How to get the most out of AdWords smart bidding

A topic that has been featuring prominently in the news, machine learning is one of the most exciting advances in technology and one which is already having an impact on our everyday lives.

It was to be expected then, with Google being one of the major players in this arena, that such advances would become available to its AdWords users.

This article will review what smart bidding is, why we should use it, and the different tools available in AdWords, with the pros and cons of each.

what is machine learning and how does it relate to PPC bidding?

Wikipedia defines machine learning as a “field of computer science that gives computer systems the ability to learn with data, without being explicitly programmed”. In the case of automated bidding in paid search, the learning goal is to set the right bid for every auction, to maximise the likelihood of a desired outcome (cost-per-acquisition (CPA), return-on-advertising-spend (ROAS), etc..), while data fed to machine learning algorithms are known as “signals” (identifiable attributes about a person or their context at the time of a particular auction).

So, why use automated bidding?

There are two main reasons that make automated bidding a sensible choice to improve performance:

  1. Efficiency: The scale of data analysed at the time of the auction is way beyond what a single person or even a team of professionals could handle
  2. Additional “signals”: Bidding strategies take advantage of “signals” that are unavailable to be optimised through manual bidding

As the latter point constitutes a significant difference, let’s look at it in more detail.

Currently, you can optimise your bids with manual bid adjustment overlays for:

  • Device
  • Physical location
  • Weekday and time of the day
  • Remarketing lists
  • Demographic

In addition to these, smart bidding can analyse and optimise according to the following signals:

  • Location intent: The algorithm would raise bids for someone looking for a holiday in New York as opposed to being physically located there at the time of searching
  • Ad characteristics: Bids adjusted on a creative level, based on each creative’s likelihood of converting
  • Interface language: Evaluates predicted conversion probability based on someone’s language preferences
  • Browser: Historical conversion rate of each browser is taken into account at the time of bidding
  • Operating system: Same as above, but with operating systems
  • Actual search query: The actual query, as opposed to the keyword that has triggered it
  • Search network partner: Takes into account historical conversion rates of each website part of the display network
  • Price competitiveness: This signal is actively comparing price points for a specific product

It should be apparent by now that, even if minor efficiencies are achieved at each of these levels, smart bidding has the potential to substantially outperform manual bidding.

What tools are currently available?

Let’s look at the tools that are currently available and the characteristics of each.

Target CPA Bidding

Goal:  Maximise conversions while maintaining CPA below a target within a single or across multiple campaigns
Pros: Allows to set max or min bids (not recommended though) and to adjust target CPA by device using bid modifiers
Cons: A potentially lower amount of conversions could be achieved as the strategy lowers bid for conversions with a potentially higher CPA
Requirement: 30 conversions in the past 30 days

Target ROAS

Goal:  Optimise bids to increase conversion value or revenue in order to reach the desired ROAS
Pros: Fully automated, real-time bidding
Cons: No bid adjustments possible, with the exception of -100% for mobile traffic
Requirement: 15 conversions in the last 30 days, however 50 conversions in the past 30 days are recommended for optimal performance. Conversion value tracking is obviously a prerequisite

Maximise conversions

Goal: Maximise conversions within a campaign while spending the full budget
Pros: Fully automated, real-time bidding
Cons: Shared budgets are not allowed; furthermore, this is not an ideal choice for most advertisers working towards CPA or ROAS targets
Requirement: Conversion tracking set up and working

Enhanced cost per click (eCPC)

Goal: Increase the number of conversions while maintaining the same CPA
Pros: This strategy makes +/- % bid adjustments on manual bids (and modifiers) in line with potential conversion likelihood, therefore enabling you to retain a certain degree of control on bidding through setting base bids
Cons: Average CPCs may exceed max keyword bid in the short term, potentially reducing traffic when operating on a tight budget
Requirement: Conversion tracking

How to get started with automated bidding?

As always in PPC, extensive testing is required to ensure the best use is made of any new features. To this end, several tools will come in handy:

  • Campaign drafts and experiments can be used to safely test a new strategy on a small percentage of traffic; this will ensure that the test will not affect overall account performance during the learning periods
  • Bid simulators are available for several of these strategies enabling us to forecast how we might perform against different targets
  • Bid strategy status can be checked to see if bid strategies parameters need tweaking for optimal performance.

If you haven’t already, we strongly recommend you give these automated bidding strategies a try. In addition to increasing performance through more efficient bidding, these will free up more time for you to dedicate to the creative and strategic aspects of PPC.

Happy (automated) bidding!

To see how we can help you to optimise your current campaigns, contact us now!


by Croud
10 April 2018



Related posts