We started our collaboration with a clothing industry company in late January-early February. The company's Google advertising was not fundamentally flawed, but due to time constraints, the entrepreneur wanted to outsource the advertising. Another one of our entrepreneurs, Mikko Lintunen, who serves as our in-house Google advertising expert, took charge of the project.
Mikko explains the initial situation as follows: "The advertising account had been set up in the traditional way, with a large number of different keywords placed in a single ad group. Based on this alone, it was clear that we could improve the performance of the advertising account, as an account structured in this way cannot meet the searcher's search intent effectively."
Meeting the searcher's search intent is one of the fundamental principles of Google advertising. The advertiser who can address information-seeking users with the most relevant ad and further direct these visitors to the most relevant landing page wins. For a user using Google, they must be shown an ad precisely about the topic they are seeking information on, and they must also be directed to a landing page that addresses that topic.
Advertising should always be developed in a way that there is no significant drop in sales through the online store when the person managing the advertising account changes. For this reason, exact copies of previous campaigns were created using Google Ads' campaign drafts and experiments feature (Drafts and Experiments). This feature allows us to compare the performance of two different campaigns and conduct controlled A/B tests. We allocated 50% of the budget to the original campaign and the same 50% to the new campaign. The targeting of the advertising remained the same in terms of keywords, but about ten times more ads and ad groups were created for the account. This way, the ads were tailored to address people using various search terms as precisely as possible.
"In addition to more specific ad texts, one of the most important changes was the more considered use of landing pages. Previously, all ads directed traffic to the same landing page. In the new campaign, all keywords were reviewed, and traffic began to be directed to several different landing pages according to the searcher's search intent. For example, keywords related to jeans now directed traffic to four different landing pages instead of the previous one. With the new ads, the aim was to address the searcher's search intent as effectively as possible," Mikko explains one of the key changes affecting the account's performance.
The importance of addressing search intent can also be seen from the results of the experiment: 12 sales were generated through the advertising, of which 11 came through the new campaigns. The experiment was conducted using Google Ads' experiment tool, so both campaigns received exactly the same visibility and ad spend, making the result entirely comparable. In drawing conclusions, we use the p-value, which is a statistical measure of probability. In this case, the p-value was 0.004, which means there is only a 0.4% chance that the result is due to chance. Normally, we use a p-value threshold of 0.05, which is a 5% threshold.
The advertising budget was evenly divided between both campaigns, so the old campaigns might have generated only 2 sales in the same period if advertising had not been developed. The new campaigns, on the other hand, with a full budget, could have generated 22 sales. This end result, which brings added value in the long run of tens of thousands, was achieved in just over two weeks. What's next?
"We follow the principle of continuous improvement in our work. After the experiment, the old campaigns were stopped, leaving only the new campaigns active on the account. Immediately after the change, however, we created new test campaigns using the experiment tool. This time, we are testing Google's machine learning-based bidding strategy against the currently used manual bidding strategy."
So, there is plenty to test and improve well into the future - and we can certainly further enhance the performance of the account.
