What is smart bidding?
Smart bidding is a subset of automated bid strategies that use machine learning to optimise for conversions or conversion value in each and every auction – a feature known as “auction-time bidding”. Target CPA, Target ROAS, Maximise Conversions, Maximise Conversion Value and Enhanced CPC (ECPC) are all smart bidding strategies.
Why use smart bidding?
With smart bidding, you get 4 key benefits that can help you save time and improve performance:
1. Advanced machine learning
In bidding, machine learning algorithms train on data at a vast scale to help you make more accurate predictions across your account about how different bid amounts might impact conversions or conversion value. These algorithms factor in a wider range of parameters that impact performance more than a single person or team could compute.
2. Wide range of contextual signals
With auction-time bidding, you can factor in a wide range of signals into your bid optimizations. Signals are identifiable attributes about a person or their context at the time of a particular auction. This includes attributes like device and location, which are available as manual bid adjustments, plus additional signals and signal combinations exclusive to smart bidding. See a list of several of these important signals below.
3. Flexible performance controls
Smart bidding allows you to set performance targets and customise settings to your unique business goals:
- Optimize search bids to your selected attribution model, including data-driven attribution.
- Set device-specific performance targets for mobile, desktop and tablet with Target CPA bidding (coming soon).
4. Transparent performance reporting
Smart bidding offers reporting tools that give you deeper insight into your bidding performance and help you quickly troubleshoot any issues. These include:
- Detailed bid strategy statuses that show you what’s going on under the hood of your bidding.
- Campaign drafts and experiments for Search and Display campaigns that make it easy to test how well smart bidding performs against your current bidding method.
- Bid simulators that forecast the number of conversions that your ads might have received if you had set different CPA targets.
- Alerts and notifications that flag issues with conversion tracking and provide clear steps for fixing them.
- Rich performance reports that help you easily track performance changes back to updates to your bid strategy status or targets. Access these from the “bid strategy” drop-down menu in the main campaigns view or the shared library.
How to use smart bidding
Step 1: Assess the suitability of your account
Google has published guidance on this, but we want to add a bit more detail based on our own client experiences.
Below is a list of possible reasons for not setting up a smart bidding strategy on your account. This list is not necessarily exhaustive, but should cover the majority of cases.
A. Too new, not enough data
This depends partly on the bidding strategy you’re planning to use. Google’s advice is “30 conversions in the past 30 days” for Target CPA and “50 conversions in the past 30 days” for Target ROAS.
What about “Maximize Clicks” and “Enhanced CPC”?
Well, even though these are included as two of the four smart bidding strategies, they aren’t really in the same league as Target CPA and Target ROAS. So let’s ignore those for now.
Google’s advice provides a useful benchmark. The most important thing is that you feel confident in the reliability of your conversion tracking.
Once you have a method you’re happy with, and a decent volume of conversions, then we would say you were ready to go.
B. Conversions aren’t segmented
This relates to the above point.
If your account includes different types of conversions, then you’ll need to make sure these conversions are properly segmented before beginning any smart bidding strategy.
C. Missing data in Google Ads
Has your revenue data been properly incorporated into Google Ads?
Are there any other key pieces of data which, if absent, will mislead Google’s machine learning?
D. Branding is the main focus
If you want to optimize for control over your account. Fair enough.
If that’s your number one priority, then perhaps the opportunity costs are worth bearing. But that’s a big perhaps – in many cases, it might still be worth exploring new tactics.
E. Lack of trust or bad results in previous tests
Now some of you may not trust Google. Even if you may have already tested Google Ads and had bad results. Keep testing.
Ideally, smart bidding should be your default. It will require a lot of refinement in some cases, but that’s the beauty of Google Ads – so much opportunity for testing and learning.
F. Unstable market
Some businesses operate in highly unstable markets, where individual days can affect entire years. In some cases, the volatility will be so much as to discourage ever using smart bidding, if this is your situation then it might be wise to wait until things are more predictable, then give it a go.
G. KPIs don’t align with smart goals
What KPIs do smart bidding cover?
Currently, it’s website traffic (Maximize Clicks), low costs (ECPC), ROAS and, conversions – which, for most advertising, is probably going to be fine.
There are some exceptions, though.
A customer lifetime value is a difficult KPI to factor into smart bidding. Google’s machine learning doesn’t distinguish between new and existing customers.
Some businesses may have very specific goals, beyond the current scope of smart bidding.
Step 2: Select an appropriate smart bidding strategy
There are two tiers of smart bidding. Our advice is to move to the top tier as soon as possible. Although Google groups them together, we don’t think that Maximize Clicks, Enhanced CPC and Target Impression Share Bidding are the in the same category as Maximize Conversions, Target ROAS, and Target CPA.
Tier 1 = Maximize Conversions, Target ROAS and Target CPA
These are the gold standard of smart bidding strategies. Google’s machine learning will look across the whole of your account, and draw upon the exclusive auction signals it has access to.
Portfolio-level optimization is a big deal. Run some tests, and you’ll see that it tends to outperform keyword-level optimization almost every time.
Tier 2 = Maximize Clicks, Enhanced CPC and Target Impression Share Bidding
These all provide a specific automated function, which is cool but:
- It isn’t possible to apply at portfolio-level.
- It doesn’t take into account the full range of auction signals that the Tier 1 Smart Bidding strategies do.
A. Choose the strategy and targets
It should be reasonably easy to link your objectives to one of the smart bidding strategies.
Google has a table with the basics, in case you’re very new to all of this.
For example, if you want to “Get more conversions with your target CPA”, then Google suggests using…”Target CPA”. And so on.
The more subtle part is setting the parameters within the specific strategy.
B. Semantic optimization
How to rank 70% of your pages higher in 5 weeks. Trusted by over 30,000 webmasters, business owners, and SEO specialists.
You need to set targets that are closely related to prior performance and aim for incremental progress.
You’ll also need to choose a single campaign or portfolio level approach. As discussed already, portfolio-level offers the best opportunities for optimization, so aim for that if you can.
C. Adapt the existing account structure/ features
Single Keyword Ad Group
This is a bit more intricate, and really depends on each individual case.
Having a single keyword per ad group structure is compatible with smart bidding, and we can recommend combining the two so that your ad copy can be optimized.
D. Audiences
You can pretty much maintain the same audience strategy, but there are a few things to note:
- If you were using in-market audiences, you need to add them in again.
- Remarketing lists don’t need to be MECE (Mutually Exclusive, Collectively Exhaustive), which is actually a lot easier!
- You can’t use bid modifiers anymore.
Step 3: Establish a testing strategy
Basically, there are two questions that testing should help you answer:
- Is smart bidding better than the original bidding strategy?
- How can your smart bidding strategy be improved?
A. Use drafts and experiments (for now)
Drafts and Experiments (D&E) is the fairest way to test in-situ, and it has the advantage of being applied across campaigns.
Note that: the bid strategy must be applied to the control campaigns
This is so that data from control campaigns are carried forward (if it wins).
D&E is the best testing software we have for now.
In due course, theoretically, Google will launch “Multi-Campaign Experiments”, which looks like it’ll be even better for testing smart bidding.
B. Choose an approach that isolates success
For the best tests, you want about 80+ conversions/ month. This will let you quickly make comparisons between the control and the test group. Less than this, and it will take longer to establish statistically significant results.
Ideally, you want about 3-5 campaigns with a full migration and similar budgets.
A few other things to aim for:
- Portfolio bidding + shared budgets on small campaigns.
- Large, stable campaigns (e.g. avoid campaigns with high seasonality).
- Avoid overlap between smart and static campaigns.
It isn’t always possible to run such ideal tests, but this is definitely what you should be aiming for.
C. Testing smart bidding on shopping campaigns
Drafts and Experiments for shopping is still in beta. So, for now, you’ll need to find alternative ways of testing smart bidding for shopping accounts.
For accounts with reasonably stable activity, one option is to do a simple before-and-after test. It isn’t the most scientific, but it’s better than nothing.
Another option is to duplicate existing manually structured shopping campaigns, then split by location, and run at the same time. It’s possible to control for location to a large extent by splitting it randomly.
We did this for a few clients trying out Target ROAS and got some impressive results.
Step 4: Optimize performance
This is very much connected to step 3, testing, but involves actually doing something based on the results of those tests.
A. Optimize your testing
Finding the most effective testing strategy is a big job in itself. And so so important. Keep testing your testing.
B. The more data, the better
Smart bidding gets its smarts from Google’s machine learning, which relies on data. The more data you can provide, and the higher the quality of these data, the better.
Keep refining your conversion tracking, keep updating your conversion data. Wherever possible, use portfolio bid strategies to make use of more data
C. Be patient
Smart bidding is based on a process of learning. Learning takes time.
There may also be big fluctuations in performance, especially for the more ambitious target changes. Be brave, too.
D. Change your targets when appropriate
There is a balance between letting it run and making too many changes.
The problem with continually adjusting the target, is that there are likely to be short-term dips in performance based on Google’s algorithm needing to re-adjust/learn.
Sometimes, though, there will be external factors that require a change of target. Like during a peak sales season, for example.
Some businesses fluctuate a lot during a year, or over many years. Your targets should be sensitive to this.
A general rule of thumb: change your targets if you aren’t happy with the performance.
E. Use your time gains to focus on other search activities
Bidding used to be one of the most time-consuming parts of running a PPC account. Smart bidding has changed that.
Brilliant, now you can focus on other stuff instead. Things like:
- Keyword additions
- Strategic work
- Ad copy