Some ‘not so obvious’ advanced test and learn ideas to help you optimise your PPC campaigns and maximise your return on ad spend.
For most PPC advertisers, the bulk of A/B testing focuses on ads and landing pages. After all, clicks and conversions are two of the most important metrics for any paid advertising campaign.
However, there are plenty of other factors that determine how often your ads are seen and how relevant these impressions are – both of which have a direct influence on the number of clicks and conversions your campaigns achieve.
In this article, we’ve got five PPC optimisation ideas. Instead of running basic A/B tests, you can apply a test and learn model that collects data from a series of variations and pinpoints the best settings for your campaigns.
Test and learn is a data-driven model that runs a series of variables within a single test to compile performance data and “learn” which combination of settings/variables is most effective.
For example, you might apply a test and learn model to ad scheduling so you can find out which times of day, days of the week, weeks of the month and months of the year your campaigns perform best.
So while A/B testing is great for finding out which ad copy or landing page design is most effective, test and learn is ideal for finding out which campaign settings get the best results. You can also apply test and learn models to external factors – for example, to find out how market trends affect campaign performance.
This data-driven model can answer complex questions about your paid advertising strategy and remove a lot of the mystery that surrounds PPC advertising. Let’s explain more by looking at specific examples.
We’ve talked about maximising search coverage before and why it’s so important for visibility, clicks and edging out the competition. One strategy Google recommends for this is combining broad match keywords with Dynamic Search Ads (DSAs).
Basically, broad match keywords increase the variety of search terms your ads will show for while Dynamic Search Ads expand this further by using the content on your website to match users’ queries. So you get the benefit of increased search coverage from broad match keywords but without the loss of relevance.
The added benefit of DSAs is that this is a fully-automated ad format but, as always, make sure you keep an eye on performance.
Now, the test and learn aspect of this strategy focuses on keywords – not only the broad match terms you’re using but also the negative keywords you use to ensure your ads only show for valuable search terms. This requires a careful balancing act when you’re trying to maximise search coverage without allowing relevance to suffer. So there is room for some trial and error here and this is where the test and learn system thrives.
Another difficult balancing act in PPC is maximising conversions while managing cost-per-acquisition (CPAs). One of the most effective ways of doing this is to optimise your landing pages to increase conversion rates so that you’re converting a higher percentage of users from the same total ad spend.
However, there are strategies you can employ at the campaign level to increase conversion rates without CPAs.
Essentially, it all comes down to increasing the quality of traffic coming to your website. So we’re kind of taking the opposite approach to maximising search coverage and focusing more on attracting visitors who are most likely to convert.
One of the most effective ways of achieving this is to test ad scheduling settings to find out when visitors are most likely to convert. You’ll often find traffic generated early on a Monday morning is less likely to convert in the near future (but they may well go on to convert at a later date) whereas traffic generated on Thursday and Friday evenings often complete the purchase right away.
Once you have these insights, you can dedicate more of your ad spend to the times where conversions are highest, increase the quality of the traffic landing on your websites and turn more visitors into paying customers while decreasing your overall CPAs.
The trick is testing enough ad scheduling variations, for long enough, to collect the necessary data for accurate forecasting. Once you’ve got enough data, you can predict the likelihood of conversion by the hour, day, month or time of year.
Finding the best targeting options to pinpoint your ideal audience requires a lot of experimentation. In the last section, we looked at just one of the targeting options at your disposal in Google Ads and it can take months to collect enough data to start forecasting with reliability.
With a platform like Facebook Advertising, you’ve got thousands of potential targeting combinations to work with. Take a look at our guide to Facebook and Instagram’s immense targeting options for more info.
The better you know your audience, the more likely you are to choose the right combination of targeting options – at least, in theory. The problem is assumptions don’t always materialise in the wild and consumer habits change. A home insurance company might assume new homebuyers are their most profitable audience when they would actually get better results by targeting couples expecting their first child.
To find out which targeting settings you should really prioritise, you have to test campaigns and settings at scale. Not only will this help you define the most profitable audiences, it also allows you to compile historical data to map out trends in your industry so you can switch to more profitable audiences as they emerge.
Earlier, we talked about scheduling bids to improve the quality of traffic (conversion likelihood) coming from your PPC ads. But this is just one of many ways you can optimise your bids to maximise performance in Google Ads.
Much like finding the best targeting options, a test and learn approach can help you find the ideal bid settings. Better yet, you can automate this entire process so that your bids are always adapting to user behaviour and maximising performance.
By using historical performance data, you can predict which devices, locations, times of day and targeting settings are most effective at any given time. You can then feed this data into an automated script for adjusting your bids, which increases your ad spend when the right criteria is met.
This means you’re spending less on keywords and campaign settings when performance is lower and upping your bids when performance is highest – all without any manual input.
You can find out more about how we do this at Vertical Leap by reading our case study on automated bidding adjustments.
As mentioned earlier, campaign performance isn’t only determined by internal factors. Real-world issues like global pandemics, economic outlooks, national holidays and the weather influence consumer behaviour and brands need to respond to these changes.
Thankfully, it’s never been easier for businesses to collect relevant data from external sources using APIs and other integrations. Now, you can pull in data from platforms like Google Trends to track Covid-19 consumer habits or weather forecasts to predict sales volumes for the upcoming weekend.
Other seasonal trends are less unpredictable than health crises and weather events, too. For example, you can compile sales figures, campaign performance and other KPIs throughout the year to understand how business performance changes throughout the year. If you have enough historical data, you can map this against influential events, such as economic downturns, political landmarks like Brexit and general elections, or even sporting events and royal weddings to understand how external factors impact your business.
The more data you have, the more effectively you can react to seasonal trends and external events. This knowledge will help you adjust campaign settings, re-evaluate ad spend and even predict positive/negative impacts before they actually materialise.
While A/B testing your ads and landing pages is crucial for improving PPC results, it’s not enough to really squeeze the best performance out of your campaigns and beat your strongest competitors.
To refine the finer details of your campaign settings and maximise the quality of traffic vs your ad spend, you have to apply a test and learn approach to your accounts. We’ve looked at five specific examples of how you can use this approach to improve performance in this article but there’s so much more you can do with test and learn PPC.
If you want to find out more about this data-driven strategy, you can speak to our PPC and data scientists by calling 02392 830281 or emailing email@example.com.
We send a semi-regular newsletter on business and other related topics, with links to the latest stories from us and what we’re reading around the web.
Categories: Data Science
Categories: Events, PPC, Tutorials