Case Google Ads: How to improve something that is already good

Содержание

1430

Content

1430

Even if you have an account that performs well, you can always do better. There is room for experimentation in your account! And if you have a separate budget for this, then it’s generally great.

You can test some hypotheses, run an experiment, look at the final result. If the result is not what you expected, you can still +/- predict the final results of the reporting week/month.

When an account doesn’t reach your KPIs, there’s no time for experimentation. You are trying to plug holes, optimize campaigns, cut off everything unnecessary that does not bring results. But when things are stable, this is the best time to experiment and scale. After all, even a 0.1% increase in the conversion rate is still a result.

With small steps, we improve performance and in the end, we see the fruits of our work. Look at you too.

Input data

Product: blinds for home and office

Country: Australia

Challenge: Scaling Results

Niche feature: a lot of competition, since in Australia it is a fairly sought-after product

Around the middle of last year, a client from Australia contacted the Penguin-team PPC agency. The average ROAS for their campaigns was 700-1400%. Agree, is it good? However, when scaling, they had problems, since more traffic didn’t bring results in the form of an increase in conversions and ROAS simply decreased. The client tried to scale for six months until he came to us.

Results BEFORE working with the PPC agency Penguin-team

Results for May. It is the last month before applying to the PPC agency:

conversions — 797;
conversion rate – 6.70%;
revenue — 188 188.95 A$
ROAS – 1,278.69%

Work stages

Working on the ad account:

We restructured the ad account.

The search campaigns were well structured, but we noticed that the devices still had different effectiveness. Most of the traffic comes from mobile, but it converts better from a computer. Because users can search from mobile and end up buying from desktop, we didn’t want to completely cut out traffic from both devices. But we need to spend budgets more rationally in order to improve efficiency.

We decided to split the campaign across different devices. This allowed us to better control budgets for different devices. Once the campaigns had enough data, we used different smart device strategies.

We saw that for mobile, the ROAS target strategy worked very well, and for computers, maximizing clicks. Traffic from computers was more converting, we tried to collect as much traffic as possible for these devices.

Structuring shopping campaigns.

A typical mistake is when all products are added to one Google Shopping Campaign at once. This account was in the same situation. All types of blinds were in one campaign. We divided all products into different groups according to the type of product, and for the more sold ones, we created Smart Shopping Campaigns.

Smart Shopping has greater reach and better performance than Standard Shopping through the use of machine learning and remarketing lists.

Display Campaigns.

In this project, the display campaigns were used to increase brand awareness, but without any structuring and tracking of user behavior on the website. There were campaigns, there were banners and responsive ads, they were somehow shown to users with similar interests and that’s it.

We structured campaigns by interests/events and created more themed banners for each audience. And showed them to users who recently had a moving event or who recently moved. The creatives were with an offer and information about installments and discounts when ordering from X pieces, as well as about delivery to the address and installation. Users who were in the process of moving were interested in targeted delivery and installation because they didn’t want to bother with that either.

For users who were just interested in curtains and window decoration, we showed banners with different types of blinds and an offer to order a sample with pieces of fabric. So people could see live how this or that fabric is suitable for their current decor. Such users have nowhere to hurry, they prefer the rational choice “to walk around, look” and only then choose. In addition, here we didn’t sell directly, we led users to certain articles about the benefits of blinds, their types, and possible options.

In the future, we tracked users’ behavior. Created a remarketing audience for users who scrolled ＞70% of the page and spent ＞3 minutes on the page. This is how we filtered users who were more likely to be interested in our product. Then we caught up with them with offers and various promotions. In general, this idea worked and the audience showed good results in the future.

Site work:

We worked not only on the advertising account but also on the website itself:

tested various titles, as well as their sizes;
tested the location of offers on the page;
tested the location, size, color of buttons with CTA.

We tested all the elements that, according to our hypothesis, influenced the behavior of users on the website and the conversion rate.

For example, all the buttons on the website were designed in corporate colors — pale pink, all in the same style. We had a hypothesis that changing the color would help to increase the conversion rate, as users are used to seeing the green “order” button more often. This testing helped us see that the green button page had a 4.07% higher add-to-cart conversion rate than the corporate color button page.

Yes, there is no dramatic increase in the conversion rate, but various tests and hypotheses help to gradually increase efficiency.

Results AFTER working with the PPC agency Penguin-team

Results after a month of work:

conversions – 973;
conversion rate – 6.32%;
revenue – 235 082.25 A$
ROAS – 1,569.13%

Comparison of results for 12 months of work:

conversions — from 3,735 to 12,344;
conversion rate – from 3.30% to 6.98%;
revenue — from A$1,176,735.35 to A$3,993,905.40;
ROAS — from 501.09% to 2,024.61%

Indicator	May 2019-May 2020	June 2020-June 2021	Absolute change	Relative changes
Conversions	3,735	12,344	8,609	+231%
Conversion rate	3,30%	6,98%	3,68%	+112%
Revenue	1 176 735,35A$	3 993 905,40A$	2 817 170,05 A$	+239%
ROAS	501,09%	2 024,61%	1523,52%	+304%