Client’s background

• 2 marketing agencies started work on the project, and the results were unsatisfactory, with a negative ROAS of 0.78
• After the partnership had ended, the agencies picked up the ad accounts and the Facebook pixel
• No tests of the used advertising creatives and audiences

 Client’s goal:

• minimum ROAS x1 - for the first month of work and tests 
• stable ROAS x4 - in 3 months' work

Analytics

First of all, we set up analytics on the website, and installed Facebook pixel, as well as all the basic events on the sales funnel, namely:

• Basic Page View
• View Category
• View Content
• Add to Cart
• Initiate Checkout
• Purchase

It is important to clarify that the Purchase event is not a payment, but an order - payment is made upon delivery. Therefore, to minimize refusals after payment and reduce delivery problems, some regions in Russia have been excluded.

What we did first of all

As we had no project background data, client database, and tests, we had to do everything from scratch. Us and business owners identified the three most popular product categories, namely kitchen hoods, ovens, and hobs, and chose the hottest items among them.

Initially, we decided to put the product category at the ad set level in the advertising campaign structure. In other words, on Prospection we have one advertising campaign with three ad sets. Also, immediately after installing the Pixel on the website, the audiences for the retargeting and the structure of the advertising campaign were created. (Below we’ll talk about this in detail).

How did we choose the audiences?

After market research and analysis, we chose the two widest audiences for the Prospection campaign. As we work with the CIS market, the audience burns out very quickly, especially with the budget given to us. Thus, it was necessary to use as wide audiences as possible.

The audiences for the Retarget campaigns covered the entire customer journey map with a maximum available window of 180 days. The audiences were divided into 3 product categories, as for the Prospection:

For hobs:

- view content 180 days

- view category 180 days

- add to cart 180 days

For ovens:

- view content 180 days

- view category 180 days

- add to cart 180 days

For kitchen hoods:

- view content 180 days

- view category 180 days

- add to cart 180 days

A separate retargeting ad campaign was created for Initiate Checkout:

- initiate checkout 180 days

Besides these audiences, a dynamic retargeting campaign with a catalogue was also set up, but it did not yield the expected results and was soon switched off.

Design creatives are not always a good choice

The business owner insisted on using beautiful creatives and involved his designer. We think it is necessary to make things as simple and clear as possible. Therefore, we agreed to test both our own and the designer's creatives. We drew up brief terms of reference for the designer, which spelled out the obligatory points to be present on the banner namely:

1. Product name

2. Price

3. Promotional price, if available

4. Brand logo

In turn, we decided to make a carousel ad for each product category - as simple and clear as possible. We placed the goods on a white background with the name and price. In addition, we used a carousel ad with photos of goods in the interior.

A week of tests later, we got the following results:

Designer's creatives:

Our creatives:

We realise this testing was not objective, but we did not have enough budget for qualitative tests, so we ran two parallel campaigns with different creatives and drew our conclusions based on the ROAS received

Problems with Analytics

Initially, everything was set up correctly, all Facebook events were handled, and everything was displayed correctly in the analytics. Ten days later we stopped getting any data and for a long time we couldn't find the problem, it turned out to be simple. The IT department responsible for maintaining the site just removed the Facebook pixel, causing us a lot of headaches and stress. Since then, we have had an agreement with the client that if any changes and adjustments are made to the site, we are notified immediately and test the pixel and all events again.

Interpretation of results in different attribution windows

At the end of the working month, ROAS x2.7 was received by attribution 1-day post-click.

If we look at the 1-day post view & 7 days post-click window, which is the default attribution window, we see ROAS x7.

We reported exactly 1-day post-click data for the client, but the standard attribution window indicates either a long transaction cycle or multi-channel marketing work.

Follow-up work on the campaign

After testing audiences, creatives, and methods for communicating with potential clients, we settled on a wide audience of people (Broad) and simple banner creatives - carousel ad. We created a new campaign on Prospection with only 1 group of ads and the entire budget was transferred there.

However, the work doesn’t end there. We are continuing the scaling-up process, testing the categories of campaigns, working with new items in the product matrix, and advertising promotional offers.

To date, constantly working on optimising and scaling campaigns, we have achieved ROAS x6.74 on a budget of $11,400, generating over $77,000 in revenue for the client (not including associative conversions and multi-channel marketing)

Result

✅ ROAS = 6.74 with attribution window 1 day post click

✅ ROAS = 12.33 with attribution window 1 day post view and 7 days post click

✅ $77,000 revenue in 30 days at the attribution window 1 day post click

✅ A campaign structure has been developed, with the ability to scale and make changes

✅ The different variants of advertising creatives have been tested

✅ Communication with the IT department for website development and support has been established