We all know that automation is key to achieving great performance in marketing. Give the machine the data and it’ll reward you with the results you want.
But how much data is enough for algorithms to work? This is a question we get all the time, but it isn’t clear cut. Media platforms provide little to no guidance on minimum data requirements, so we’ve spent some time analysing various data sets from a variety of clients to provide this handy guide for you.
So, to begin, what do we mean by data? In the world of performance marketing, we mean signals of intent that a platform (Google/Meta etc) can use to understand if traffic being sent is valuable. Does the traffic convert, and if it does find more traffic like that.
In the good old days, we didn’t need to worry about this. We just focussed on the keyword being searched or the audience a user was in, set a bid and off we went. We’d review performance and adjust the bid or targeting accordingly. Simple times.
This doesn’t work in modern performance marketing. Media platforms work off millions of data points that users give them, and the more data we can provide on valuable traffic the stronger performance can be.
Okay, so we know we need to tell platforms what good performance looks like with data, but how much data do they need?
| Platform | Minimum number of conversions for automation | Ideal number of conversions for automation | Time period automation uses |
| Google Ads | 15 | 75+ | 30 days, with emphasis on recency |
| Microsoft Ads (Bing) | 15 | 75+ | 30 days, with emphasis on recency |
| Meta Ads | Unknown | 50+ | Unknown, but assume 14 days |
| TikTok Ads | Unknown | 50+ | Unknown, but assume 14 days |
| LinkedIn Ads | Unknown | 50+ | Unknown, but assume 14 days |
Google is pretty good at guiding us on the minimum amount of data required (reading on this here), but extensive testing tells us the minimum really isn’t enough to achieve strong performance. Social media on the other hand copes better with less data in comparison with Google.
Note this is accurate at the time of writing, but media platforms update their technology faster than we do prime ministers, so it could be out of date. Always check with your media provider before making strategic decisions.
What’s the risk in applying automation without enough data?
In PPC you could be exited from the auction because the platform won’t believe it can achieve your goal, and your traffic will decline. In some cases we’ve seen platforms bid highly aggressively to test traffic, resulting in CPC’s above £80. Be wary of this and always monitor performance closely.
On Google, can’t I use the ‘maximise conversions’ option, it says there’s no minimum conversion number required?
We wouldn’t advise this unless you reach the minimum thresholds outlined above. Google still needs data to work to, and whilst it can bid to get the ‘maximum’ number of conversions possible for your budget, it’s likely to cost you. If your conversion action is fairly easy to achieve this is ok, as you’ll gain data quickly and the algorithm can learn effectively. However, if you’re only likely to get a few conversions per day this is a risky option. Your CPC’s could go through the roof with little to no return, proceed with caution.
What should I do if I don’t have enough conversion data to use automation?
You have two options here:
- Provide the media platform with a conversion data point higher up the funnel to give more data. This can be called a micro conversion. For example, if your goal is to drive sales but you only make 10 sales a week, you could create a micro conversion for when users view specific product information or add to cart. This would provide much more data that you could bid towards.
- Go manual. There is still always the option to bid manually. This is not ideal but a sensible stop gap if you can’t provide enough conversion data. Ensure you check your bids are appropriate regularly, and update as needed.
Both options are a great alternative to using automation to bid to your macro conversion, but ensure you keep an eye on things and move over to the macro when you have sufficient data.
We hope this blog was useful in navigating the world of data and automation in performance marketing. If you have any questions or would like advice, please drop us a line.
