Most advertisers view programmatic in the same way that Field of Miracles do a black box: with a mixture of interest and doubt. After all, it is completely unclear what is inside and whether it is worth the offered money.
It is really difficult to understand how programmatic works. Customers usually do not have access to advertising offices, and they have to be content with the knowledge that the agency buys “some users” on “some sites.” In fact, the users are targeted, the sites are relevant, and at the bottom of the black box is a valuable prize – a real contribution to increasing sales.
Let’s take a look at how programmatic works and how it helps the advertiser’s business.
Programmatic and contextual advertising
Programmatic is a method of purchasing online advertising at an auction, which involves the use of automated systems and algorithms. An attentive reader will ask: “And how is this method different from buying the same contextual advertising? Yandex.Direct and Google Ads have both an auction and automation. “
The answer is simple: it’s about the audience. Contextual advertising is aimed at people who are already choosing a product or service (and setting appropriate search queries). Contextual campaigns work with the established demand and help to get fast sales. The catch is that these sales are limited by search volume. Programmatic helps to find a new audience and bring users to the site who are at the very beginning of the sales funnel.
Programmatic and Display Advertising
The attentive reader will ask here too: it turns out that programmatic is just a new name for display advertising? After all, media formats are usually connected precisely to increase coverage. Programmatic is really like display ads, only it is much more accurate and smarter.
Let’s imagine that we have an offline store of expensive watches. To make more potential buyers know about the store, we can buy a banner on the Forbes home page. And the site, for example, will show a banner to all site visitors for a month, including girls who have come to look at the photos of millionaires. Of course, there will be targeted users among the visitors, but it is not at all necessary to shoot at these “sparrows” from all media tools.
Programmatic helps us to solve the same problem an order of magnitude more efficiently: select an audience entirely consisting of target users (for example, Muscovites with high income), and show ads only to them – but on different platforms. Among the sites may be the Forbes website, and the impressions on it will most likely be cheaper than with a price purchase. In addition, we will be able to place not only the usual media formats (banners, videos, text and image ads). In addition to them, the programmatic arsenal includes native advertising and Rich media (ads with interactive and multimedia content – ed.).
The fundamental difference between programmatic and media placement is that we buy out the audience, not the space on the site, and in most cases this happens through an auction. It is also important that programmatic opportunities are not limited to increasing reach or increasing brand awareness. This tool can work as part of a performance strategy.
How programmatic works
The market for programmatic services consists of three types of services: user data providers (Data Management Platform, DMP), advertising inventory platforms (Supply Side Platform, SSP), and advertising purchasing systems (Demand Side Platform, DSP).
DMPs collect data about the behavior of Internet users from different sources (websites, ISPs, plugins, etc.), store, process and segment them using machine learning. As a result, user segments are formed that may be of interest to advertisers. Standard segments are categorized by demographic characteristics, location, and interests. More complex (custom) segments are developed at the request of an advertiser or agency.
It is DMPs that select the audience to which we will show ads. The largest providers of user data in our market are Google and Yandex. There are also specialized companies such as Weborama and AmberData.
SSPs are platforms on which individual sites or entire ad networks automatically sell their inventory. Both Yandex and Google and ad networks (Begun, Between, and others) have their own SSPs. DSPs help to buy ads for given targetings and rates, the largest of which also belong to two search engines.
DSP and SSP interact in real time. As soon as the user opens the webpage, SSP auction the ad display on it. DSP evaluates the relevance of the user and offers a price. SSP selects the best deal and displays ads. The entire process takes a split second and ends before the page has even loaded.
The same auction model (Real Time Bidding, RTB) is used by contextual advertising platforms and targeted advertising services on social networks. However, many consider RTB and programmatic to be synonymous. In fact, RTB is just one of the programmatic elements, just like an engine is a part of a car.
This is not the most difficult process, but any advertising tool must be used correctly. First of all, we need to accurately define the target audience (CA) and the tasks that the placement should solve.
A programmatic goal can be either to increase reach or increase sales. In the first case, the number of targetings will be minimal, as will the cost of advertising. In the second, ads will be shown to narrow segments of users, which always costs more.
As a target audience, you need to ask users who are really interested in our product or service. Here you need to honestly face reality. After all, as much as some advertisers would like to see their target users active and ready for adventure, sometimes they are just people with an excess of free time.
Read on the topic:
We transfer the description of the target audience to the DMP, which selects suitable segments and loads them into the DSP. At the same time, DMPs comply with all legal requirements: user data is impersonal, restrictions on data collection are taken into account (for example, you cannot collect segments consisting of children).
After that, we upload creatives to the DSP, and also set bids, budgets, and other settings. You should pay particular attention to the rate management. You shouldn’t set the same bids for campaigns with different objectives. Let’s imagine that one of our campaigns is aimed at increasing reach and takes into account a minimum of targeting, and the second is aimed at increasing sales (and works with a narrow custom audience). If you set the same price, then the first campaign will collect all impressions in the world, and the second will not receive them at all.
Another plus of programmatic is the ability to run in the “we need tomorrow” mode. Campaign preparation takes just a couple of days: that is how much time it takes to load data into DSP and pass moderation. But such terms are possible only if all creatives are ready and the terms of placement are agreed.
Now we just have to track the performance of different segments, formats and platforms – and optimize campaigns. It is important to remember that programmatic creates new demand, and does not work with already formed one. It is wrong to wait for direct conversions and evaluate placement results by the last click, as in contextual advertising. An attribution model that takes into account the contribution of each advertising channel to conversion provides an objective idea of the impact of programmatic on sales. And here we cannot do without end-to-end analytics.
We have all the moves recorded
Setting up end-to-end analytics starts before advertising is launched: to get statistics for each session of each user, we need to embed tracking pixels in all creatives in advance. This way we can collect and then combine data from any source – from contextual advertising and SEO to CRM systems and call tracking services.
All statistics are uploaded to a secure cloud service (our agency, for example, uses Microsoft Azure) and transferred to the Power BI data visualization platform. In Power BI reports, we will be able to see how many advertising contacts each user had before the purchase, which channels were involved and in what order. And now nothing prevents us from distributing the conversion value between different channels in accordance with the chosen attribution model.
End-to-end analytics allows you to track how the programmatic audience returns to the site through contextual advertising, organic links or other traffic sources – and ultimately makes a purchase. As a result, we can assess the impact of programmatic on sales in other channels and make informed management decisions.
Programmatic in performance strategy
Let’s see how programmatic works in real life – as part of a performance strategy. In 2018, our agency, together with colleagues from the programmatic platform Omniscienta, carried out an advertising campaign for the immersive show “Faceless”: a theatrical performance of a new format that maximally engages viewers in what is happening. The goal was to increase ticket sales through the show’s website.
We developed a strategy that used two tools: programmatic and contextual advertising. Programmatic was supposed to increase reach and drive brand traffic to the site. The purpose of the contextual campaigns was to convert interested users into buyers.
The programmatic audience was selected using neural network analysis, the material for which was data on the behavior of site visitors. The collected segments were tested to determine the optimal demographic groups, delivery times, and device types. So, we found out that the best time to display ads is from 12:00 to 14:00, and the highest conversion rate is observed among women aged 25-32.
After a week of testing, we optimized the campaigns. The budget was redistributed in favor of segments that are optimal in terms of the number of users, their presence in ad network traffic and efficiency.
For three weeks of an advertising campaign using programmatic, we managed to attract 1400 interested users to the site. The plan for impressions was exceeded by 177%, the KPI for clicks was fulfilled with a margin (104%). At the same time, the programmatic audience turned out to be more converting than ordinary site visitors. For example, the share of clicks from a show description to a page with tickets increased from 29% to 42% during the campaign.
As you can see, the programmatic black box is quite decipherable. You just need to competently use this tool and clearly understand what you want to get at the end.