How to scale to 4 mln revenue with ROAS 9.5 in pandemic times

• To make ROAS 5+ so that the business could steadily be profitable and 2X or 2.5X its current revenue.
• To adjust funnel structure in such a way so that it would be possible to steadily track all the key KPIs on every stage.
• To set up retargeting.
• To resolve the issue with audience overlap, and start testing new creatives and ad copies.

This kind of structure allows distributing traffic more effectively on all the funnel stages and helps to achieve the best possible results while scaling. To implement it, your creatives, texts and formats should be carefully prepared for every stage of the funnel.

On TOF, we used videos of several bestselling products and DPAs as collection ads with products on the bottom and a video or an image on the top. On MOF, we added video feedbacks from previous customers and videos with influencers. On BOF, we ran DPA carousels, texts with promo codes, and text feedbacks. We also made upsells and cross-sells with special offers for previous customers.

Many believe that ad copies do not really influence campaign effectiveness. But this project proves such belief to be wrong. One of the key objectives was to find new ad copies variations that could keep ROAS stable. For this, we had to create a system of continuous testing because the client was using short cookie-cutter texts back then. We started testing various texts for certain funnel stages:

1. short
2. long
3. feedbacks
4. with and without discounts
5. lists of characteristics and advantages

We tested a fair amount of new texts in Split test campaigns, and within the first week, we managed to find several variations that allowed keeping ROAS at 6-7. Ad copies variations that helped to generate max ROAS in Split Test campaigns were carried over into the main TOF campaigns.

Along with ad copies, we also tested new creatives. We chose Collection ads as the key format for this. Generally, it’s nice and easy to use because you can add dynamic catalogue to it. We mostly utilized it for TOF, but also tested this format for other funnel stages. We started testing Collection ads using brand products videos, and then we switched to testing more vibrant videos that we edited ourselves. We also told our client more about the types of videos that would perform the best and advised them to start cooperating with influencers.

And we kept testing different videos.

The format of ads that we used for this client started showing good results pretty soon after we had combined best performing ad copies and collection ads. We also noticed that this method could 3X ROAS.

Audience overlap is one of the most frequent issues on Ecommerce projects that start scaling. Most often it causes some adsets to either stop spending at all or to spend far less than they should. It happens because Facebook algorithm detects high audience overlap in certain adsets and starts showing other adsets more often. Audience overlap on this project was up to 60% in some adsets and we had to resolve this issue somehow. Since we had experience with this kind of problem, we knew how to deal with it. Here’s how:

• Kill adsets with the highest audience overlap if they perform much worse than the other adsets (it’s not always the case, though)
• Turn on CBO (this client wasn’t using CBO)
• Add exclusions to the adsets. You have to at least exclude your customers and VC for the last 30-180 days (amount of days depends on a project). You can also make broader exclusions, for example, of those who watched videos for the last 30-180 days.
• Try broadening your current audiences in adsets. I mean, if you use certain targeting settings by interests, you can try to broaden these audiences. Or if you use 1% LAL, you can try 3% or 5%

We applied the 2nd, 3rd and 4th tips listed above, and almost did not use the 1st one because we had to keep certain adsets running. First off, we started testing CBO on several campaigns, and within a week all the other key campaigns were also switched to CBO.

Then we addressed the problem of exclusions in the adsets. While making an audit for these guys we noticed that they weren’t excluding custom audiences in most of the campaigns. We excluded all those who made a purchase and VC audience for the last 180 days.

In addition, this client was using a 1% Purchase LAL audience for 30 days in the majority of the adsets. We decided to test 3% and 5% LALs in most adsets.

All these steps helped to make audience overlap on average three times lower.

The client didn’t have permanent retargeting campaigns and that was the problem we had to deal with immediately. Not targeting users who had abandoned carts, the company failed to have at least 20% of additional sales.

The first thing we did was creating and launching the retargeting structure. Our perfect structure for long-term retargeting campaigns is to gather VC, ATC, and IC audiences for 3-7-14-21 days and show different ads variations to each group. You can also test broader time periods.