In internet marketing, there are many ways to find new customers: search engine, display and targeted advertising, video, SEO promotion. What else can you think of? How do you find users who haven’t visited your site yet but will be interested in the offer? One solution is to target a look-alike. Let’s see how it works.
How look-alike works
A look-alike audience is a segment of users who are similar in characteristics and behavior to website visitors or customers. It is formed by advertising systems based on their own segments (in Google Ads they are called remarketing lists) created by an advertising specialist: these can be audiences generated by analytics systems or customer lists (their contacts and other parameters).
For example, you have collected an audience of your website visitors who have made a purchase for more than 5,000 rubles. The system analyzes what topics these people are interested in, where they live and “what they breathe”. This data will be taken into account and used when creating similar audiences – users who, in terms of their interests and behavior on the Internet, are similar to customers who ordered goods worth more than 5,000 rubles.
Thanks to machine learning, Google’s Lookalike Audience lists are updated in real time to reflect the changing behavior of users on the remarketing list. No additional actions are required from a specialist to update them. It’s worth noting that similar audiences won’t include visitors from your original remarketing lists. Ads will only be shown to new users.
Ready look-alike segments in Yandex are also regularly and automatically updated – new information is added and irrelevant information is removed. The segments only contain data about those users who were active in the browser or application in the last 30 days. If the original segment changes, the same audience will be affected as well.
About source lists
To form the initial audience, you can choose the key parameters that are important to you. For example, it can be:
users who added the product to the cart;
customers whose data is in your CRM;
those who live or work near you;
buyers with a large average check;
those who have seen your display ad, etc.
Please be aware that Yandex and Google do not support creating audiences based on sensitive and personal parameters (certain aspects of medicine, health, religion, race, etc.). Also, you won’t be able to create an audience based on segments that others have shared with you.
The time to create a Lookalike Audience can vary from a few hours to a few days.
For advertising in Yandex.Direct, segments of similar users are created in Yandex.Audience. In the statistics of the intended source segment, you can see how the Yandex system estimates the degree of user similarity. This information helps to determine if the selected segment is suitable for creating a similar audience: the higher the similarity, the better. It is also important that the source segment contains at least 1000 anonymous identifiers. And, if the degree of similarity is high and the reach is at least 1000, it makes sense to test this type of targeting.
Google determines which remarketing lists are suitable for generating similar audiences. To do this, your original list must have at least 100 users (more than 500 recommended).
According to Google, when Google Remarketing and similar audiences are used together, conversions can increase by 41%.
Briefly about similar audiences on Facebook, VKontakte and myTarget
Similar audiences work well on social media too. Each ad site has a completely different algorithm for finding common user characteristics and different requirements for the original audience. Facebook recommends choosing an initial audience of 1000-50,000 people, VKontakte – at least 300 users, and the minimum number of users in the downloadable myTarget list is 1000 people. But it’s important to understand that the minimum audience size may not bring the expected results. As well as a very large audience – in this case it is better to divide it into several parts.
Look-alike audiences can be collected from the following sources: lists of customers from CRM systems (for example, email or phone numbers), users who interacted with your ad, subscribers in social networks, website or mobile app visitors, and so on. Visitors who have performed a certain action on the site (for example, made a purchase or left a request) can be collected using the pixels of social networks. As with contextual systems, similar audiences on social networks are recommended to be used in conjunction with other tools. For example, you can add targeting based on demographics or interests. And, of course, you need to keep track of the timely updates and quality of the original audiences to get the maximum impact.
Selecting the target audience for creating a segment of similar users should be thoughtful and clearly defined. If you need:
New audience – form a look-alike based on all visitors to your site.
Interested audience – for look-alike, take those who left their contact information on the site or, for example, put a product in the basket, but did not buy.
The target audience – collect users who have converted on your site.
Potentially loyal audience – take as a basis regular users who regularly make conversions on the site.
To expand your reach, you can use the audience of users who are at the beginning of the sales funnel (for example, all site visitors). This makes sense if you want to tell as many people as possible about your product or announce a new promotion.
If you have your own database of contact information of visitors (email, phone numbers), then on its basis you can create a segment of similar users who, presumably, will be interested in your offer.
But to increase the number of conversions, the above audiences will not work, since there is a high probability of getting non-targeted visitors. In this case, you should focus on the audience of users who have made at least one conversion on your site. The system will identify the distinctive features of buyers and, based on these characteristics, will find potential customers who do not yet know about you.
If your offer is a highly targeted profile, then the likelihood of finding customers using similar audiences increases significantly.
Divide your original audiences by sales funnel (e.g., visited a website, made a conversion, made a conversion more than once).
Place a campaign in ad networks targeting your most valuable customers (for example, those who look like high-order shoppers).
Don’t focus on one audience, test different options. For example, target a look-alike audience of buyers with different check amounts or visitors to certain sections of the site.
Use look-alike audiences for bid adjustments in search campaigns. For example, set up adjustments for an audience of loyal customers. This will increase the likelihood of your ad being shown to those who are interested in your offer.
For audiences that look like early funnel users, lower your bids. And for an audience of users close to converting, increase.
Apply geographic and demographic targeting to show your ads to the most targeted users.
Don’t be annoying and limit the frequency of your ads. We recommend setting it 2-4 times a day.
If your reach is small, add larger similar audiences to your targeting or increase your bids.
Regularly update customer data that will create a similar audience. It is important to use up-to-date information because the user’s interests, like their online behavior, can change.
Test different creatives.
Let’s sum up
With the help of a similar audience, you will increase your reach and attract new customers, as you will show ads to those people who are most similar to loyal visitors to your site. But its quality directly depends on the parameters that you set when creating the initial audience, and on what stage of the sales funnel it is at. therefore choose users who are closest to making a conversion… This will help the system find an effective, similar audience.