Beautydiscount.ru, an online cosmetics and perfumery store, turned to ZV.Digital with the task of increasing traffic to the site, increasing sales and reaching a KPI for DRR of 14%.
At that time, the overall DRR indicator for contextual advertising was 33.05%. Four campaigns were launched in Yandex.Direct, which worked with 70% of the DRR, and a trading campaign for all products worked in Google Ads.
In this case, we will describe how to work with project advertising from September 2019 to June 2020. Running a little ahead, we note that the annual plan of the online store was closed in June: revenue from advertising in Yandex.Direct increased by 490%, from promotion in Google Ads – by 106%.
Collecting the semantic core
We started by collecting keywords. At the first stage, we selected masks – the broadest keywords – in Russian and English. Then they parsed it using the KeyCollector.
Further, we divided all pure semantics by campaigns depending on the types of keywords.
Key words were also grouped in KeyCollector using the “Group Analysis” function. There, all keywords are divided by repetition – focusing on this, the words were divided into groups.
Types of ad campaigns
Different types of campaigns have been set up in ad offices:
by brands of cosmetics manufacturerspresented on the website. This was the bulk of the campaigns. The most popular, demanded brands with potentially good conversion rates were selected for advertising;
by the brand of the online store itself Beautydiscount;
wholesale campaign – for wholesale, dropshipping and fulfillment sales, the client has a separate page on the site. For it, we collected two advertising campaigns: for macro and micro brands with two types of queries: “brand 1 wholesale”, “brand 1 wholesale”, “brand 2 wholesale”, “brand 2 wholesale” – and chose the strategy “Conversion Optimization”, where Specified the purpose of completing a wholesaler form;
dynamic search ads in Direct and Google Ads;
for general inquiries about cosmetics and by certain categories of the site;
and smart Shopping campaign on Google Ads…
For brands of manufacturers with a large number of series of cosmetics, separate campaigns for each episodeFor example, we launched separate campaigns to advertise the Avene brand lines: Cleanance, Xeracalm and others. For some brands, separate campaigns by product categoryeg for Vichy shampoos and Bioderma sunscreens.
If the brand didn’t have a lot of semantics, we loaded all the keywords into one campaign for this brand.
Dynamic search ads (DSA) in Yandex.Direct were configured both by feed and by site pages for all products. These campaigns have increased bids for users who have left items in their cart. Also, these DSAs were divided according to geo targeting: separate campaigns for Moscow and the Moscow region and separate ones for all other regions of Russia.
DSA in Google Ads was configured for site pages for all products and for three regions: Moscow and Moscow region; St. Petersburg and Leningrad region; Russia (excluding the two previous geos).
This is how our creatives looked in the Display Network (advertising and remarketing):
For more information on how to organize campaigns in Yandex.Direct and Google Ads for an online cosmetics and perfumery store, read our article on the structure of advertising accounts with tips for promotion.
Optimization of contextual advertising
In the course of the campaigns, it became clear which keywords were effective and which ones were not bringing results at all. The most effective were general keys, phrases with commercial links (buy, price, online store, etc.) and keywords for specific products.
We regularly monitored and optimized all campaigns.
To optimize some campaigns in Yandex.Direct, we connected Origami: we used the Conversion Optimizer and set the Target RRP. The service did not allow us to set the required DRR value, which we needed (14%), the minimum could be set to 26%. This was explained by the fact that there were few statistics and data on conversions, it was necessary to set a more realizable KPI; we set the DRR to 26%, and after two months we reached this level for campaigns that were connected to the optimizer.
We chose campaigns to connect to Origami based on analytics data: we looked at which campaigns had an average result – not the best (so as not to spoil effective campaigns), but not the most ineffective. And, of course, an important requirement was that in aggregate, for all campaigns included in this Optimizer folder, there were 30 conversions per month.
Working with keywords and bids… Based on the data on consumption and the level of RRR in Google Analytics and Yandex.Metrica, we made a decision to adjust bids for specific keywords or to disable ineffective keywords. In some cases, we changed the keyword match type to exact to be displayed for relevant queries.
Adjustment of branded campaigns by target audiences… Here we created two copies of the RK by brand:
one excluded users who visited the site (Site visitors 30 days -100%);
in the other, on the contrary, they set up various increasing adjustments for the necessary target audiences (for example: looked at the N brand without making purchases in 30 days + 1200%). In this campaign, key bids are initially set to low.
We also divided them by geo: Moscow and Moscow region. Then, if they worked well, they added St. Petersburg and the Leningrad region, excluding Moscow and the Moscow region.
Working with budgets… Budgets were added to effective campaigns or redistributed from other campaigns with low efficiency.
Checking landing page links in your ad… It often happened that a product was out of stock, and, accordingly, an ad with this product had to be disabled. In addition, we have occasionally found more suitable landing pages for the ad in the process, for example, when the ad was promoting a particular product.
Disable ineffective campaigns… There was a big problem with categorical advertising campaigns, since the site does not have a filter in the link, and there are practically no category pages for brands (only some categories for some brands). Let’s analyze the situation:
The buyer enters the site to select the Avene cream, opens the category “Face Creams”.
Below he chooses the right brand – Avene.
Clicks “Show”, and the filter in the URL is not fixed like that. Accordingly, we could not use this link as a landing link in advertising campaigns for the Avene Facial Creams category.
In such cases, we were forced to conduct ads based on “brand + category” queries on the general landing page for a brand with a large number of products, where the user had to scroll for a long time to find the desired product.
This flaw can lead to an increase in the bounce rate. In our case, campaigns with such landing sites turned out to be ineffective. Therefore, for advertising certain categories of goods, it is extremely important that the filters on the website of the online store work correctly and are transmitted to the URL via GET parameters – then they can be used in advertising campaigns. Well-designed landing pages will be able to quickly lead the user to what he was looking for, and convenience is critically important here, the conversion of the campaign depends on it.
Minus search queries… Quite often, we did some workaround to exclude impressions for irrelevant requests. Here, special attention was paid to display by synonyms – it happened that an advertisement for one brand was shown at the request of a perfect other brand (most likely, due to the fact that it is a competitor in the market for the advertised brand). For the backing track, the SEMtools for Excel add-in and the PPC tool were used.
We unloaded a list of search queries from statistics for a certain period and checked in two ways:
- We checked in the SEMtools add-in the number of repetitions of certain formulations, consisting of 2-4 words, irrelevant of which we can minus.
The advantage of this method is that not only individual words, but also entire irrelevant phrases can be minus.
- Next, we checked the list of queries in the PPC tool program: we uploaded the entire list there and divided it into words, immediately looked at the number of repetitions of each word, marked irrelevant ones and then also included them in the list of negative keywords.
Working with bid adjustments… The adjustments were made in accordance with the performance by mobile devices, by gender and age, by audience segments. The rates were changed by segment as follows:
Reduced for users who made a certain number of visits without purchases within a certain period.
Increased for conversion users and audience of users who have achieved certain goals on the site: adding an item to the cart, adding an item to the Wishlist, micro conversions, visiting more than N pages on the site, viewing certain pages, categories, brands, etc.
Disabling ineffective sites with high spending and low CTR. This applies both to RKs in networks and to search RKs.
Impressions on other search sites, besides the main Yandex search, for example, in a search engine using images, significantly reduce the CTR of campaigns, so they also need to be minus.
When working with online store advertising, it is important to look at e-commerce reports, statistics on specific products, highlight the most successful products and categories, understand which products have sold well before, and promote them first.
We have set up contextual advertising for all product categories almost from scratch. KPIs for DRR and online store revenue were achieved in June. Thus, DRR in Yandex.Direct decreased by 33 pp, ARR in Google – by 15 pp. Income from advertising in Yandex increased by 590%, from advertising in Google – by 106%.
Now the client is moving to a new site, and we continue to cooperate on it.