Understanding the future of SEA

How not to let Google have the upper hand on your campaigns?

The announcement made by Google last month according to which it would now limit the report on search terms has sent quite a shockwave through the SEA community. 

So far, that report has allowed advertisers to know most of the terms which users had typed for each purchased click. Just as happened earlier regarding SEO, Google has now chosen to become less transparent regarding queries. 

This news does not come as a complete surprise. After all, Google has become more and more of a black box regarding Search over the past three years. With Search accounting for more that 60% of Alphabet’s revenues, it is obviously quite an important matter for them. (see a chart from The Economist).

There is not a single advertiser who can oppose the Mountain View firm and call out the new direction it has been giving to its SEA strategy. What would their alternatives be ? 

SEA experts like me have even less influence. Yet our opinion should matter, as a warning to advertisers and certainly as an alert to competition authorities. 

In this article, I would like to make advertisers aware of how Google has increasingly been abusing its obvious dominant position - and has grown more and more opaque over the past few years regarding how it manages its main source of revenue - under the guise of doing machine learning. I also wish to give advertisers some tips on what to do now. 

What has changed over the past few years ? 

1. Keywords have become less and less precise

The first strong signal came in 2018. That year, Google decided to broaden the meaning it attributed to keywords (especially with exact matchs). This allowed it to show your ads on broader keywords than those you had chosen (including variants, synonyms, etc.)

The recent decision amplifies this trend and makes it increasingly difficult for advertisers to control which search terms lead to their ads. 

2. Google introduced new campaign types with limited targeting

Also in 2018, Google introduced UAC campaigns, a new type of campaign designed to promote mobile apps. It may have been a prefiguration of what future search campaigns might look like int the future, as it suppressed keyword targeting and the use of negative keywords. In UAC campaigns, advertisers can choose a budget, a geographic targeting, and the components of their ads: Google then places the ads on whichever queries it deems relevant in order to get the expected conversions. Advertisers are no longer involved in selecting the keywords, and cannot even exclude such words as their brand name or the mention “free” ;). 

Google also started pushing DSA campaigns around that time as well: a new type of campaign which scans the advertiser’s website and shows the ads on relevant keywords.It sounds simple, but it makes things more opaque since Google is free to decide which queries it deems relevant. 

3. Well-established management methods have been reconsidered

In 2020, an increasing number of keywords fell into the “Low Search Volume” category and stopped leading to any ads. This subtle change specifically took aim at those advertisers who favor a highly granular strategy, since it forced them to reassess their usual models and to modify the structure of their campaigns. 

Many advanced accounts are indeed structured with a multitude of ad groups, which ensures both a high level of precision in the ads and some control in the bidding process. The use of negative keywords in ad groups helps to avoid cannibalization. 

By making it impossible to show ads on keywords which are too precise (such as +training +adwords +paris, for instance), Google de facto imposes a new campaign structure in which ad groups include broader keywords. Grouped under a single keyword, bidding differences become impossible. 

Google has been accelerating this trend over the past few weeks by pushing more and more keywords (even those which have quite a bit of volume) into that category. Advertisers have no choice but to adapt or not be shown anymore. 

4. The supremacy of smart bidding

Finally, and alongside these changes, Google has been pushing advertisers to opt for smart bidding strategies. These automated bidding strategies leave Google free to set bidding prices while taking charge of hitting the campaign’s goals (in terms of clicks, conversions, target CPA, etc). 

This harassment has taken several forms: 

• Phone calls from account managers in charge of pushing these strategies
• Recommendations posted on the account and claiming the superior effectiveness of these strategies
• And finally some background work conducted with many intermediaries, especially the Google Partner agencies. 

Google’s recommendations regarding the effectiveness of smart-bidding strategies are rather unequivocal and leave no room to compare performance with other solutions. Furthermore, Google all-too frequently relies on the argument that smart bidding is more time-efficient and more massively used than manual bidding.

=> Smart bidding does work, but many users dread the removal of manual bidding, a system which still allows them to make their own choices. A few years ago, Google seemed ready to give some leeway to large advertisers while also offering smaller advertisers a range of products suited to their needs (such as Adwords Express). Over the past three years, Google has been leaning towards a single model for all advertisers. 

Why does Google want to keep the upper hand? 

To attract more and more advertisers

Those who seek to understand Google’s goals regarding SEA (and incidentally YouTube and Display) need look no further than Facebook. 

In just a few years, Facebook managed to attract almost 8 million businesses. How did it do that? Mainly by offering a simple, fluid onboarding process. As opposed to Google Ads and its piles of data which scare away many small businesses.

=> Google has been targeting the “long tail” of small businesses with:

• a new, slightly more user-friendly interface 
• a campaign onboarding process that puts forward goals which speak to all businesses
• the launch of Discovery campaigns - which manage ad placements on Display, YouTube and Gmail - and the recent beginning of smart campaigns
• DSA and UAC campaigns

To make advertisers bid for more queries

By depriving advertisers of part of the keywords they purchase, and by broadening the definition of some keywords, Google is sending advertisers a message that says “I know more than you do about which search terms are relevant to your activity. Therefore, I am offering you the possibility to show your ads on terms you may not have selected (or that you have not parametred well, for those with less experience).” 

While this approach may be valid for novice advertisers who have not yet mastered keyword nomenclature, or the semantics of their business, it is less relevant for larger clients. 

=> The broadening of potential search is first and foremost a way for Google to expand the queries covered by the advertisers and to increase the competition between them:  

For small advertisers

Due to a certain lack of knowledge, many advertisers do not cover the appropriate queries, or place ill-adapted biddings. Google does the job so that their campaigns are adequately shown and get results. For many small advertisers, this is certainly a good thing. 

For large advertisers 

The thinking should certainly be different for seasoned advertisers, because most of them do have some level of expertise. Why does Google not maintain the status quo for them? In other words, is Google seeking to make larger advertisers spend more? 

• Maintaining a constant spending level

Google can probably get away with maintaining revenue by offsetting the lower investments observed over weekends, holidays and vacation periods - i.e. times during which many businesses reduce their impression volumes, especially in B2B. 

There could certainly be growth opportunities for Google there. Thanks to smart bidding strategies and the broadening of search terms, Google’s algorithms could easily maintain stable spending levels during quiet periods if goals are met during busy periods.

Some advertisers have seen it during the lockdown: the level of spending for their target CPA campaigns generally remained unchanged during the first days, or even the first weeks of quarantine. Which was made possible, as Google warned users, by the fact that CPA targeting always has a margin of error. Let’s just say that sometimes that margin is rather wide. And easier to smooth over when it’s just a weekend. 

• Suppressing spending thresholds

Furthermore, by broadening search terms and controlling bidding, Google has blown up the bidding caps and impression ceilings which respectively limit the CPC for some keywords and the potential impression volumes of a campaign. Absent these “thresholds”, Google can push up the spending limits as long as it still manages to get conversions. 

Displaying the status “limited by budget” on a campaign is particularly effective in that regard, even when the target CPA goals have not yet been met. 

One could extrapolate that the Google advertising universe (Display, YouTube, and Search) might become part of a global Google inventory on which advertisers could show their ads. They would “entrust” Google with a budget, and Google would then proceed to reach their goals through different platforms. 

Why is this evolution in SEA not a good thing ? 

A disturbing lack of transparency

Google sometimes suggests search terms variants which are not relevant: for instance, it has been known to propose “email advertising” along with “email marketing”, even though these two words obviously have very different meanings. With a shorter search term report, users may legitimately wonder on which keywords Google has been showing their ads. One of your competitors? Or variants that are too far from your business? Many advertisers have learnt at their own expense how careless Google can get regarding YouTube ad placement. 

If Search campaigns were to take the same form as UAC (Universal App) campaigns, one can fear the worst. Google might resort to the same strategy as a fruit seller who gives you a flat price just to unload his stock of rotten produce on you, all the while taking advantage of your honest interest for the fresh stuff. 

Google becomes both judge and jury

Google Search is based on a bidding system. However, when the organizer of an auction starts playing the parts of both the auctioneer and the buyer, one is allowed to wonder to what extent they are serving their own interests. 

Search has been particularly profitable for Google because it is based on a CPC model. Advertisers are ready to pay for clicks (as opposed to impressions, as in the case of Display or Facebook) because they expect a high conversion rate. A limited number of advertisers thus compete for target keywords, and agree to pay the maximum price to reach a level at which the campaign becomes profitable. Advertisers are then able to touch a potential customer at just the right moment. 

Bidding has reached high levels lately because all queries are not equal, and advertisers specifically “target” those queries which are most relevant to their businesses. For instance, the bidding for “accounting software” is higher than that for “free accounting software” because many advertisers have calculated that the former should generate more paying customers than the latter. They certainly would not pay the same price for both queries. 

With smart bidding, Google deftly swamps those differences, all the while claiming that this system allows advertisers to both meet their goals and save time.  

Campaign optimization will become more limited

As an SEA expert, I have made some accounts grow their monthly budget from some tens of thousands euros up to the hundred thousands. If there is one acquisition channel I believe in to make a business take off, it is SEA. Account optimization has been key to push up conversion in those cases.

If you’ve tried your hand at growth hacking, you know just how important it is to “test & learn”. And you know that, in order to do it right, it is crucial to compare information on equal perimeters, i.e. by isolating possible causes as much as you can. Fully-automated campaigns make it extremely difficult to analyse the precise impact of a single action, because there is an infinite number of optimizations which may come into play. I once saw the numbers of an account I was working on plummet after we moved to target CPA. Analyzing the possible causes of that drop turned out to be impossible, and we had to switch back to manual. It is just not desirable to become dependent on Google’s algos to understand a trend, analyze performance, and improve campaign management. 

Let’s take the example of RSA ads (or Responsive Search ads). This ad format allows advertisers to set up a group of titles and descriptions for their ads. Google then takes care of switching around the combinations and shows those which continued learning has shown to be more effective. It all sounds so promising on paper. 

But it’s a whole other story if you’re trying to keep optimizing your performance. Let’s pretend that you’ve started a RSA with 6 titles. The algos learn to show the best combination for every query. A year later, you decide to add a new title to tell the world about your free product. Which titles are you going to delete? How do you make sure that the title about your free offer is indeed shown? How can you assess the impact of this new title on CTR, if your bidding level is set by Google all along? It’s impossible! Most advertisers (even the seasoned ones) will therefore just run the same message over and over again. The reasoning is the same for campaign structures with extremely high automation. 

Contrary to what some have said, ads and landing pages are not the only way to refine a campaign in this future model. Advertisers will need to keep just as close an eye on other factors as well.

How can advertisers adapt to this development? 

Track & Analyze by yourselves

Google is all-powerful when it comes to conversion tracking and to the attribution models used by many players. And Google Analytics is present on all websites. 

All ad platforms want to impose their own models (just like Facebook did with Facebook Analytics) because advertisers use these platforms to report their conversions. The Google Ads platform only needs the user to have one single point of contact with an SEA ad to register 1 conversion. While in reality that particular point of contact may have only played a minor role in the conversion process. 

Let’s pretend that I am looking for a project management solution. I hear good things about a solution called Monday.com, and I visit their website. A few days later, I type “best website for planning management” into Google. At this stage, we can expect that Google is going to show me an ad for Monday.com at the very top of my search page. If I click on this ad, Google will credit itself for that conversion. Google: 1 - Brand Awareness: 0. 

With the broadening of possible queries (and the fact that they are often hidden), Google maximizes the chances that a point of contact does happen before conversion. It’s all the easier because Google knows what websites you’ve been visiting. 

Regardless of their effectiveness, both advertising giants have high value in advertisers’ attribution models. What everyone must understand is that they are a quasi-inevitable point of contact with a lot of web users on whom they have a great deal of data. 

=> Advertisers must become able to analyze their conversions themselves, and to get a better understanding of their own attribution model through independent or proprietary models. Only then can they adequately quantify the acquisition costs they are ready to pay. Furthermore, they must remember that all conversions do not have the same value. Now more than ever, they must analyze with great precision the revenue generated by different conversions. 

Isolate “brand” queries as long as possible

Google is aware that many conversions can be obtained through “brand” queries. These “easy” conversions logically have very low acquisition costs, and advertisers should always isolate them.

Is Google planning on preventing the use of negative keywords, as is the case in UAC campaigns? It would make goals easily reachable, while cleverly lowering campaign acquisition costs. Which would free up some budget for other things. One can see how advertisers could easily get tricked if they are not careful enough. Some advertisers sometimes boast very good performance for their UAC campaigns, but it’s a whole different analysis if 80% of their conversions come from the brand. 

If you have any doubt, I suggest you check the search terms of your DSA campaigns. Even if you have excluded your brand name, you may just find some close variants in there. 

=> Google has managed to impose a model in which advertisers ultra-frequently purchase their own brand names. Intuitively, it is quite amazing to realize how much money Google is making from the welle stablished idea of protecting your own brand. As long as Google has that option, advertisers must conduct their brand and non-brand campaigns separately. Unfortunately, the automated model adroitly blends both types together. If the distinction becomes no longer possible, campaign managers will need to make careful use of audience exclusion. And in order to bother Google, advertisers should just stop bidding on their own brand name. They’d come out a lot stronger. 

Keeping control, in a different way

Advertisers are letting Google have the upper hand. Yet as long as Google does not handle 100% of keyword placement and biddings, you can keep some level of control: 

1. In your most important campaigns (i.e. those whose keywords are linked to your core business), do stick to manual bidding. Contrary to what has been said, bidding management is integral to SEA expertise.
2. Keep segmenting your campaigns, even the automated ones: it will allow you to pay the right acquisition cost per segment, and to avoid being charged for a fuzzy “global package”. Sticking to a set of homogeneous keywords for each campaign will always give your ads a more effective message and will push up your CTR.
3. Choose your ads with an aim to control their message as much as possible. If RSAs (Responsive Search Ads) become a norm, you’ll have to be very meticulous when you pick your ad’s components and their respective positions. Above all, avoid going for a host of completely different titles.

=> Many advertisers are ready to give Google full freedom to handle their campaigns. I have even read that it would give them more time for Facebook campaigns ;). In concrete terms, it means that they will allow their campaigns to run without doing the least bit of optimization. In such conditions, the slightest advantage will be decisive to create important performance differences. 

LET’S CONCLUDE by saying that most advertisers believe that the important debate to have these days is about the efficiency of smart bidding. This is totally misleading since the real debate is about transparency and control given to advertisers.

I truly believe in the efficiency of Google algorithms since I am using them. However the absence of choice would not benefit advertisers that are relying on SEA to get conversions.

Google Search is a fantastic channel to generate traffic with a high conversion intent and to grow a business. However, it is not necessarily suited for all businesses, and what works for Facebook may not work for Google. For example YouTube is bringing 15$bn of annual revenues to Google, but it still lacks the appeal that Instagram has among advertisers. In that sense, unexpected placements on YouTube have cooled several advertisers to the idea of spending heavily on this platform. It should not be forgotten that advertisers have reached those high levels of spending on SEA because it has given them access to premium and trustworthy traffic.

This “search-term gate” will not completely change SEA. But there is not doubt that the removal of manual bidding or keyword targeting would cause an outcry. It is in the interest of the entire market that such an earthquake does not happen.