The bulletproof recipe for achieving success with Smart Bidding
Smart Bidding has been bashed a lot in recent years, but in 2019, the sentiment started to change. Advertisers started seeing decent results. Now, more advertisers are using Smart Bidding than ever before.
As an agency that has no allegiance to specific vendors, tools, or ways of getting Google Ads to work, we experiment a lot. With Smart Bidding, there is a big benefit compared to other third-party tools: It’s free.
So our biggest question has been: Under what circumstances is Smart Bidding the right choice for an account?
I can easily say that it’s not always the case. But, we’ve distilled it down to a couple of key principles combined with ways to mitigate the disadvantages of Smart Bidding.
And that’s been the key to us finding success with Smart Bidding in some cases; that we know what makes it tick, what makes it perform well, and why it performs poorly. Then, we can take actions that build upon this rather than just giving up when it’s not working.
The last thing we can do as an agency is to say, “We’re running Smart Bidding, so there’s nothing we can do.”
That’s not good enough. We can’t afford those excuses when we’re paid well. We’re paid to be the experts, so experts we shall be!
This article has a positive sentiment towards Smart Bidding. I’m not generally THIS positive about it, but I believe there are enough articles bashing it. We don’t need more of those. If you want it to work, this article can help you get as close as possible to making Smart Bidding work for you.
The truth is that it’s free, and it works pretty well in enough cases that it’d be reckless not to see if you can make it work for you.
Note: I’ve gone in-depth with the pros and cons of automated bid management in general already. This article will solely focus on the Smart Bidding feature of Google Ads.
The Three Principles For Making Smart Bidding Successful
I work with three principles that need to be in place before Smart Bidding can be successful:
- Data-volume must be significant
- Stability in your data
- Predictability in the market
These three principles should be in the back of your head every time you consider whether Smart Bidding works, or if you should try it out.
Let’s briefly go through them one by one:
Data-volume must be significant
I like to have at least 250 conversions a month per campaign type (search/Shopping) before I venture into Smart Bidding. That’s not to say that this is the only time it works, but based on our cases, that’s when we typically see it working most of the time.
My thinking is that the more data Google has, the more they know about their users, which they can use to make Smart Bidding work.
Stability in your data
Your data absolutely has to be stable. If the average order value changes too much from week to week, then you will have some days with very high ROAS and some days with very low ROAS then Smart Bidding will have a challenge working with it.
The same goes if you have conversions scattered across days of the week. This could be if your sales are very heavy on the weekend, then Smart Bidding will have a tough time figuring out what to bid when.
I’m sure Google is working on making Smart Bidding better at these fluctuations, but it’s a tough nut to crack. Because sometimes the instability just means a temporary dip (like right before payday) and you don’t want to stop Google Ads those days.
Predictability in the market
Smart Bidding works well when it can predict things based on historical data.
Are your Wednesdays always bad conversion-wise? Smart Bidding can use that.
Is keyword X converting at 10% on desktop and 1% on mobile? Smart Bidding can use that.
Are your products on sale converting better than non-sale products? Smart Bidding can use this, too.
The easier your market is to predict, the better Smart Bidding will perform.
Smart Bidding got much better in 2019
I know huge agencies who have built proprietary bidding engines, but have scrapped their entire teams to use Smart Bidding instead. Where Smart Bidding was awful in the early days, nowadays it’s getting better than the typical PPC manager.
So just because you’ve tried it in the past and it’s failed doesn’t mean that it won’t work for you in the future.
Why Smart Bidding (on paper) is better than third-party tools
Basically, the biggest argument to why Smart Bidding is better is because Google has access to more data than third-party tools. The more data, the better the bidding will perform.
With that being said, I’m not convinced that they actually use the data. The minimum amount of conversions for Smart Bidding to be enabled is 50ish. And if Google is using 30+ signals, then they have an incredibly narrow dataset.
Another big benefit is that it’s free. Few challenges have annoyed me more than finding a bidding software for SavvyRevenue since we started. It’s almost impossible to find a solid bidding software that outperforms manual and is affordable (not cheap—we don’t need it to be cheap, we need it to outperform us with more than it costs—it’s that simple).
Smart Bidding also bids according to Search Terms, which is impossible for any third party software to do in Google search ads or Shopping ads. For search ads, you can add the keywords and bid based on them, but you can’t do the same in Shopping. So Google has a huge advantage there, and it’s typically why Smart Shopping can perform much better than when you try to manage it manually.
At the same time, Smart Bidding can bid 1-to-1 on audience data, device data, search terms, etc. Third-party bidding software can “only” make bid adjustments on an ad group basis. Keeping track of the various bid adjustments combined with the individual keyword bids is a lot for any tool to handle.
And I honestly haven’t found a tool that can do it super well. They’re out there, and they’re better than manual, but I’m not super happy with some of them.
The biggest mistakes many make with Smart Bidding
I’d like to square away three mistakes that everyone seems to make when it comes to Smart Bidding:
- ROAS targets are too high
- It’s not a magic pill
- Not getting into the weeds when optimizing
ROAS targets are too high
I simply don’t know why this is, but most people make the mistake of saying, “We want a 500% ROAS, so we are going to apply a 500% ROAS target for our Smart Bidding.”
And then they get disappointed when their campaigns tank in terms of revenue.
Well, of course, it will. Your campaigns were performing at a 300% ROAS, and you asked the system to get you a 500% ROAS.
It’s not a magic pill
Smart Bidding only solves bid management.
Again, it ONLY solves the bid management aspect of optimizing your Google Ads account.
It doesn’t improve your ads, select better keywords, choose the right products, or add audiences.
Not getting into the weeds when optimizing
Similarly, when you start Smart Bidding, many people will get out of the habit of getting into the weeds of the account. They will forget to do basic optimizations or understand trends in the account.
Again, this will hurt your overall optimization efforts and, therefore, your account performance. It’s not Smart Bidding’s fault that you, all of a sudden, stopped doing the work. But it’s a side-effect that you should know.
How to Get the Most Out of Smart Bidding
Set appropriate targets, and launch in stages
When you’re starting out, set targets that are equal to the performance for the campaigns you’re applying them to:
This actually works.
The idea behind setting your Target ROAS to the same as your campaign(s) have performed is twofold:
- It will maintain the same conversion volume as before while;
- It learns what works and what doesn’t work
So during the initial phase, Smart Bidding will decrease bids for search terms that perform below your target and increase bids for search terms that perform above the target.
The higher you set your target ROAS, the more strain you’ll put on the system to perform.
So start with baby steps.
Launch in stages
When you launch, Smart Bidding needs time to learn and make changes to bids. I have seen it work from one week to the next, but it’s very rare.
My best recommendation is to stagger your launch, so you take five campaigns at a time like this:
Slow and steady wins the race if you want it to work.
Note: I highly recommend choosing top-performing campaigns for your testing. Maybe not your #1 or #2 campaigns, but campaigns #3 – #8 based on revenue would be ideal.
Don’t try to take your worst-performing campaigns and apply Smart Bidding. It’s not a miracle cure (and again, I can reference the three principles for making Smart Bidding work).
Portfolios Generally Perform Better than Individual Campaigns
You can apply Smart Bidding to a group of campaigns sharing a single target (called portfolios) on an individual campaign basis.
Never do this on an ad group basis. It’ll rarely be necessary to do so.
I have routinely seen a better performance with portfolios.
There are a couple of reasons for this:
- It allows Google to better balance the need for every keyword to hit the target (some campaigns can be at 400% and another at 600% averaging to your 500% target)
- It gives Google more data to predict performance, and set bids
- It’s easier for you to control instead of having a 100+ different targets in large accounts
Also, Google is adding more and more features to portfolio bidding strategies. The first one is Seasonality Adjustments, which is out already:
But more are hinted to come (like date blackouts, which is in beta right now).
Create 2-3 Portfolios
I like creating 2-3 portfolios with different ROAS targets.
Let’s say my overall account target is a 500% ROAS.
I will then create three portfolios with the following targets:
- 300% Target ROAS
- 500% Target ROAS
- 700% Target ROAS
This allows me to move campaigns around when I need to. One reason could be that I need to increase the exposure of a certain category, which is moving into high season. So, I move it from the 500% target to the 300% target. All of a sudden, Smart Bidding has a big campaign that can get higher bids and higher exposure.
And the same works the opposite way.
Your budget can be a bidding signal
If you’re spending $1,000/day and set your budget to $10,000/day, Smart Bidding can take this as a signal to increase spend.
It’s not a bad thing. Just use it appropriately, especially for Dynamic Search Ads campaigns.
Always set minimum and maximum bids
This is another reason to use portfolio bid strategies. It allows you to put a cap to how high (or low) Google can set your bids.
I always set a maximum and minimum bid:
For some campaigns, I set it to twice the normal maximum CPC. For others, I recommend you set it to a more conservative.
Google will tell you when your bidding strategy is being limited by the maximum or minimum limits, so no need to be too careful here. You can always update it.
Google starts bidding an insanely high CPC because something has happened in the market, so make sure you watch for that.
A great example of this is when COVID-19 was discovered in China, and some markets saw an increase in demand:
ROAS increased a lot, so Smart Bidding thought it should increase bids. However, it didn’t need to increase bids. We were already maximizing our Impression Share, so the increased CPCs were not needed and just decreased our ROAS.
This was actually a case where we moved off of Smart Bidding for a while until the markets became more stable.Again, I’m referencing the three principles for Smart Bidding success.
Be careful with advanced portfolio controls (seasonality bid adjustments)
The idea behind seasonality bid adjustments for portfolio bidding is that you can tell Google you expect your conversion rate to increase during a date range.
So far, so good.
The problem is that the system is a tad bit too simple in its current state.
When you create a seasonal bid adjustment where you predict that your conversion rate increases by 100% (not unlikely during big sales), then Google will increase your CPC by 100% (or around there).
So when you make these seasonal bid adjustments, make sure you set them to how much you want your average CPCs to increase.
Play with it, and make sure that the first time you use them isn’t during one of the biggest sales events of the year (or Black Friday).
CPA vs. Target ROAS: Which is best?
Duh, this is eCommerce, Andrew!
Yeah, but not always.
Smart Bidding can sometimes struggle when your average order values are really high. If your AOV is above $2,000, then your keywords will routinely have a ROAS of +100x.
Smart Bidding has no idea what to do with 100x ROAS keywords, so it just increases your bids a ton:
That’s where Target CPA comes in.
It takes the conversion value out of the equation, so Smart Bidding is just bidding based on conversions, which is not as unstable (again, back to the three principles).
It does require that you know exactly what you want your Target CPA to be to achieve a certain ROAS. Play around with it. If you set a Target CPA that results in a too high ROAS, then you decrease the CPA target next month. Easy peasy.
Your account segmentation should reflect smart bidding
The idea with Smart Bidding is that the machine uses the data it gets to set the correct bids.
Many of the highly granular ways to structure Google Ads accounts in the past was meant to segment an account, so we could bid according to
- Hour of day (for restaurants)
And a few others as well as assign budgets to match-type specific keywords.
It resulted in very granular accounts that, especially in big markets (like the US), could perform really well.
But, if you are going all-in on Smart Bidding, this is not necessary. All the segmentation that you did before to help you make better bidding decisions should be dismantled because you will not use it anymore. Actually, you can hinder Smart Bidding by making too granular of a structure.
Instead, your account structure should reflect ROAS targets. I’m very big on not overcomplicating account structures unless I have a specific purpose for doing so.
There can still be value in splitting campaigns in mobile or desktop campaigns, but it shouldn’t be because you want to set specific mobile bids. It should be because you want your mobile ROAS to be different from your desktop ROAS.
With that being said, then you should obviously not just lump all your campaigns into one campaign. Our article on account structure for ecommerce still holds true: The Ideal Google Ads Campaign Structure for eCommerce Accounts: Work Faster & Drive Higher Performance.
Just don’t over-segment it. It’s not needed anymore.
Note: When you get started with Smart Bidding, I suggest you don’t make changes to your account structure. Just enable Smart Bidding instead and verify that it works for your specific case.
Once you’ve done that, you should start reorganizing your Google Ads account. Otherwise, you might be doing a lot of work that you’re not going to use.
Don’t use Smart Bidding for all campaigns
We should again revisit the three principles for when Smart Bidding can be successful:
- Data-volume must be significant
- Stability in your data
- Predictability in the market
A common misconception is that either you’re all-in on Smart Bidding, or you’re not.
It doesn’t have to be like that. You can apply Smart Bidding on campaigns that fit into those three principles, and then let other campaigns run on manual bidding:
- Brand new campaigns
- Low-volume campaigns
- Moving into a high season
- Campaigns for volatile categories
Don’t forget the rest (opt calendar)
The number of people I’ve talked to that haven’t touched their account in 6-12+ months after enabling Smart Bidding is insane.
It’s like people don’t think they ever have to optimize ads, keyword selection, product prioritization, etc. again just because they switch on Smart Bidding.
Use a solid Google Ads calendar to make sure you optimize all the needed areas of your account.
7 Things to Keep In Mind When Working with Smart Bidding
1) Profit-calculations can be key
Profit margins on different products can kill your profitability when you enable Smart Bidding.
You might have manually preferred certain products that you knew you had a great profit margin on when starting out. But when you launch Smart Bidding, it will blindly optimize for a higher ROAS no matter what other factors (like profit margin) you had taken into consideration previously.
We’ve seen a couple cases where Smart Bidding has increased bids for products with great ROAS, but has resulted in a low profit margin for products with a lower ROAS despite a great profit margin. It created a perfect storm of higher revenue, higher ROAS, but when the finance people closed the books a few months later, the business had actually lost money instead of earning more.
The best way to combat this is to use the priorities level for Google Shopping campaigns:
- High priority: Products with high-profit margins
- Medium priority: Products with medium-profit margins (set the target ROAS higher)
Note: You can’t do this with Smart Shopping. You’d have to put your high-profit margin products in Smart Shopping and the rest in regular Shopping campaigns.
2) Google overestimates Smart Bidding’s abilities, but that doesn’t mean it’s all bad
All of Google—from engineers to account representatives—are way too optimistic about Smart Bidding’s ability to improve an account.
I believe it’s one of the reasons why some eCommerce companies haven’t started using it. The aggressiveness to which Google is pushing Smart Bidding on accounts without the slightest regard for nuances is giving people the wrong impression.
My advice is this: Don’t listen to what they say. Make your own tests and see to which extent you can make it work for you.
3) The “don’t let the fox guard your chickens” argument
The argument is that Google is interested in the highest CPCs, so they make the most money.
That should immediately make everyone concerned about what goes on in the algorithm-building offices of Google Ads.
And yes, I’ve seen many cases where the CPCs were not at all aligned with the value, but I’ve just as often seen the opposite—that Google decreases bids too much so that we lose out on revenue.
As mentioned, Google is bad when working with large value-per-conversion numbers. All of the sudden, a keyword can have a 65x ROAS, and with an overall ROAS target of 10x, then it’ll increase this keyword bid from $20 to $50.
But I have never seen anything built into the Smart Bidding algorithm that would make me think that it’s built to increase profits for Google.
Again, I refer to the three principles. I personally believe that one of the problems is that Google’s recommendations for how much data is necessary for Smart Bidding is too low. So they get a lot of low-volume accounts with horrible performance, which drives the negative impression.
4) Smart Bidding for Shopping vs. search campaigns
There isn’t much difference in my recommendations for how to use Smart Bidding for Shopping vs. search campaigns. In my first point about profit calculations, I listed an example of how to use campaign priorities for Shopping to steer Smart Bidding towards bidding higher for certain products rather than others.
That’s pretty effective, but going into more detail would be another 1,000 words and this article is already long enough 🙂
5) Smart Bidding vs. Smart Shopping
This is an interesting conversation.
First, I’d like to highlight that there is a difference between Smart Shopping and Smart Bidding.
Smart Shopping is the complete black box where you can’t see search terms, audiences, control product prioritization, set negatives, etc.
Smart Bidding is just laid over your regular Shopping campaigns to manage your bids.
I’ve gone into depth with this subject in my article on Smart Shopping.
6) You lead; you don’t do the work anymore
I often use the example of a ship vs. a speedboat when comparing manual bid management to automated bid management. But being a “leader” might be a better example.
You no longer have to (rather, get to) make changes to bids anymore. You need to make sure your Smart Bidding portfolios have the right datasets, and then steer it.
It frustrates many that they can’t make a change to a bid when they want to. But you have to trust the machine and review what it’s doing right instead of wanting to micro-manage it.
An example of this can be when you move into a high season, and you need to push a certain category more than it has been so far.
You have two choices with Smart Bidding:
- Decrease the ROAS target for the campaign(s)
- Move them to manual bidding for a couple of weeks
Both actions will increase your exposure, which is your ultimate goal until you start focusing on ROAS again.
7) Bid adjustments no longer have any effect (except for -100%)
Just so you know – you don’t need to set bid adjustments anymore (devices, audiences, and ad schedule). Unless you set a -100% bid adjustment, then Smart Bidding disregards the number you have.
This doesn’t mean you should delete your bid adjustments when you’re testing Smart Bidding, though. Keep them. They don’t hurt anything, and if you decide to rollback to manual, then you’re not starting from scratch again.
The same applies to times where you want to set manual bids. You might forget to apply a decrease bid adjustment to mobile, and then you are spending way too much on mobile until you realize it.
My Aggressive Opinion on Smart Bidding
I believe that 80% of PPC managers are bad at bid management.
15% are good.
5% are really good.
On top of that, a very large percentage of PPC managers are not performing bid management according to widely accepted best practices:
- Make bid changes on the same day of the week
- Use the same date range every time, but include longer/shorter ranges occasionally
- Calculate for lag-conversions (conversions that happen several days after a click)
- Increase/decrease bids based on a set list of percentages
- Override as needed
- Calculate bid adjustments (device, ad schedule, audience) into your keyword bids
It comes as a surprise to many in-house teams when our Google Ads Roadmaps recommend Smart Bidding to their case. But in all instances, my argument is:
You’re not doing a good job managing your bids in the first place. So even if Smart Bidding was only 80% of the way there, it would still do a better job than you could.
And that’s really the main point I have. So many people manage bids poorly that Smart Bidding should be absolute crap not to improve the performance.
Do We Use Smart Bidding in SavvyRevenue?
I feel I have to answer this question as I got a lot of feedback on my other article on automated bid management and Smart Shopping.
The answer is sometimes.
We use what works the best in the particular case, and always according to the three main principles.
We have big accounts in stable markets that are predictable year-round. Smart Bidding performs very well in those cases, and we’d be foolish not to use it. It’s free, it performs well, and we have a framework in place to mitigate the disadvantages.
EXTRA: How to Treat Smart Bidding During Coronavirus Days
As a friend of mine noted the other day when he was asked about Smart Bidding and Coronavirus: For some industries, it’s just like a reverse Black Friday event.
So if you’ve read this article through and through, then you should already know what to do with Smart Bidding during the coronavirus outbreak.
We, again, go back to the principles:
- Data volume
A couple of scenarios can be reviewed in combination with the three principles:
If your data volume has decreased to the point that you are on the verge of not having enough data for Smart Bidding to work effectively, then you should go back to manual bidding.
If your ROAS has decreased, but you want to maintain volume, decrease the ROAS target to maintain the revenue volume.
If ROAS is decreasing and you can’t afford a lower ROAS, then this is where Smart Bidding should come in handy as it will bid down.
In industries that are down by 30-50% you should keep an open eye on Smart Bidding, and based on the three principles, move to manual bidding until the new normal has been found.
You can also apply higher minimum bids and lower maximum bids for Smart Bidding to do what you want it to do manually. Try to think about the mechanisms that are available to you rather than trying to revert back to the “way you’ve always done it” and be upset that you can’t change a bid for a specific keyword.
In Summary, Smart Bidding Can Help You Manage Your Account More Easily, But It’s Not The End All
I think Smart Bidding is a step in the right direction for many advertisers. It allows you to focus on other aspects of your account. If you’ve spent 10 hours on your account per week so far, and half of that time has gone to bid management, then you’ve probably felt pretty good.
Now that we’re taking those actions away, you will suddenly have more time to optimize other parts of your account that you have neglected.
I hate best practices (see why here). I hate advice that people think will work for everybody. And the same goes for Smart Bidding.
Smart Bidding doesn’t work for everybody. But it will work for some. And that might be you. You just need to know what works well and what doesn’t work well with Smart Bidding so you can mitigate the disadvantages.
Otherwise, you might run into a big problem as soon as something goes wrong.