The Ultimate Guide to Google Smart Bidding for Display Campaigns

PLUS Smart Bidding Face-Off: “Maximise Conversions” vs “Target CPA” – The Blueclaw Experiment & Travel Sector Results

To encourage more extensive – and effective – use of Google Ads, Google has been continually adding Smart Bidding options for ad management to automate certain functions, reduce complexity and (ideally) improve ROI.

To de-mystify the selection of smart bidding options available in Google Ads, we decided to put them to the test.

Whether you’re an experienced Google Ads user managing accounts that are already doing well, or a marketer totally new to Google Ads, it’s important to understand how Smart Bidding can help – and what its shortcomings might be.

Working within the highly competitive travel sector, our experiment focused on the ‘Maximise Conversions’ and ‘Target CPA’ goals for a Display campaign.

But before we start, let’s establish what smart bidding is and why it’s taking off.

What is Smart Bidding?

Smart Bidding is a subset of automated bidding strategies that optimise a Google Ads goal set by you, using thousands of real-time bidding signals.

Because of this, all existing bid modifiers are ignored except for 100% negative bid adjustments (device exclusions, audience exclusions etc).

Essentially, Google is trying to accomplish the mission you’ve set for it.

Why Change to Incorporate Smart Bidding

Smart Bidding frees up time.

According to Google, as of 2019, about 80% of digital marketer’s time is spent on manual tasks like bidding while only 20% is spent on strategy.

Smart Bidding allows (in principle) for more intelligent working by automating routine tasks.

Of course, this idea isn’t totally new, as custom scripts have been used by PPC managers to automate different aspects of Google Ads.

The key reason why Smart Bidding is a definite step forward in the evolutionary chain is not just due to it being the far smarter cousin to scripts – they make automation more accessible than ever before.

With the support of Smart Bidding, PPC managers and marketers can eliminate (if they choose) the need to optimise each campaign for a specific audience, device, demographic, location etc – and the list goes on!

Smart Bidding scours and applies a huge range of user and environmental signals.

With the user journey becoming more complex than ever, the complexity of bidding is on the rise too – particularly in fast-moving sectors like travel and e-commerce where a myriad of different customer audiences search in different ways about a rapidly changing inventory of experiences, offers and products.

Contextual signals considered by Smart Bidding include everything from time of day, geolocation, device, user demographics, interests and even prior user behaviour.

These signals are used to predict conversion likelihood and subsequently, a bid will be made which mirrors this likelihood to maximise your chances of a profitable conversion.

For example, if Google knows that a user aged 18-30 is 1.5x more likely to book a flight to Ibiza than any other age group, it will bid 1.5x higher for users in this demographic.

Although we might hate to admit it, because smart bidding can analyse 70 million signals in 100 milliseconds, its computing capabilities are far superior to even the smartest human brain.

Essentially, you’re getting more value for less time spent – if applied correctly.

Smart Bidding decides bids in real time.

Auction-time bidding can tailor bids for every single auction. This level of precision and efficiency will likely drive better results than one set, manual bid.

Smart Bidding uses adaptive learning at the query level.

Rather than optimising at the keyword level as manual bidding strategies do, Smart Bidding can optimise for conversion rate across different queries.

For example, if a user’s search query for ‘pink Nike trainers’ is triggered by the broad match modified keyword ‘+Nike +trainers’ in your account, Google will remember not just the average conversion likelihood for the keyword in your account, but also conversion likelihood for the query of ‘pink Nike trainers’.

Deeper, faster algorithm learning.

If a search query has already been matching with other parts of your campaigns, the algorithms and conversion data carry over into the rest of the account.

Types of Smart Bidding Strategies

There are a variety of Smart Bidding strategies you can apply. By setting a certain goal, Google will then adjust bids to meet the outcome that you’ve specified.

Enhanced Cost Per Click (CPC).

GOAL: Although Enhanced CPC uses machine learning, it’s only a semi-automated strategy and is a hybrid of smart bidding and manual bidding. It works by automatically adjusting bids for clicks that are likely to secure conversions.

For example, for a US-based travel company, users in California are most likely to purchase a holiday to Cuba, while users in North Carolina are least likely to do so.

Therefore eCPC automatically up weights manual bids for users in California to match the increased conversion likelihood for this browser location, and down weights bids for searches from South Carolina.

Target Cost Per Acquisition (CPA).

Unlike Enhanced CPC, this bidding strategy is fully automated - as are all the strategies detailed below - so experiment with caution!

GOAL: Target CPA Smart Bidding works by aiming to get as many conversions as possible while staying within the target CPA.

On average, the CPA for Display campaigns will be closer to the target than for Search campaigns, as Google can apply acquisition data more rapidly to the campaign.

CONDITION: Google recommends that advertisers have at least 30 conversions in the past 30 days before using the Target CPA.

BENEFIT: Google has identified a typical 31% uplift in conversions by using Target CPA Smart Bidding. It is particularly well suited to e-commerce businesses where manual optimisation over a wide range of products isn’t practical, and competitors may be adjusting their own pricing to be more competitive.

WARNING: The cost-per-conversion (CPA) will sometimes go above the Target CPA that you’ve set, particularly at the start when the machine is still in its learning phase, so you must take this into account when budgeting.

Also, the Target CPA may not work effectively if your campaign is budget restrained. This is because it will aim to meet your target CPA over the course of the whole day, so if the budget runs out at midday then the average CPA may be way above the figure set.

This can be managed by being conservative with the CPA figure you set and building up campaigns so that Google has more data to optimise with greater accuracy.

Maximise Conversions.

GOAL: As one of the newer additions to the Smart Bidding set, it aims to get as many conversions as possible, without exceeding your daily budget.

This approach differs from Target CPA as the primary goal is the number of conversions, with the CPA as a secondary consideration.

This approach is thereby best used by acquisition-focused businesses who can anticipate a longer-term customer value as they build market share.

CONDITION: This bidding strategy will aim to spend as much of your daily budget as possible, so make sure you can afford to spend the amount that your daily spend cap is set at.

BENEFIT: Helps you to spend your budget as efficiently as you can and works best for budget-constrained campaigns. If your focus is on conversion volume and you’re happy not to carry-over budget to the next day, this approach is definitely one to adopt.

WARNING: This strategy is not suited to campaigns with a strict CPA or Return On Ad Spend goal as the systems will prioritise the number of conversions – rather than the profitability of those conversions – within your ad spend for the day.

This strategy would be particularly useful for industries that experience peak trading periods, for example, fashion e-commerce sites around Black Friday where the conversion rate is extremely high and there is intense competition.

Maximise Clicks.

GOAL: Choosing to Maximise Clicks will, as the title suggests, aim to get as many clicks as possible within your budget.

Choosing to have Google focus on clicks (rather than conversions) means that Google will be focusing just on this area of the customer journey to drive traffic.

BENEFIT: This is a great option for those wanting to drive more traffic to their site when conversion volume is already strong.

This strategy would also work well alongside other time-restricted marketing campaigns to support offline channels such as TV advertisements or billboard campaigns if you have considerable budget to spend.

WARNING: If a maximum bid limit is not set, your CPC bid will be adjusted to gain as many clicks as possible - certainly not recommended unless you’ve got budget to blow!

Target Return on Ad Spend (ROAS).

GOAL: To achieve higher conversions without exceeding the target you’ve set for return on ad spend.

CONDITION: For Google to effectively apply ROAS data, your campaign must have at least 15 conversions in the past 30 days.

Also, before you can apply Target ROAS to campaigns you’ll need to set values for the conversions that you’re tracking – if Google doesn’t know the revenue generated from your conversions, it can’t calculate ROAS.

BENEFIT: Google claims that advertisers see an average uplift of 35% in conversions when using Target ROAS on Display Ads - the biggest uplift of all Smart Bidding strategies.

More than this, it is also the only strategy that factors in the predicted conversion value of a user when deciding how much to bid – and calculating bids based on accurate financial value is arguably the most meaningful way to optimise PPC.

Shopping campaigns work well using Target ROAS as product prices can be accessed through Google Merchant Center – helping to join up data and complementary channels.

Pitfalls of Smart Bidding: What to Watch Out For

Selecting the wrong strategy will drive less than impressive results.

Picking a smart bidding strategy that fits with your goals is crucial to success. As with any manual bidding strategy, you’ll need to have conversion tracking in place to measure the KPIs you want to optimise towards.

A reasonable level of conversion data is needed for smart bidding to work effectively.

Although Google states that you don’t need historical conversion data for most smart bidding strategies, in our experience a solid number of conversions (at least 30 in the last 30 days) is needed for campaigns to succeed.

Less control bidding on specific keywords.

You can’t selectively choose to bid higher on keywords that are more critical within the account as you can with manual bidding. If certain keywords have more focused user intent, you may want to ensure that those appear higher up the SERP than weaker keywords.

Less control bidding on keyword match types.

Within a campaign using manual bidding, a stacked bidding strategy can be used to ensure that higher bids are made for Exact match keywords over Broad Match. With smart bidding, however, you can’t ensure this. Since you can’t edit bids, you can’t tier match types.

To help eliminate these drawbacks, there are some basic guidelines that we recommend:

• Avoid making big changes to the experiment once it is up and running. Making changes means that the results won’t be a true reflection of either condition you set. By altering your Target CPA too much, you’ll likely also bring the spend right down.
• Leave the experiment running for at least 30 days. This gives the algorithm enough time to gather the performance data it needs to optimise bids.
• Be patient. It’s unrealistic to expect performance within the learning period to be phenomenal straight off the bat. And when analysing the data, it’s worth reviewing performance both including and excluding the learning period.

Smart Bidding Face-Off: “Maximise Conversions” vs “Target CPA” – The Blueclaw Experiment & Results

The account that we used to test ‘Maximise Conversions’ Vs. ‘Target CPA’ belongs to a global travel & group tours company.

We set up a base display campaign that used ‘Target CPA’ with a target of A$35. We then drafted a campaign experiment with the bid model set to maximise conversions as the testing variable.

We then left the campaigns running for 30 days - this is the minimum length of time Google recommends for an experiment.

Due to monthly budget changes set by the client, 15 days into the experiment we had to halve the daily budget for the Display campaign from $150 to $75.

Ideally, we would have kept all variables constant, but budget changes are common real-world occurrences so are worth testing in conjunction with smart bidding strategies.

Original Display Maximise: Conversions.

Display Experiment: Target CPA.

Despite producing vastly different results (which we’ll get onto shortly), you can see that clicks and conversions followed a very similar pattern in both campaigns.

In a nutshell, as a Smart Bidding methodology, Target CPA outperformed Maximise Conversions in terms of both conversion rate and click-through-rate (CTR).

The Target CPA campaign achieved a 100% higher conversion rate and 300% higher CTR.

The likely reason for this is that Target CPA tends to be more reserved with the auctions that it enters as it aims to stay below the specified CPA, whereas Maximise Conversions will try and get as many conversions as possible for your budget but doesn’t have a restrictive goal to meet.

“Maximise Conversions” had a greater share of impressions.

The reasoning that Target CPA is a more reserved bidding strategy is reflected in the percentage split of impressions between the two campaigns.

Despite being set a 50% even split of budget, only 33% of total impressions came from the Target CPA campaign, whereas 77% came from the Maximise Conversions campaign.

When splitting campaign budget evenly in an experiment, you would expect to see impression share be split evenly too - or at least that’s what we expected to see!

However, this didn’t happen, nor did each Display campaign spend the same amount. It’s not 100% clear why.

Speaking directly with Google, the feedback suggested was that the Smart Bidding settings applied may have overridden the experiment settings - although a definitive answer wasn’t provided. Then again, would Google be Google if there wasn’t some mystery involved?

“Maximise Conversions” had a lower share of conversions and a higher CPA.

As the Target CPA campaign was seemingly more selective with which auctions it entered, the average CPA for the Target CPA campaign was 63% lower than the Maximise Conversions campaign.

The Target CPA campaign did not meet its target CPA of $35. However, the campaign is still relatively new and was in the learning period for most of the experiment duration.

Even so, it wasn’t far off target, producing 43 conversions at a $45 CPA, especially when compared to the Maximise Conversions campaign which only produced 21 conversions at $121 CPA.

The Final Takeaway.

For our Travel client, Target CPA was the clear winner with a 100% higher conversion rate at a 63% lower CPA.

Perhaps the high-value nature of purchases in this industry worked in favour of Target CPA’s more reserved bidding strategy.

How Blueclaw Can Help Your Business

Machine learning is here to stay - and it’s not just taking over the search industry, it’s being used for everything from data protection to financial trading to marketing personalisation.

50% of Google Ad spend is already through Smart Bidding - and if search giant Google says it’s happening, it’s happening, so the more experience you have with it the better.

As this blog piece demonstrates, we are always seeking to optimise our clients’ paid media activity to deliver the best results possible.

Optimising accounts isn’t always about following “best practice” as dictated by Google or Facebook.

We pride ourselves on being pragmatic and objective – and this enables us to exceed our client’s KPIs by optimising towards what’s best for their businesses.

If you’d like to get more from your advertising spend, get in touch online or call 01132343300 to speak to Blueclaw’s Paid Search team.