Bid strategies have become one of the cornerstones of a good Pay Per Click (PPC) campaign.
Google has done its best with smart campaigns (and it continues to improve), but a click-and-go solution which is actually effective still belongs in the realms of science-fiction. Therefore, successful bidding still very much relies on how you set up your bidding strategies and campaigns.
Bid strategies are now more malleable, with a growing number of features. This means they’re complicated. Even when Google moves towards simplification, such as consolidating bid strategy types, understanding how to get the most out of them can be tricky.
What are bid strategies?
A bid strategy, also known as smart bidding or automated bidding, sets bids algorithmically for each individual auction based on the targets you input.
There are six main bid strategies:
- Maximise clicks
- Maximise conversions (with optional target Cost Per Acquisition or CPA)
- Maximise conversion value (with optional target Return on Ad Spend or ROAS)
- Target impression share
You can learn more about how to select the right bid strategy with our guide.
Best practice for bid strategies
We’ve got years of experience with bid strategies of all varieties. And we've used them to drive everything from leads to product sales. However, bid strategies are not a black box solution to all your problems. Here are our top tips for how to get the most out of them:
- Don’t use bid strategies if you don’t have enough data
- Get the basics right
- Feed bid strategies as much relevant information as possible
- Set realistic targets
- Set minimum and maximum CPCs
- Don’t change them too frequently
- Adjust them to unique circumstances
- Understand how bid adjustments work within bid strategies
- Keep analysing them
1. Don’t use bid strategies if you don’t have enough data
Different strategies and platforms each have thresholds for how much conversion data they need to optimise, and many platforms will now let you set up bid strategies without any historical conversion data at all. For example, Google Ads now claims to be effective with no historical conversion data at all, but Search Ads 360 (at the time of writing) recommended at least 20 per week in a portfolio.
You should however be cautious of approaching bid strategies without any historical data. All machine learning, by definition, relies on data from which to learn. If it doesn’t have any historical signals around what has led to success, it cannot learn.
Setting strategies live without decent historical conversion data can cause potentially crippling volatility. Seemingly good or poor performance based on a low number of clicks can lead strategies to push too much or too little. For example, two conversions from two clicks.
Bid strategies are getting better at this, but it does make you reliant on your competitors’ conversion settings and rates as bids will largely run from these signals rather than your own data.
For example, if your product targets a unique niche or conversion action, your conversion rates are likely to be completely different, even on similar keywords. If this is the case, we’d recommend building your own conversion data before launching.
Instead, think about optimising your manual bidding instead. Test bidding formulae and bid adjustments. For high value, low volume industries, like legal or automotive, look at aggregating your data to give you applicable insights:
- Change bids based on how a particular group of keywords is doing
- Adjust to strong periods of the week rather than specific hours. For example, think “weekday mornings” rather than “Monday 9-10am”.
- Analyse how your audiences have been doing and set adjustments here. Build similar audiences off the top performers to give yourself more data points to optimise from.
Lastly, be wary of running with enhanced CPC as some kind of middle ground. In Google Ads, “Enhanced CPC” means that you can set max CPCss to benchmark your bids, but it adjusts this based on the chances of converting. It sounds great, right? However, I’ve never seen this perform effectively, and certainly not as effectively as manual bidding. There’s a simple reason for this: If you don’t have enough first party data to justify bid strategies, you likely don't have enough for enhanced CPC either.
2. Get the basics right
Bid strategies can’t fix a broken account.
If you haven’t got the fundamentals right, bid strategies will struggle. Without cross-match negatives, for example, bid strategies will get stuck in an endless cycle of raising and lowering bids. Searches will leap from keyword to keyword, and no actual learning. Here is a simplified diagram of the cycle of poor performance:
Likewise, make sure your tracking is accurate and meaningful. In particular, make sure you’re obeying the following guidelines
a) The conversion events you are optimising for are all meaningful
I’ve often seen businesses unintentionally optimising to page views as a conversion. This gives very little idea of user value
b) If you have many conversion events, tell the system which one you value more
Unless you tell the bidding platform otherwise, it will value all conversions equally. You would not value conversions and newsletter subscriptions equally. So, tell the bid strategy this. You could apply a mock conversion value.. For example, setting that membership is five times more valuable to you than subscriptions
c) Check your conversion events
Make sure the conversion actions are still tracking what you intend them to. Check that you have a reasonable conversion window set-up. Make sure that they’re actually still tracking. The last thing you want is for Google to think your ads have stopped converting when, in fact, a developer has just knocked your tag off the website.
Machine-learning is only as effective as the data going in allows it to be. If the dataset is too small or flawed, the results will be similarly grubby.
3. Feed bid strategies as much relevant information as possible
Google has a plethora of real-time information about each searcher, but the more signals from our account set-up we can give the algorithm to work with, the more we can help it learn relevant patterns to bid towards.
Add first party audiences that have been thoughtfully segmented based on behaviour, along with similar audiences of users who’ve taken key action, and group keywords by intent as well as by semantics. Make sure what you’re telling Google to look out for aligns with your ultimate goals. For example, is your conversion window aligned with your actual customer journey, and does your chosen attribution model align with how customers interact with your business?
If you have low conversion volumes, you could even deploy bid strategies optimising towards a higher section of the funnel to ensure Google has enough data. This is a strategy that has paid dividends for many of our clients here at Ayima. The key is selecting an action that’s still highly indicative of ultimate intent to purchase. A rough rule of thumb is that you want to identify the top 1-3% of your users with this conversion action.
For example, optimising to the number of people adding items to basket, or engaging with a particular part of the website. At Ayima, we’ve been developing ever more granular ways of capturing these identifiers, but which don’t quite fall within the scope of this guide.
If conversion volumes differ by campaign, try using upper-funnel actions for just the low-volume campaigns, which are likely to be customers sitting higher up in the funnel anyway. Deploy conversion action sets to truly optimise towards what matters for each campaign.
4. Set realistic targets
Bid strategies are not miracle workers. No matter what platform you use, target CPA and target ROAS strategies can’t work towards a target that is miles away from historical performance.
While platforms differ, you generally want to stick within 30% of performance over the past one to two weeks. If bid strategies are always struggling to spend or they fall below target every week, move the target closer to actual performance.
For example, if your current ROAS target is 500% and you're only hitting 250% on a good week, lowering the target to 300% can lead to your actual ROAS rising.
This is because bid strategies get more of a grip. It would be like driving a car and shifting straight from 2nd to 5th gear. The car is not going to gain enough traction. Similarly, it can cause your campaigns to stall.
For example in the above strategy, be wary of changing your target ROAS to above 279% (215*1.3 = 279.5).
Bid strategies are getting more robust at dealing with this, but there are still limits.
5. Set minimum and maximum CPCs
Setting limits gives you a degree of control in your strategies.
You should set realistic minimum CPCs to help ensure that your keywords never fade into nothingness. Minimum bounds give them a chance to perform. Without this, a bid strategy might be pushing a keyword to £0.02 bids. This means they’ll never show, and the platforms never see if it’s performance has improved.
The algorithms on many platforms now periodically push bids on each keyword to test how they perform. Max CPCs restrict Google’s ability to get carried away with your bids. Google isn’t a non-profit, and at the end of the day they do like to make money!
6. Don’t change them too frequently
Bid strategies work best when they’re allowed to build a consistent picture of performance. If you make constant changes, the strategies will be constantly re-assessing, and never get the chance to settle on one of them.
Most platforms need about a week to learn after major changes to the targets. They will also re-enter a learning phase with major changes to the portfolio such as the keywords within a search campaign, or the products within a shopping campaign.
Most tell you when they’re in a learning phase. Don’t keep on changing the performance target, no matter how tempting that is. It’s machine learning: give it enough time to learn.
7. Adjust them to unique circumstances
Bid strategies are getting better. But, they still aren’t fantastic at dealing with seasonality that is unique to your business. They are still mostly a reactive rather than a predictive tool.
If you’re having a sale, set seasonality adjustments. These are available within most bidding platforms. These give bid strategies advanced warning of when you expect conversion rates to change.
There is no need to set these up for broader sales events with most bid strategies. Google Ads, for example, adjusts bids globally for events such as Black Friday.
If your bidding provider does not have this feature, use your targets to inform the bid strategy where to go. For example, if you expect conversion rates to rise temporarily, you can give the bid strategies a boost by telling them to aim for a 30% lower ROAS target. The bid strategy will then raise bids preemptively, right as you hit the sale.
The same applies if you have a conversion blackout, or if your site has gone down. The bid strategies will see this as your conversion rate plummeting, and they will want to lower bids. Most platforms (including Google Ads) offer blacklisting events. This makes the bid strategies exclude certain time frames when learning.
8. Understand how bid adjustments work within bid strategies
When you launch bid strategies, your current modifiers will change roles. This is particularly key when it comes to audiences.
Rather than adjusting the actual bid for an audience, bid strategies use audience modifiers to signal how important an audience is to you. This helps them to define the audiences.
Users will fall into whichever audience they’re a part of with the highest bid modifier. If a user fits into two audiences, for example:
- An in-market audience for home furnishings - (+30% bid adjustment)
- A retargeting audience aimed at converters in the last 7 days - (+25% adjustment)
In the above case, the user will come in through the in-market audience because you have given it the higher adjustment.
9. Keep analysing them
Like all Artificial Intelligence (AI), the best way to see the inner workings of bid strategies is to see what results they produce.
You can test different modes of bid strategies using campaign experiments. Test them against other bid strategies, test different ROAS targets. Bid strategies are constantly changing, the best today may not be the best tomorrow.
Bid strategies have the capability to drive great success in accounts. But only if you do them well. And, they rely on you having a best practice account with solid foundations.
Like all automations, they’re only as good as the data they use, and the account structure they sit in.
They are certainly not a one-size fits all solution to your Pay Per Click problems, but they can be a great weapon. Use bid strategies responsibly.
If you have any questions or would like to learn more about how we manage bid strategies, we'd love to hear from you.