A deeper dive into Google’s advertising business
I wanted to give Alphabet investors or those interested in learning more about Google’s search and ad business a deeper understanding of how Google Ads work and why it’s a particularly powerful form of advertising. This is a little bit more granular with detail because FinTwit investors already know almost everything about Alphabet from a high level point of view. This is for those interested in the nuts and bolts: the how and why.
First things first
It’s important to know that there are multiple dimensions of Google’s search business and a variety of ways to approach getting traffic from a merchant/website standpoint. Technically, you don’t need to advertise to get traffic, but if you are a small business operating on planet earth in what is even a semi-competitive industry, you might not have a viable path forward without ads.
There’s a couple reasons for this: First is that 75% of all Google queries never make it past the first page, and second, well, look at the screenshot below of the organic search results for the query “white running shoes”. In order to rank organically on the first page of this query, you’ll need to displace one of the top 10 results. These are 5 results, but let me know which business you’ll knock out.
Ranking organically on the first page for important queries is almost impossible. Google’s algorithm considers hundreds of on-page signals when comprising its rankings, and page and domain level relevance and intent are extremely important. The issue is that Google learned a long time ago that any business could game the algorithm by having the phrase “buy white running shoes” (intent: “buy”, relevance:”white running shoes”) appear thousands of times on a page, and voila, they would rank on the first page for that particular query.
To combat this, Google introduced the concept of PageRank into their algorithms (more info here: https://en.wikipedia.org/wiki/PageRank). Your page still had to be relevant, but now it had to be proven with social cred -- which in internet terms simply means backlinks. The quantity and especially the quality of links back to a particular page mattered a lot to Google algorithms. This made a whole lot of sense in a number ways: for one it made the system harder to game, and two, it used social signals from around the internet to determine which pages to display for a particular query. The end result is a much better user experience with higher quality search results.
A secondary effect of PageRank was that it now became harder for small businesses to outrank established ones, short of unbelievably savvy and viral marketing. Before you cry foul, just remember that most consumers would rather see Amazon rank organically than a small business based in Missouri, for example. That’s just the way things work.
So now that we’ve established that it’s not a viable strategy to create a small business and expect results without advertising, let’s look at the options these businesses have to advertise their product.
The first Google Adwords product was a text-based form of advertising centered around keywords. The underlying mechanics of why brooksrunning.com is ranked first and nike.com is ranked second in the query above is a bit complicated. The first thing to know is that every commercial query on Google results in a real-time auction to determine order of ads. Here again, there are multiple components to how Google ranks ads, with bid on a keyword only one part of the equation. Remember, Google wants to show relevant results, period. Accounting for the highest bids only in the Ads algorithm is a recipe for a terrible user experience.
To make Ad results as relevant as possible for the user, Google introduced the concept of AdRank. Essentially, it is multiplying the advertiser’s bid with the Quality Score (QS itself consists of multiple components, which I’ll explain in a second). The higher the Ad Rank, the higher the page will rank.
Here’s a great graphic on how it works:
What goes into Quality Score? Again, Google wants to show relevant results. Let’s look at the ad results for the “white tennis shoes” again for example. There are multiple possibilities as to why the Ads are showing in the order that they are. First, we don’t really know which keywords either company is targeting. On keyword-based advertising, advertisers can target queries in a few different ways:
Phrase match: Google allows you to target a keyword in phrase match format. So for example, you can bid for the keyword “running shoes” and rank for “white running shoes”. Going one step further, you could also target “shoes” and rank for “white running shoes”. With phrase match, you’ll rank for the keyword itself and any other query with words before and/or after the keyword.
Exact match: You can choose to rank only for queries that exactly match your keyword. In this case, you’d need to target [white running shoes] exactly to rank.
Broad match: This is a very loose match; so if you target white running shoes, you might also rank for the query “white tennis shoes” because they are generally related.
Modified broad match: Allows you to rank for words out of order as long as all words you are targeting are inside the query. So for example if you target +white +running +shoes you will also rank for a query like “white nike running shoes”.
You can supplement these match types with negative keywords that block your ad from showing up for certain queries. So for the above example, if I wanted to rank for “white running shoes” but don’t carry Nike brand, I might want to have “nike” as a negative keyword. This helps achieve higher ROI over the long term.
So Quality Score is determined by the quality and relevance of the landing page (the page the user lands on after clicking on the ad) as well as how closely the keyword being advertised matches the user query. So for example if brooksrunning.com is targeting an exact match [white running shoes] while nike.com is targeting a phrase match “shoes”, you can assume that the Quality Score for brooksrunning.com is higher, and thus the brooksrunning.com Ad will rank higher, all things being equal including bids. Because of quality score, it is also possible to rank higher despite a lower bid. Why? Because the CPC is actually determined by dividing the AdRank (remember, this consists of QS and bid) of the next bidder with your AdRank plus 1 cent. The end result is that those businesses with high quality scores, and better relevance, will have much better economics. This dynamic alone helps Google deliver better Ads results as this plays out over time.
Here’s a good graphic explaining that.
Another advertising option, primarily for e-commerce businesses, is Shopping ads.
Businesses would need to upload a spreadsheet of products and bid on the product itself. This is different from text ads in that the bidding is on the product level. Relevance is achieved by matching the name and description of the product passed in the feed to the user query. The closer the match, the higher the quality score.
Google itself will determine which queries your product should show up for, and although the advertiser has less control, the ROI is higher for Shopping Ads than for Text Ads. The reason is that the picture and price act as a filter for attracting the right buyer. With Text Ads it’s much more difficult to convey your value proposition. Otherwise, the auction mechanics and bidding process are similar to text ads.
Because Google will choose the queries that a certain product will rank for, advertisers can expect Shopping ads to rank for a much wider range of queries, increasing almost exponentially for those products with higher relative bids. Here’s an example of how that works out, form the Google Ads back end itself:
You can see that with a $0.79 bid, advertisers for a particular set of products would get 109 clicks, but with a 3x bid of $2.50 they could get approximately 10x more traffic for ~25x the cost. The reason is that as the bids go up, the likelier the product is to rank for less relevant queries (remember, because of AdRank, if QS score is low, bids will need to be really high to rank). It’s up to the advertisers to find the sweet spot for profitability. Which leads to another note: those businesses with higher gross margins can grow revenue very fast with high bids that competitors with lower gross margins cannot realistically meet.
Another form of advertising is Google’s display business. It’s essentially about showing ads on Google’s partner websites, and bears more similarity to Facebook’s ad business than the other options listed above. While advertisers can layer in additional keyword, topic, and audience targeting, advertisers in this case are bidding to show banners ads on websites. These ads tend to be called “top of the funnel” and are more branded in nature. As a result, the ROI is much lower than Text ads and Shopping ads, and hence the CPC’s are much lower as well.
Because of the enormity of the internet, and the prevalence of Google’s AdSense product, advertisers can get a whole lot of traffic from Google Display ads. The quality of traffic is debatable, but again, because Google Ads are auction based, the CPC’s usually end up being where they should be.
Advertisers have the option to layer some other strategies on top of what’s mentioned above. This includes having a bid modifier on devices, specific locations, specific audiences, hours during the day, etc. The level of advertiser control is extremely granular.
Bid modifiers on devices is an important consideration given that more than 65% of all clicks are now on mobile and conversion rates still lag desktop. Google allows advertisers to apply a negative modifier to normalize bids across devices while still giving you data on cross-device conversions.
Advertisers can easily push their campaigns to YouTube by checking a simple checkbox. Obviously for video ads, it would require creating video/content, but nonetheless, advertisers can access Youtube very quickly. In fact, the ability to push campaigns out immediately to Google Search, partners, and Youtube is done with a few checkboxes on campaign creating.
Recently, Google has invested a lot in machine learning, with the idea to help advertisers automatically achieve higher ROI across all their campaigns. Advertisers can use Google’s machine learning algorithms in multiple ways. First is to give Google a little bit of control over the bidding process and allow them to bid on the ads based on an ROI the advertiser would like to achieve. Google will begin to serve ads and then continually learn over time, differentiating bids in real time by taking into account keywords, products, conversion rates, browser information, device information, location of user, previous conversion data of the user, and so much more.
“Smart campaigns” takes this idea to another level, effectively giving Google complete control of the campaign and bidding process and allowing Google to determine where and when to show ads to achieve the highest possible ROI. This might include showing ads in partner websites and YouTube, and blurs the lines between the search, shopping, and display businesses.
Essentially, with a single click of a button, Google will do everything to get the most possible conversions at the highest possible ROI. Google is actively trying to convince advertisers to merge their current campaigns to “Smart Shopping”. For new advertisers with little experience and little time it makes a lot of sense to give it a try from the get-go.
One of the really powerful things about Google Ads is an almost instantaneous feedback loop. The data on the campaigns filters in almost immediately and advertisers can make changes to their campaigns, messaging, website, products, pricing, and ads quickly.
The data itself is incredibly precise. Advertisers can see the exact queries, costs, conversion rates, and so forth. Advertisers can also see their impression shares relative to other advertisers in a particular campaign.
The data itself is given in tables and charts and advertisers are able to compare the data with previous time periods. It’s an incredibly intuitive backend tool for creating and managing campaigns and using the data to adjust on the fly.
This leads me to my last thoughts, particularly about competition in the advertising space. Google owns intent when you consider its Text ads and Shopping ads. Although Facebook is obviously its nearest digital advertising competitor, Amazon is the only other business that can match Google’s conversion rates, and may actually be higher.
What makes Google a better advertising business for merchants is that it’s strictly a CPC model whereas Amazon is a CPA + CPC model. Remember, advertisers on Amazon are third party businesses that pay Amazon a cut of the acquisition. Certainly, it’s getting harder and harder to rank organically on Amazon and some third party businesses are forced to spend on advertising there. The question becomes how much margin in addition to the acquisition cut they are willing to forgo.
Another thing to consider with Google Ads is that it encompasses a wide range of industries whereas Amazon focuses mostly on retail. The full scope of Google Ads is wide and retail is just one vertical. Travel is another large segment of Google Ads, including flights and hotels. Digital services and software and local services such as restaurants, therapists, doctors, lawyers are other examples.
Facebook is a formidable foe. If Google dominates the bottom of the funnel then Facebook absolutely dominates top of the funnel and branded advertising. What’s interesting is that while Google is swimming up the funnel with display and Youtube, Facebook is trying to get closer to the purchase point with Facebook Shops and various payment initiatives. So in a sense they are complimentary advertising products but converging at the same time. What makes Facebook especially impressive is the unbelievable understanding and targeting of people and demographics, to the point of surgical precision. On top of that, their Ad formats are second to none with rich full-screen mobile ads.
Bing is another competitor, albeit to a lesser degree. The ad formats are essentially identical to Google and this is done on purpose. In fact, Bing allows you to import Google campaigns with a click of a button or even allow it to import campaigns on a predetermined schedule. Nevertheless, there are still certain things that make Google ads much more powerful.
Bing is not just not quite as well rounded as Google. Data collection and presentation -- including charts -- on Bing feel like it is 5-10 years behind Google. Secondly, the underlying machine learning and to some degree, Artificial intelligence that makes Google Ads so powerful is missing in Bing.
For example, Bing copied a Google feature with “close variants” keyword matching. The issue is Bing has a very loose definition of what “close variants” actually means, and in some cases advertisers may find that they’re paying for unrelated keywords, resulting in much lower ROI. This is probably because Google understands user queries a whole lot better than Bing does.
Critical options and settings are missing in Bing. Negative keywords don’t work effectively in some cases. Bulk changes are needlessly difficult. The bottom line is that there are a whole set of issues on Bing’s backend that make it an incredibly frustrating experience for advertisers. The end result is that the ROI is consistently below what can be achieved on Google. Combined with a much smaller user base for Bing search, advertisers tend to focus on Bing much less. Remember, Google Ads is Alphabet’s core business and they continue to innovate to separate it from the pack.
Having said all that, Google continues to innovate around it’s ads business. For example, last year they announced Gallery ads, basically allowing business to add images to Text ads. To an outsider this may seem like a minor thing, but images make it easier to convey the value proposition and if that leads to higher ROI, that will lead to more spend on Google Ads. Underpinning ad campaigns with machine learning and artificial intelligence is a huge opportunity that can lead to higher ROI for advertisers as well, so Google is investing there as well. Overall, I would expect Google to keep innovating and providing advertisers with amazing tools to reach new customers.
This was an excellent breakdown.
For the CPC calculation, did you mean to say "by dividing the AdRank (remember, this consists of QS and bid) of the next bidder with your QS plus 1 cent?" This is what the table shows.
Thank you so much for the write-up!