Yandex Direct for real estate developers: how to set up campaigns that lead to deals

Search ads are shown to people who have already typed a query, so they are good at capturing existing demand. The Yandex Advertising Network, Yandex's display network of partner sites and apps, together with retargeting helps add volume and bring back people who have already visited the project's site. What Direct can hardly do is introduce a new residential complex to people who have never heard of it and are not looking for an apartment in that district. That takes outdoor advertising, display formats, property classifieds and content, which I wrote about in detail in the article on promoting a residential project.

This leads to the channel's main limitation: the number of relevant searches in your city is finite. When the inquiry plan is larger than that demand, campaigns start expanding into generic and neighboring queries, bids go up and the quality of inquiries goes down. So before launch it helps to check in Yandex Wordstat, Yandex's keyword volume tool, how many people a month search for apartments in your district, and to estimate honestly how much of the plan Direct can cover.

As a rough guide, for a mid-market project in a city of a million people, an inquiry from search now usually costs between 4,000 and 8,000 rubles, and one from the ad network costs about half as much. However, noticeably fewer people from the ad network pass the sales manager's check, so the price difference is largely misleading.

It is worth splitting campaigns by their role. If you mix brand and generic queries in one campaign, the cheap inquiries from the project's name will mask the real cost of hot demand, and you will notice it only once the budget is already spent. If you combine search and the ad network, the strategy will start moving money to wherever inquiries are cheaper, which is usually the network with its lower-quality inquiries.

By default, Yandex now offers the Unified Performance Campaign, a single campaign type that combines search and the network. At the start I still keep them separate, because that way you can see which of the two brings in buyers and which simply brings cheap inquiries. You can merge them later, once the strategy is learning from qualified leads and there is enough data for both.

Advertising on competitors' project names is a separate question. Direct allows ads to show on such queries, but you cannot use someone else's name in the ad text. These campaigns usually convert poorly, because the person is looking for a specific project. They start to make sense when you have a clear advantage in price or completion date and the landing page shows it right away.

A developer's keyword set is fairly compact: district and street, type of housing, number of rooms, mortgage programs and the project name. Expanding it with generic words like "apartment" or "real estate" is almost always expensive and adds little. It is much more useful to work in the opposite direction and cut off unrelated demand with negative keywords.

A basic list for new builds usually includes words related to renting (rent, daily, to let), the resale market (resale, from owner), informational queries (calculator, forum, what is), construction jobs (vacancies, foreman), as well as the names of districts and cities where the project is not located. The word "reviews" is better kept in the brand campaign, because someone looking for reviews of your project is in the middle of choosing an apartment, and it makes sense to send them to the page with construction progress.

By default, autotargeting extends impressions to related queries, so it is worth restricting it to the targeted categories. During the first six weeks it helps to open the search query report once a week, because it almost always turns up new phrases for which the ads are shown for nothing.

Manual bid management in real estate is mostly a thing of the past, because Direct's automated strategies do better as long as they have enough data and the right goal. The second condition matters more than the first. The strategy looks for people similar to those who have already completed the goal, so choosing the goal essentially decides who the ads will bring in.

Form submissions or clicks on the phone number are reasonable only for the first couple of weeks, while there is no other data. The next step is a targeted call, meaning a call longer than a minute recorded by call tracking. The best option is a qualified lead: the sales manager marks in the CRM that the person fits by budget, timeline and need, and this status is passed to Yandex Metrica, Yandex's web analytics service, as an offline conversion.

To learn, a strategy needs roughly 10 conversions a week per campaign, and a single campaign rarely produces that many qualified leads. That is why it is convenient to use a composite goal in which a qualified lead is valued, for example, five times higher than an ordinary targeted call. This gives the algorithm enough signals to learn from, while it still understands which inquiries really matter to you. I described how to connect ads, call tracking and the CRM into one setup in the article on end-to-end analytics in real estate.

A person sees several new-build ads on the results page at once, and what helps an ad stand out is concrete facts: the starting price, the monthly payment, the completion date, the walk to the metro, the finishing. Words like "comfortable" or "premium" appear in almost every ad and tell the reader nothing. A price in the headline may lower the click-through rate a little, but that is more of an advantage, because you stop paying for clicks from people who cannot afford the project.

The product campaign is built on a feed, which is a list of apartments with floor plans, areas and prices. It needs to be updated every day from the same system that holds the unit availability chart. Otherwise the ads keep showing apartments that are already reserved, and a buyer who calls hears that the apartment has been sold and leaves irritated.

A landing page built for a specific segment works noticeably better than the project's general website. If someone searched for a two-bedroom apartment with a family mortgage, the first screen should show them exactly that: two-bedroom apartments, a payment calculation and a button to book a viewing. A quiz along the lines of "find your apartment in 30 seconds" usually increases the number of inquiries but lowers their quality, so if you do need one, the strategy should still learn from qualified leads.

It is convenient to calculate the Direct budget from the plan: how many qualified leads the channel needs to deliver and how many inquiries that will take. If it turns out the money is enough for only four or five inquiries a week per campaign, the strategies will learn slowly, and it is better to reduce the number of campaigns by merging similar ones. A detailed budget calculation from the sales plan is in the article on the developer marketing budget.

It is worth planning the test for six to eight weeks. During the first two weeks the strategies are learning and the numbers jump around a lot, so it is too early to draw conclusions from them. Each week it helps to look at how many inquiries passed the sales manager's check and what one qualified lead cost, after a month at the cost of a sales office meeting, and after two or three months at the cost of a reservation. It is better to change one thing at a time, because otherwise you cannot tell which change worked.

If a contractor's report contains clicks, click-through rate and cost per inquiry but no qualified share and no cost per meeting, it describes the work of the ad account. Such a report tells you nothing about how the advertising affects sales.

Almost every project I have audited has the same problems. Brand queries are mixed with generic ones, so the real cost of hot demand is invisible. The strategy spends months learning from form submissions. Nobody cleans up the ad network, even though mobile apps and games bring in cheap random clicks. The feed is updated once a week, and the ads show apartments that have already been sold.

Response speed deserves a separate mention. A person who comes from Direct is usually looking at several projects at once, and if you call them back an hour later, you reach them far less often than if you call in the first few minutes. No campaign setup can make up for that.

At Matveo we start work on a developer's Direct by checking exactly these points and setting up the transfer of statuses from the CRM. Most often the first results come from changing the strategy goal, cleaning up the ad network and separating out the brand campaign, and new campaigns are added after that. I described how Direct fits into a developer's marketing as a whole in the article on the real estate marketing system.