Case Study. Health supplements store. USA.

About the client: This is an e-commerce store, which is selling Lab-verified Clean Food under their own brand in the US market.

For all e-commerce projects, we are using ROAS + transactions volume as main KPIs. ROAS provides you an understanding of profitability; i.e., if you have invested $100 and received $700 in sales, your ROAS would be 700%.

At the start of our collaboration, the client tried to set up AdWords campaigns, but this resulted in only 3 transactions with an ROAS 10.95%

Screenshot google analytics before set up a ppc campaign

After 7 months of collaboration with us, he is now getting 187 transactions (last 7 days) with 777% ROAS.

Screenshot google analytics after set up a ppc campaign

List of experiments:

Experiment 1. Campaigns in the broad match based on product feed. Budget – $400. Result – 287 clicks, 0 transactions.

Target keyword was the name of the product in the broad match (must be a long term keyword), for example, Organic Extra Virgin Coconut Oil, 1 Gallon. Our search report for this keyword looked like this:

Target keywords

As you can see, the generated traffic was relevant; however, there were numerous brands, which the client did not have on their website.

Campaigns were segmented by their profit margin with different bids based on product price:

AdWords Segmentation

Output: You need to work out a good list of negative keywords before starting this test and never use keywords, which contain two words because this would generate many irrelevant search terms.

Experiment 3. Search remarketing (RLSA): 22 transactions, 513% ROAS during a 30 day period.

Search remarketing proved to be one of the most effective targeting methods; to start, you need to create an audience and wait to receive at least 1000 people in your list. Audiences can be added in active search campaigns with bid adjustments or you can create a separate search campaign and pick more general keywords, which usually have a low conversion rate.

We have been testing search remarketing for only one month before we have been banned by Google due to unclear reasons. We did not break any rules but since we are selling health supplements (for example chlorella), their policy team decided to block the remarketing feature for our account.

Experiment 4. Display and Dynamic Remarketing: 4 transactions, 57% ROAS.

We just started to test these remarketing strategies but were banned. This is really frustrating because we have invested considerable time to fix bugs with the dynamic remarketing integration.  

Output: Be aware of AdWords policies, sometimes it can be a good idea to ask your account manager in Google for advice regarding specific business models that you want to promote.
Experiment 5: DSA campaign based on product feed. 91 transaction, 1,018.89% ROAS during the last 30 days.

After the first experiment, we have already worked out a long list of negative keywords; so, our idea was to create a separate ad group for each product supplying AdWords with its target URL so the system can pick up keywords based on page content instead of keywords.

This experiment always requires a lot of attention, especially for the search report, as some ad groups might produce considerable irrelevant traffic.

Experiment 6. Shopping campaigns: 446 transactions, 485% ROAS.

We have created two campaigns:

  1. A campaign for all products. Make sure that you have a separate goal for all products (if there is one group in the campaign, this is called ‘Everything else’ in ‘All products’). Make sure that you have separate goals for all products (if there is one group in the campaign, this is called ‘Everything else’ in ‘All products’).
  2. A campaign for most profitable products. Once you have statistics from your main campaign you can choose the most profitable products from Shopping in the dimension tab. Alternatively, you can add custom labels for your top sellers in the shopping feed and reallocate them to separate campaigns from the start.

Experiment 7. Display campaigns: 880 clicks, $35 spend, 0 transactions.

Display Network (not including remarketing) is not very effective for e-commerce; however, we have decided to run a small test and target selected websites, which are related to the products we sell. This did not work out.

What’s next?

Right now we are setting up new experiments such as:

  1. Testing Video Shopping ads
  2. Beta features, which we get from Google as part of the Red Badge Program
  3. Smart Display Campaigns