Written by George Crosby, Director - Paid Search

This is no doubt a period of rapid change and advancement within digital marketing. The recent landmark announcements of AI Overviews & AI Mode are front of mind across the industry. However, AI has existed in the world of Paid Search for a long time. Use of automated bidding is a fundamental part of Best Practice and the ability to customize it remains a critical lever for optimisation. Whether it’s the conversion goals we target, the first-party data we ingest; or the settings we establish within bid strategies, each element plats a vital role. These all have the potential to differentiate an advertiser’s approach from their competitive set, particularly when other levers are being eclipsed by the rollout of broader AI capabilities. 

The importance of bid strategy targets 

A specific feature to focus on though is the humble bid strategy target, as these are still crucially important in directionally driving activity to meet the performance goals we require. There haven’t been many recent big releases from Google on how targets work, but in the backend, there has been a significant shift in the way we change these to meet our advertising goals over the last few years, and this is having a profound impact on the way we might optimise today. 

What’s the history? Haven’t we always been able to do what we want? 

In the pre-Auction Time Bidding era, when intra-day adjustments were needed to steer performance, the standard  guidance was to only change the target by increments of +/- 20%. This would prevent strategies from re-entering a learning period and keep performance smooth. Auction Time Bidding entered the scene, and this guidance continued loosely in the industry, although Google’s official advice around it had changed. +/- 30% became the new threshold, and the guidance around learning periods became somewhat fuzzy. Then, quietly in the background over a number of years, those rules changed. The parameter for changes became “whatever you need”, and learning periods didn’t really exist anymore. The frequency at which you could change the target was also somewhat blurry, with “as little as possible” being the prevailing direction, although this hasn’t particularly updated over the years. 

What does this extra freedom buy us? 

This allows advertisers to more easily pivot to their actual desired performance level with more agility. Gone are the days of ramping up bid strategies in anticipation of a significant peak in spend, as not only are strategies more equipped to deal with this in their considered BAU range, but advertisers are now able to jump their target to the desired point and see an impact almost immediately.  

How does this benefit us in the long run? 

Another benefit of this targeting flexibility is slightly less well known. It’s not necessarily bid strategies’ use of AI, but the enabling of other areas of AI that makes this so critical. One of, if not the, greatest blocker to consistent performance is the “Limited by Budget” status on campaigns. When this label is active, your campaigns are limited in the backend for when they can serve and the AI is not able to ingest the full range of data as it should. This has a profound impact on the decisioning it then goes onto make. Clearing this label should be an absolute priority. You might be thinking at this point, “Why does my target matter here, it’s surely just a matter of budget and if I don’t have enough, I don’t have enough”.  

This is a popular misconception, as the target set in your bid strategy controls the level of aggression with which it seeks out clicks to achieve the desired outcome. If, in a Target ROAS strategy, you have a low target and a low campaign level budget, the bid strategy is going to push hard to pull in volume but then encounter the cap very early in the day and stop serving.  

Instead, what we should be doing is moving our target to better accommodate the demand in the market. If we increase the Target ROAS target, this will mean that we don’t serve with the same level of aggression, but we are able to stay in the auction for longer in the day and therefore gather more data to enable better decisioning, which will go on to improve performance and reduce fluctuations. 

Google’s advice here if you have a bid strategy with “Limited by Budget” campaigns, is to increase your ROAS target up to the level of efficiency it’s been hitting over the last 30 days. This could be a minor shift, or it could be significant, and could also depend on how capped your campaign budgets are. In doing so though, it should drop your Impression Share Lost to Budget down to 0 and mean that you are present in far more auctions. With this extra data, it can learn faster and be more accurate in predictions to steady performance. 

Does this work in practice? 

We tested this approach with one of our clients in the retail industry. Within their bid strategies, many campaigns were limited by budget, and there was a sizeable difference between the ROAS we were asking the bid strategy to hit, and the ROAS it was actually achieving. We made 2 rounds of large target adjustments to 4 of our bid strategies, increasing the ROAS we were asking for up to the level of efficiency currently being achieved.  

What we saw after comparing two full conversion cycles to the previous period was a flat conversion number, but a 59% higher ROAS. Cost had decreased 36%, and CVR had increased 15%. 

What became clear is that allowing the bid strategies to serve more consistently and for longer in the day allowed them to gather more data, make smarter decisions, and pull in the same number of conversions as before but far more efficiently. This then starts a conversation about how to reinvest that saved budget into other areas, or push the volume in this area more effectively, now that the bid strategy has more direction. 

There’s no denying that budget is still an important factor in delivering against a target. However, if your hands are tied, setting bid strategy targets with this approach allows advertisers to see great performance and more effectively understand and adapt to the level of demand in their market. Importantly, we can also do so with a much greater degree of agility than was the case a few years ago. If you haven’t already, it’s worth experimenting with your targets to see if you can generate similarly efficient performance. 

We’ll be exploring this topic in more detail in a series of articles which will be available through our website here: https://www.dentsu.com/uk/en/blog