Case: Sold $160,992 worth of women’s clothing thanks to targeted advertising

Implementation

First of all, we analyzed the advertising cabinet. As it turned out, the client was buying traffic to the site on his own, but due to a lack of time and energy it was inefficient. Therefore, we studied in detail the previous launches and their performance, analyzed competitors’ activities, and developed a strategy for buying traffic. Eventually, we tested three different purchasing formats before getting results on models that were worth scaling up.

Transferring events, setting up Conversions API, and unloading the catalog are the main necessities of working with the online store. We did all the integrations with the help of Google Tag Manager. For today, the integration of LiqPay online payment to the site is in progress.

Next, we implemented analytics and organized the process of direct communication with the sales department. A Google spreadsheet was used as daily analytics to track sales dynamics. User behavior on the site and in-depth metrics were tracked using Google Analytics. After numerous tests, we determined that the actual sales results are estimated based on data from the CRM-system Bitrix24.

The next step was to test hypotheses, optimize effective bundles, analyze user interaction throughout the marketing funnel, and make changes to advertising campaigns to increase conversions.

The structure of advertising campaigns is an important part of buying traffic with this volume of models. Creating a single one was inefficient. We have made the theses that we constantly adhere to when receiving new offers.

1. Segmentation by price, category, and seasonality
2. Budget allocation for the test is determined by the sum of 2-3 model sales values per day (~$20-40 per ad group)
3. Test with optimization at the ad group level (ABO)
4. Test of 2 creative formats: video and static

We regularly structure advertising campaigns to fit a specific situation. As a result, we manage to increase the final number of successful models in our work.

Conclusions

It took quite a long time to accomplish the task. Basically, the purchase was carried out with the goals of “Conversions” and “Sales by catalog”.

Our collaboration can be divided into three stages based on the change of strategies. We calmly analyzed the data and drastically changed the strategy on the basis of the results obtained.

The first solution was to buy traffic according to the usual model of working for e-commerce. The logic was as follows: we developed creatives leading to the homepage of the site. A woman saw a beautiful design banner and moved to the home page to interact. Next, everyone interacting with the site was caught up with retargeting on a range of 7-28 days. It looks easy, but real things are quite different.

The main challenges we faced were:

• the abundance of choice – when seeing 100+ items on the site, women got lost and ended up going away to reflect. The emotion gained from viewing the creative faded away while scrolling the site.
• it was impossible to control the flow of leads to specific models, since it was difficult to predict user behavior.

As a result, the actual return on traffic is affected, as well as production, which is customized to the format of a particular model’s work. We came to the conclusion that we need to change the purchase of cold traffic. Partially the approach is still the same today, but globally, we made changes based on the results.

The second solution was cold traffic for a specific model. The point of this approach is that new people see a specific model in the ad and after clicking on it they go to the card of this product. In one test we launched from 4 to 10 different items. The main advantage of the solution – 90% control of the traffic.Still, some people from the general flow of visitors got to other models and ordered them. Globally, though, we controlled the attention and interest of the audience.

As a result:

• marked trends – by analyzing the effectiveness of the models, we had a chance to determine how the demand and trends of the audience were changing. Unfortunately, when working with the main page of the site, such tracking was difficult to access. The range of model types was very wide.
• efficient production operation

We adapted quickly enough to work model by model and the results started to improve. The structure of advertising campaigns, was completely rebuilt and is still under improvement. Challenges included frequent restarts, ineffectiveness of 80% of positions, delays in getting materials to develop creatives, etc. But if you keep up the pace, set up communication and processes for launching advertising campaigns, the result is worth it.

Dynamic retargeting works with both emotional and rational communication. It is driven by the need to bring users who have different shopping reasons back to the site.

The third solution was to adjust the structure of advertising campaigns. With a large database of feedback from managers and the client, we had ideas for improving a working strategy. In terms of settings for buying cold traffic, the broad audience of women works best. We don’t limit ourselves to interests or other settings, because our goal is to scale the bundle in the future. To do this, we need to have a large audience capacity. In our case it is ~8-9 million women. At the same time, we have dynamic retargeting in order to catch up with the warm audience who visited the site.