a big campaign myself but I have bookmarked this to keep in mind when the time comes next spring.
Excellent article Alan, huge amount of detail there. I’ve been staying away from PPC advertising because as you point out it can be a deep hole that I can’t afford to fill!
Alan,
Great post and this reinforces my logic on keyword funneling. One question – where did you get the $.60 average CPC (scenario 3) from?
Hi Paul,
The average CPC in scenario 3 can be anywhere between the $0.30 CPC in scenario 1 and the $0.70 CPC in scenario 2, with any amount between $0.30 – $0.70 resulting in the same outcome of lower profit for advertisers, and higher profit for Google.
The actual CPC value for keyword 1+2 will depend on the relative click volume for keywords 1 and 2. So if keyword 1 has considerably more click volume than keyword 2, the average CPC when grouping 1 and 2 together will be closer to keyword 1 ($0.30). Similarly, if keyword 2 has considerably more click volume than keyword 1, the average CPC of both keywords together will be skewed and weighted towards keyword 2.
So the average CPC of $0.60 I used in scenario 3 suggests that keyword 2 (CPC of $0.70) has a higher click volume than keyword 1 (CPC of $0.30), hence the combined CPC of $0.60 being closer to $0.70 than to $0.30.
You can choose any CPC amount between $0.30 – $0.70, and the end result will be the same (albeit with slightly different numbers).
Cheers,
Alan