Let’s talk about automation. Do you automate your ad placements? Do you consider your company automated? Most likely for you and me, this concept has a different meaning. And now I will tell you what automation means for newage.
What Do Ad Campaigns Look Like?
Before talking about automation, we need to understand what we are working with now. Consider a typical media campaign.
Let’s say we use 5 services:
- Google Ads;
- Facebook Ads;
- Direct placement.
Each of them has separate segments: cold, warm and hot audiences. We get 15 types of settings. Each segment has sub-segments with their own creatives — that’s another x3.
Approximately so many slices of data need to be considered for further optimization of the campaign. But the data that we would like to analyze in each of the slices is several hundred.
In general, if we multiply everything, then we get a table with several hundred thousand cells. Specialists should launch a campaign on it, and then analyze it. This is a rather labor-intensive process. According to my calculations, to support such a campaign without automation, at least 2-3 specialists are needed who would devote all their working time to the campaign.
In fact, many do not have the resources for these specialists. Therefore, advertisers launch campaigns for 2-3 segments and optimize them according to basic models. Naturally, the quality of the campaign suffers from this. Naturally, the quality of optimization suffers from this and budgets are not used effectively enough. Let’s see what we can do about it.
Effective Automation: Marketing vs MarTech
In digital, we work not with marketing, but with MarTech – marketing technologies. If we are talking about marketing, then we are talking about strategy, creativity, understanding the user or client. And MarTech is adding a technological part that helps us make campaigns more effective and work faster.
Previously, digital was perceived as another media channel: they made a creative, placed it on a wide or narrow audience, and that’s enough. Now everyone understands that everything is not so simple here and there are many nuances, digital is developing with opportunities and, most importantly, the client understands that it works and will grow multiple times.
When I talk about efficiency, I don’t mean cases like “we increased conversion by 20% and made placement more effective by 40%. I’m talking about multiplicity. We want to use even more targeting, even more settings, and analyze even more data to make advertising as effective as possible. And automation, MarTech is about efficiency.
Next, let’s talk about priority and important automation tools at my discretion, a lot of things can and should be automated, but I chose what I consider a priority for today
What to Automate in Advertising Campaigns
Autoload is essentially a template table that we can populate with settings and parameters to import into systems like Google, Facebook, DV360, etc. The table is individual for each of them, but the template is the same for all campaigns in this service. That is, tables for Facebook and Google Ads are different, but they are suitable for any campaign on Facebook and Google, respectively.
In such a table, you can enter parameters, specify a large number of targetings and settings. It’s definitely faster than through the interface. Yes, you first need to figure out how it all works, but then you can save time all the time.
Basic steps with autoload files for different systems:
According to our forecasts, the campaign that I gave as an example at the beginning, our rather smart and strong specialists would launch in 10 hours of work – 2 days. And having an autoload file, this could be done in 2-3 hours.
There are already cases on using dynamic ads in Search, dynamic creatives in Search Shopping and for retargeting campaigns, and everything is more or less clear here. As for Display and Video, it’s more complicated. Surely everyone has seen ads for T-shirts with your year of birth. A person is more likely to buy a T-shirt with their month and year of birth than with random ones.
How long does it take to make such creatives? Let’s take an audience of 15-45 — that’s 30 creatives by year. Let’s add different designs, let’s say 5 – we get 150 creatives. And if we take also different months, then we multiply all this by 12. Imagine how much time it takes for a designer to prepare this campaign, and then we add the time of a specialist to start it, not to mention the analysis.
There are tools, for example, Creative Studio, in which you can create a master banner (1 creative) and set the parameters by which the year and month in the picture will change in this creative, the price or store address will change.
Yes, you need to spend time to understand, master and configure. To make 10 creatives with 5 resizes, that is, 50 images, a strong designer will need 16-24 working hours. With automation, it is enough to make 1 master banner and prepare the parameters by which it will change depending on the audience.
If everyone is set up and you know the system, then a similar task with only x10-100 variations can really be done in 2-3 hours.
If you need effective creatives for your campaign, write “I need banners” to [email protected]. Or leave your contacts on the “Design” page.
When it comes to automation in digital, the first thing that comes to mind is how to upload data from everywhere. But the problem is not so much how to get the data, but what kind of data to pull up and how to process it further.
As usual, reporting occurs without automation (we also had this about 4 years ago). On Monday, the traffic manager starts uploading actual data from the systems. He visits dozens of direct platforms, collects information about different formats and targeting, looks through several advertising accounts…
The manager spends the whole Monday, Tuesday on this, and only on Wednesday by lunchtime the client receives a report and the conclusion: “This hypothesis was confirmed, and something there does not work”. That is, three days of optimization are lost, and this is significant for large campaigns.
If we want to develop and taking into account our specifics, that we look at a lot of numbers, then it was quite logical to develop this direction and automate.
We have developed a dashboard that collects, compares and organizes data on its own. Now the newage manager spends not three days, but three hours preparing a report, and this time is spent not on “Ctrl + C, Ctrl + V”, but on formulating hypotheses, searching for growth points and suggestions for optimization.
In order for the dashboard to work, we connected the unloading from tracking systems, auditors, web analytics, CRM, etc. to a single database. Somewhere we set up integrations, somewhere we agreed with the sites that they regularly send a report according to a certain template . So in dashboards for our clients, more than 500 parameters are collected:
- placement quality control and fraud protection metrics;
- data reconciliation parameters;
- campaign media metrics;
- KPI data for each segment;
- financial indicators.
If you give out all this information in the form of a table, it will be too huge for a person, in principle, to understand what is written and at least somehow evaluate what is happening. Therefore, our dashboard automatically visualizes all this in different views and sections.
You can see how it works and spin the dashboard template. To do this, write to [email protected] and read the instructions in the article Dashboard newage. 2021: How to automate the work of a media agency?
One of the most important elements in automation is that the systems themselves optimize campaigns based on our data. Marketing is not so simple as to somehow shove it into an algorithm: a person, a specialist, must understand what data to collect and feed to the system.
I find the standard optimization models that exist in Facebook and Google to be impractical, because all campaigns are different. For example, a client needs to bring a new user to a certain section of the site. Accordingly, we first need to set the correct KPI, and then feed the correct data to the system. For another client, we work through a funnel, consistently showing videos with different CTAs for different purposes.
It is important to have data to show the system: “I want to improve this particular KPI.” At this stage, artificial intelligence will help us a lot. A computer has long been able to beat a grandmaster at chess. But the chessboard has a limited number of squares and a limited number of certain pieces.
And marketing is not a chessboard, it is much larger, it has an unlimited number of fields, figures, rivals and third-party circumstances. Artificial intelligence will help us, but we must limit the number of variables for it, draw a board and set clear conditions.
When we talk about this level of automation, it’s not about how quickly we can optimize campaigns. It’s about how often you can iterate.
Conclusion: What Should We Strive For and How I See the Future of MarTech
Automation for digital is a kind of ecosystem, the number of components of which is constantly growing and will continue to grow. We have touched on four main issues, now let’s look at how they will work in conjunction.
- A strategist or marketer prepares a campaign media plan in a certain format, according to a given template.
- Autoload files are automatically generated from the media plan, which can be uploaded to advertising accounts.
- Dynamic creatives are created from the same autoload files with changes for each audience. With parameters that are passed under each audience.
- From the same media plan, the structure of the dashboard with the desired KPI is unloaded.
- As soon as the campaign is launched, the dashboard is ready. He immediately collects information and analyzes it in the form in which we want.
- When a campaign is running, we analyze what and how we work in it, optimize it, and we can loop the work using scripts and algorithms.
All this optimizes marketing and makes it more effective.