How I build successful mediabuying teams
Through the past years, I happened to work with numerous people on online mediabuying and advertising, some juniors and some pretty senior. My ultimate goal was trying to help them to make their campaigns successful. What I’ve learned by experience was that different people usually have different strategies in their mediabuying and my role as a leader / coach is to help them find their winning strategy, rather than dictating my own strategy to them.
However, having different strategies does not mean that there is no framework. In fact, mediabuying has a solid framework and it is important to follow it rules whethere you are mediabuying on Google Adwords, Facebook Ad or any other DSP and adnetwork. Below I try to explain the rules of this framework.
1. Define your campaign Goals
The fundamental rule is to set your campaign goals. You should define what determines success in your campaign, and what variables you should compare it with. You should know why you are advertising, what you want to reach and at what cost.
Here are the things to do:
- Define a user action as your goal: In other words, what is considered as a conversion for you. It can be a product purchase, website registration, mobile app installation, product add to cart, etc. This will be your conversion and all your calculation and measurement will be done based on this. If you have flexibility on choosing your conversion, try to be creative and test different things. For example, if you don’t get so many sales on your product or your sales price is very high, having the sales as the conversion will make your optimization process very difficult.
- Try to add a conversion pixel to the your conversion page. Most of the advertising platform have this functionality and it will save you a lot of time to check stats.
- Set a target CPA (cost per acquisition) for you conversion. You should measure the value of a conversion.
2. Plan your media and campaigns structure
Here is a part that many mediabuyers even the experienced ones miss. You should always plan your media and structure the campaign setup. In the article, [Media planning article], I tried to explain media planning in details.
In this article, I have assumed that you’ve already chosen the advertising platform that you want to use. So I am not going through it. Before planning your structure, you should know about the followings:
- The campaign setting and targeting options in your ad platform. Most of ad platform are providing the followings:
- Cost Model: CPC or CPM or OCPM (optimized CPM)
- Geo / Country
- Device type: (Desktop, Tablet, Mobile)
- Operating System
- Internet Service Provider
- Site targeting (Run on network or specific sites)
- Ad position (Skyscrapper, Header, Footer, etc)
- Time targeting
- Others (like interests)
- Your Target Group. Try to take a broad target group, but stick to your campaigns specifications.
Once you have this information, try to structure your campaigns in a way that you would be able to cover most of the areas of your target group. Don’t create very niche campaigns in the first place. It is better to keep the campaign a bit broad, and as soon as you have enough data you can go niche. But also don’t make them too broad that your campaigns would get too messy and it would take you a long time and big budget to find the good performing setup.
I usually start with creating campaigns based on Geo + cost model + Device + Group of site targeting / RON + Demographics (depends on my target group).
A good example is that I never start my first campaigns using time targeting unless it is a requirement by the advertiser. Time targeting is something that you can find out after receiving some data.
3. Set your test budget
This is more of an obvious rule. First it is crucial to set a daily budget for your campaign in order to avoid spending too much money on bad performing ads. This gives you the chance to optimize your campaigns every few days and see improvements. Your daily budget should not be so little that it would take you so long to gather enough data, and it should not be so large, that it waste your money.
Second, you should set a test budget. This means, how much are you willing to accept an under-performing campaign. Sometimes it happens that you do what you could do but the campaign is not getting positive and you should stop it.
4. Never Stop A/B Testing
A/B testing is a topic that many people paid attention to and talked about. This proves the importance of A/B testing. In this article, I am focusing on A/B testing for your creatives and landing pages.
When you start a campaign, it is important to run multiple creatives (whether banners or text ads) and landing pages. Depends on the vertical, I usually start with around 10 creatives and 5 landing pages and split the traffic equality between them. After a few days, you probably should optimize the setup, and remove the bad performing ones. In this article. [A/B testing article name], I describe different A/B testing techniques in details.
What is very important is that you should never stop A/B testing. There is always room to improve the performance of your campaigns, even if you think you found the best creative, I promise there will be something else beating your creative. So keep continuing on A/B testing. I usually, set between 5% to 20% of my campaign traffic for A/B testing.
Also as a side note, try not to use assumptions about a creative before running A/B testing. Test as many creatives as you can even those you are not optimistic about. You will be surprised.
5. Use dynamic parameters to measure more variables.
Most of the ad platforms do not have a very comprehensive reporting and analytics systems. Which means that either there are variables that you cannot find in their reports or there is no report for the combination of a few variables that you’re looking into.
So if you’re using a tracking or analytics platform, you can usually send extra variables using dynamic parameters.
This also would be very helpful if you are not able to see all conversions in your ad platform.
6. Pay attention to statistical significance
I am going to start discussing about optimization, but beforehand, I believe it is important to emphasis on the importance of statistical significance. Either in your A/B test or in your optimizations, take the statistical significance into consideration. Any action without considering statistical significance may result in bad performance, fluctuations or missing opportunities in your campaign.
Of course, I am not saying that you should use super scientific techniques like finding the probability distribution associated with your stats (as most probably it is not a normal distribution) and then find the statistical significance method related to that probability distribution. This is a waste of time.
The important thing is first to understand the concept of statistical significance, and then to use in your decision making in a way that there would be very low chance of making mistakes.
7. Optimize wisely
Ok this is the last rule and perhaps one of the most important ones. The basic rule is that you should optimize all elements and variables of your campaign such as sources, devices, demographics by filtering the bad performers and boosting up the good performers
I mentioned “wisely” because it is very important to optimize with a clear mind and in a smart way, rather than just removing any bad performing items.
It is very important to think that you are optimizing based on the past performance for the future. So basically, based on the past performance you are expecting to see the future performance similarly. So obviously this is an assumption and even with a high statistical significance, there is always a chance that the future would be different than past. My point here is that you should be very careful not to over-optimize your campaigns.
Over-optimization is the worst mistake that you can make in your optimization. By over-optimization, I mean excluding a potential good performer. The challenge here is that in many cases, you exactly don’t know what are potential good performers and the uncertainty is high. In many cases, because the mediabuyer is getting impatient and wants to reach a positive campaign, he filters some items which had bad performance in the beginning. However, it is important to note that when you filter them, you stop getting data on them, and you may miss a good performer forever. At the end of the day, your campaign always will have good and bad performers, but the key is that your good performer would make it up for you.
I personally, usually keep some bad performing campaigns (on a low budget) only to collect data and search for opportunities.