The Basics of Migration to Automated Bidding

Here’s a common statement you might hear from a Google Ads professional: “Our manual CPC campaigns just stopped performing. CPA bidding dominates performance.” 

Advertisers have been switching from manual to automated bidding over the last several years in Google Ads and other ads platforms. In general the user experience, and narrative from Google, has been that CPA bidding is better for driving performance than manual bidding. 

Why is CPA bidding better? In theory, the idea is that Google is aligning conversion data from your Google Ads account to all the many data points that Google might have about each user, to determine how likely a given user is to convert. Google uses that to predict the best bid. Seemingly both in theory and in practice, Google does this more efficiently than a human can. 

A Basic Example of Automation Creating a Zero-Sum Environment 

• Imagine 10 advertisers all selling the same product, “blue widgets”
• All bid against each other in the same search market, competing for the same users
• If all 10 use manual bidding, all will be manually be trying to bid on the converting users
• If then 1/10 switches to automated bidding, better data and math allows this user to win
• However if all 10 advertisers switch to automated bidding, the efficiency is lost 

In other words, 10 people cannot all bid against each other more efficiently at the same time. The efficiency of automated bidding has come at the expense of the non-automated bidders in the auction. Once all advertisers adopt the same automation the efficiency is gone. In this case automated bidding has become a zero-sum game.

When Automation Creates a Positive-Sum Environment 

In the first scenario we imagined 10 advertisers selling the exact same product. All other things being equal, however they bid against each other, the net is zero-sum for the system in total. However there are many cases where net-positive outcomes could occur through better bidding.

One main reason is better sorting of users to advertisers. Even in competitive markets most products and services are not total commodities. Competitive products have different value propositions, target users, brand message and so on. Imagine two shoe stores bidding against each other for “women’s shoes”. One brand favors older women and the other younger women. Smarter bidding better allocates traffic between the brands, leading to a mutually winning auction scenario  in which more total shoes can be sold. 

A related underlying factor here would be that the output of most digital markets is variable dependant on user experience. There is not a fixed number of users who will buy shoes in any given month. More users can be convinced to buy with better pricing, offers, user experiences, and overall marketing strategy. Thus automated bidding can be a factor in driving total positive-sum growth of a market. 

What About Negative-Sum Outcomes

It appears possible that negative-sum total outcomes are possible due to broad advertiser adoption of smart bidding. Negative-sum in this case meaning that more total profit on ad spend is consumed by Google than in previous manual bidding scenarios. 

To elaborate: 

• In a manual bidding landscape, 10 advertisers compete, manually choosing their CPC
• Generally they have selected rational CPCs leading to profit
• There is an arbitrage gap in this market favoring the advertisers (rather than Google) 
• Each of the advertisers likely would be willing to “pay a little more” for their current traffic
• Lack of precision, coordination, and data keeps this arbitrage intact for advertisers 
• One advertiser migrates to CPA bidding
• This one advertiser realizes a short-term gain as they outcompete the other 9
• Once the other 9 move to smart bidding, the whole group has realized a net-negative
• Google has eaten the arbitrage that existed in the gray space of the manual CPC bids

There is Still Lots of Technical Opportunity in this Market

The good news for marketing operators is that there is still considerable execution opportunity in the overall space of bidding strategy and technology. Most advertisers are very far from having a perfect data and bidding environment. The average advertiser has both the opportunity to win zero-sum games by outcompeting and positive-sum games by generally using data to make user experiences better. 

An example of the ladder of sophistication in bidding automation:

Step 1: We don’t know how to bid

Step 2: We use manual CPC bidding based on the market average

Step 3: We use manual CPC, with various modifications based on conversion data

Step 4: We use a “maximize conversions” bidding strategy based on pixel data

Step 5: We use max conversions based on CRM data

Step 6: We test tCPA vs. max conversions based on CRM data

Step 7: We import user / deal value from the CRM and use tROAS

Step 8: Bidding on CRM data includes calls, chats, and other hard-to-integrate sources 

Step 9: We can include COGS in CRM data and bid to profit (tPOAS)

Step 10: We can adjust bidding for attribution-weighted incrementality (tPOIAS)

Summary

• Advertisers should still be adopting smart bidding
• Smartest advertiser in the auction will win: they capture efficiency from other advertisers 
• Get smarter with CRM imports, real dollar values in conversions, and so on 
• This is sometimes a Red Queen scenario where you are running a zero-sum race
• In the long run automation drives growth in market sizes and customer success
• However it might also drive marginal profit to asymptotically approach zero