Because digital marketing is always changing, advertisers must also be able to constantly evolve.
Testing and experimenting is extremely important to the success of your evolving PPC campaigns. They help you understand if proposed changes to your account will support your overall goals.
Today, let’s take a look at the highlights of a new guide AdWords put out on the best practices of confident testing, and how you might use them in your own campaigns.
1. Decide How and What to Test
Use campaign drafts and experiments to test out the changes you want to make, concentrating on the high-value levers for the most impactful tests. This is an organized approach that allows you to quickly see results and make the changes to your original campaign.
Campaign drafts and experiments are AdWords features for drafting changes and experimenting with the results. Drafts allow you to create multiple changes that you can either apply to the actual campaign or to an experiment. If you opt for experiments, these measure the results of your changes before you actually apply them to the live campaign.
Drafts work by mirroring the real campaign’s setup, though some features and reports aren’t available in drafts. Experiments then take a portion of your traffic (determined by you) and test the changes. Afterward, you can compare the experiment to the original campaign.
When you run experiments on drafts, test out the potentially high-value changes to things like bidding strategies and ad extensions. Lower-value outcomes and changes that are simply interesting, such as tweaks to the color scheme, are more likely to waste your resources.
Remember that drafts and experiments are not the only way to test, and can’t provide testing data on everything in your campaigns, such as certain automated strategies. Try attribution models for account-level conversion tracking, geo experiments for learning the effectiveness for whole channels and Brand Lift to grasp the way YouTube ads impact brand perception and user behavior.
2. Create Experiments for Clear Results
You want your experiments to yield fast results, and for those results to be relevant to your campaign so that you can quickly make changes for better performance. Ensure this by keeping the experiments simple and efficient.
First, focus the tests on one pre-determined variable at a time, as you cannot tease out the effect of change if several of them happen in one experiment. Segment the experiments into stages—first test the bidding strategy segment, for example.
Strive for efficiency by designing tests to reach statistical significance (and conclude) quickly. One way to do this is by having the experiment and your original campaign split traffic 50/50 (keeping more traffic with the original is less risky for performance, but will take longer).
Be aware that the more subtle the changes, the longer they can take to create a change (good or bad) in performance.
Measure the success of your tests with one pre-determined metric, such as total sales, to more easily identify what’s working. Use secondary metrics for added insight.
Finally, don’t alter original or experimental campaigns while the tests are running—it’s possible but not advised if you want to avoid biased results. Any necessary changes should be mirrored across the original and the experiment.
3. Analyze Results and Choose Winners
This is about determining what worked so that you can implement it and get those great results across the board.
Give the experiment enough time to give you confidence. AdWords indicates how confident you can be in your results by little arrows within your experiment results. If your results have no arrows, it means there’s not enough data. Allow your chosen metric time to build up enough data so that you can be confident in the outcome and expect similar results in the future.
Always look for outliers that don’t fit with the high-level headline result. High-volume factors can skew results, and certain areas might perform differently.
Implement any relevant changes in future campaigns, but consider how they affect original campaigns. You have the choice, with an AdWords experiment, to update original campaigns or convert to new ones, and have to decide on a case-by-case basis.
Generally, updating original campaigns allows you to port the changes over and keep the history of your campaign.
Converting to a new campaign pauses the original campaign and keeps the experiment going as a normal campaign. This is ideal for testing new campaign structures or if you want to hang on to findings from an experiment on bidding strategy.
The final step is to keep records of your experiments so that you can refer to them in the future. It’s up to you if you want to simply keep the list in AdWords or create a spreadsheet with tracked progress.
AdWords testing is a way for you to take advantage of the tools and data available, and make the best possible decisions for your search advertising. Observe these best practices, implement your results and enjoy improved performance!