I have implemented many marketing automation solutions over the past decade and one of the perplexing findings is how organizations put the cart before the horse when they are installing their solutions. I like to say marketing automation solutions are "dumb". Not the kind of dumb as in "this is stupid, why are we implementing these solutions, why not do something else". They are "dumb" in the essence of they need help from something else to be successful. They cannot work on their own.
Marketing automation tools are a slave to the underlying data. All marketing automation tools do is query data and create metadata that is used to create content and messaging for your customers. Now I am minimizing the importance of the marketing automation tools in that sentence, but from a high level, it works.
Since the underlying data is what drives the marketing automation tool, that data is the first step in implementing the tool. Without the proper data, your implantation will fail. Getting the data into the proper format for consumption from the automation tool is the most important step of marketing automation.
Understand the problems to be solved
Write out all the different types of campaigns or communications to be run with the automation tool. This step is vital to understand if there is a gap in your data collection strategy. Also, this identifies if the data is structured properly to even run these types of campaigns. This step comes before buying a marketing automation tool.
For example, I want to send a reminder email to all customers who bought a television that specific cables will enhance the performance of their new purchase by 50%. For this, the data will have to be structured to understand which customer bought a television set, along with cables because you don't want to sound like you don't know your customers, within X amount of time, their email, mailing or app device ID, and the channel they prefer to be communicated with. Now the data team can make sure they have the proper structure for just this one use case. If the data can't be structured accordingly, then the marketing automation tool will not be able to deliver this campaign.
Define success for the campaigns
This can be a simple sentence in each case. What this determines is how the analysis of the campaigns performance will be achieved. Analysis is also part of the marketing automation tool implantation, because I guarantee you that the executives will want to know the impact of this large investment, so the data needs to be prepared to answer these questions.
For example, I want to see the redemption rate and revenue generated, along with the expenses for delivering and cost of goods for the customers who returned to the store and purchased upgraded cables for their televisions. For this the data will have to meld together the ID for the offer, in this case the cable, along with the purchase item along with the expense data from the marketing automation tool and the sales system. These tasks aren't easy, but they will pay dividends if this legwork is done upfront. There is nothing worse than flying blind with your marketing automation..
The expectations for campaign execution times
This is one that almost always gets missed. I have heard of campaigns that run almost all day because the data is not organized in a fashion that is not optimized for the marketers. That kind of performance may be acceptable if the campaigns are run once a month, but for most businesses that is not the speed of digital marketing.
For example, I want to be able to run the campaign for the television purchasers every day. This includes time to run the automation, send out proofs for the collateral and have the deliveries out to the customer by 10AM. This allows the data team to be able to optimize the data structures to make sure the data can be pulled fast and efficiently for all your automation campaigns.
This by no means is an exhaustive list, but it is a start to having a successful marketing automation implementation. No matter how many bells and whistles the marketing automation tools have, if the data does not support the wants and needs of the marketer, it doesn't matter because the tool is "dumb". It needs the data to perform magic.