I wrote this post for MMT
Transitioning to data-driven marketing, which has become a must-do for business performance in the last 5 years, is not uniquely a mindset to adopt.
Using efficient data for decision making is not about building data-heavy graphs in business presentations to help you determine your next business steps. You need an infrastructure setup to collect those data sets but also to compile and often visualize the data in dashboards so that you can leverage it. You need to be able to obtain, organize and read the relevant data so that you can make smarter business decisions.
Manual preparation of reports and data can be time consuming and resource heavy, which can lead to duplication and errors that can easily be overcome by automating fetching flows and updates.
Table of content:
- Is there a specific issue to data collection in the advertising industry?
- How do you connect data flows?
- Next steps
Is there a specific issue to data collection in the advertising industry?
From the beginning of the media digitalization era, each individual publisher and service provider has done its best to keep up with technological developments on the market. However, the industry at large has not managed to set standard processes for buying and delivering media (some countries have independently created a standardized approach, but this is not the norm). The absence of standards makes it challenging to find fast and unique data flow connection solutions that work for everyone. Marketing teams, wishing to get a clear view of their media activities, have to individually connect the data from the necessary platforms like social media, programmatic, DMPs, publishers, TV booking systems, etc.
How do you connect data flows?
Solution 1: Third-party provider
Some companies have made it their mission to help marketing teams easily get these connections by acting as a third-party service provider. They have developed adapted solutions that deliver formatted data to their client’s warehouse (or theirs if their client does not have one) where they organize the data. This has made connecting data provision fast and resulting in quick launch times for data-driven projects, offering speed and simplicity at a cost. If businesses do not have the team or the experience needed to build the data infrastructure necessary to run their own solution, these providers can be extremely helpful.
This connection still needs to be leveraged through an additional visualization solution, as the connection system does not include a direct interface to “read” what it has extracted. What’s more, the meta-connectors do not offer ad-hoc offers, and small businesses end up paying for a more complex service that they do not need.
This also deprives marketers of transparency regarding data delivery and control as it means being dependent on the provider’s ability to offer the required connection and regular updates no matter what happens and because update issues that may arise are not always anticipated (e.g. the database does not update properly the day you have the largest budget presentation of the year).
|Positive: Data connection is made fast and easy to a very large number of partners. API speaks directly to your applications.||Negative: This can lead to a lack of transparency and flexibility on a couple different fronts: data-availability (when new data is ready to be consumed) & troubleshooting. No savings possible: You pay for a Ferrari when you need a Fiat.|
Solution 2: API connection
Modern platforms mostly offer API access to catch this data – API being the layer of software which allows communication with another application – to allow data extraction at regular times and maintain an updated database. Some make it possible to generate data ingestion as well, which makes it possible to create a centralized self-service solution that can give commands as well as receive information.
But again, API is built into the code of an application and does not follow a particular standard in the marketing industry, which forces businesses to address these connections one by one in order to obtain their own data. Connecting to some of these APIs is extremely challenging. Only veteran engineers can code a working connection and manage to fetch the relevant data.
On the positive side, APIs offer transparency and control over the data flow and engineers can quickly resolve issues and troubleshoot connections if anything happens.
|Positive: Provides direct contact to your dataflow from a particular platform with control and transparency.||Negative: Requires qualified engineers to connect and maintain your connections to your application. It embeds the API in your application => which means that your application has an API too.|
Solution 3: RPA
Robotic process automation (RPA) is a surface level automation software that can be used to allow applications to talk to each other.
For those who haven’t yet incorporated a dedicated protocol for data sharing, human actions can be “mimicked” with a piece of program that retrieves the necessary information. This is called robotic process automation, or RPA, and once coded it can perform tasks the human hand would perform otherwise, like basic chatbot solutions or a report generator on a platform UI. It is a simple piece of automation, not an AI, and performs repetitive tasks like copy and paste, logging to applications, scrapping data and making a calculator very well. It is fast to implement and does not affect the structure of any of your software applications, which keeps it from disrupting everything that is already in place. However, it can easily crash or underperform of there is a disruption of the environment. A basic RPA can only deal with the keyword it knows and will not identify spelling mistakes. However, it would be a perfect way to update a database when using forms to fill dropdown menus and ticked boxes.
|Positive: RPA is a good solution for fetching data when no API is available. It is cost-efficient and can be an effective way to replace repetitive human tasks, thereby reducing potential human errors.||Negative: RPA does not determine whether the task has been well-executed. It can only rely on preformatted information to guarantee quality results.|
At MMT, we have studied the connectors available on the market, and our team is able to interact with API as well as write RPA. We adapt our client offers to the most relevant solutions for them, their business, their needs, both long term and short term. Get in touch with us for more details.