As a startup company, we at Valohai didn’t have a clear idea about what kind of marketing message would resonate in our audience.
For sure, we had an idea about our customers and what problems they might have but we hadn’t really studied the effectiveness of different messages in any systematic way.
We had had advertisements but they were always separate ads thus the results could not be compared. It was impossible to say whether the difference in CTR (click-through-rate) was due to the marketing copy, ad placements, or due to an image for example.
I red couple blog posts about defining the marketing message and here is the basic idea of what I found:
- Know your audience
- Define the market and what is your position in it
- Identify the problems in the market
- How will your solution answer to that problem
- Present the outcome of using your product
I feel that this is really for-dummies-level, we all know these, but more interesting is how we started defining these steps.
We have a remote team so I prepared an excel sheet for us to go through in a Google Hangouts meeting.
You can copy it here.
I’ll briefly explain the sections in the sheet.
- Who are your target audience? We added titles here, but defining your customer can vary from defining demographic information to personal traits. What ever is relevant in your case.
- What are the problems of your desired audience? Think about all kinds of problems. Not only about the problems that you can tackle. We wrote down problems that aren’t currently addressed by our product and this might give a hint to the product team about future improvements on the product.
- Rating. In a scale of 1 to 5, rate how big is this problem to your customer and how well your product solves the problem. Customers might have huge problems that you don’t solve and you might solve something that is not that big problem. When you scale a problem to 5 and you rate that you can solve it fully, there is your place to strike.
- How you solve the problem. Just a pure engineering response of what can be done with your product.
- Then my favourite part. Add fairy dust and write drafts of your marketing messages. At Valohai, we like unicorns, fairies and all the magical stuff so we wanted to add some glitter to the engineering explanation of our solution and this is where the marketing message comes from.😉✨
The excel is missing market positioning since we had done that before and adding that part would have unnecessarily prolonged our meeting. Add positioning row if you haven’t really defined how you stand out in the competition.
Even though many things that we went through were obvious, I feel that as a newcomer, it was very useful to me to go through our most potential customers and their problems once again. I have worked in Valohai for three months now and since I don’t have a daily contact to our customers, it is good to remind myself of the old things and find new golden nuggets as we go.
To be honest, our excel didn’t look pretty after we had finished working with it. There were multiple different marketing messages with different angles so I chose the most promising ones and added them to Google Docs file and went them through with our engineer. I didn’t want to steer the marketing messages too far away from our target group who are also engineers or data scientists.
We had four different customer segments initially but ended up merging two of those and left one out since it was too hard to reach just the right people in that target group in our industry. Now we had two main audiences and four different marketing messages to each.
Audience 1: Machine Learning engineers and Data Scientists
Audience 2: Head of Data
Where to test the messages?
Thanks to Joona Heino’s great idea, we chose LinkedIn as our channel since it allows defining the target audience precisely when advertising based on job roles. Below you can see our finished messages. Note that the image stays the same in each since we don’t want that to affect on CTR.
Messages to Machine Learning Engineer
Topic: Time saving
Messages to Head of Data
Topic: Continuous integration
Topic: Team is changing
Topic: Know what the ML team is doing
Settings and pro tips
Each ad set is shared in the U.S., Singapore, Europe and UAE equally. I could have tested whether people in different parts of the world react to different copies, but that is not part of this experiment.
In addition to targeting solely based on job roles, I added skill targeting also. Not all Head of Data Managers know machine learning and there are different types of Data Scientists. We wanted to target those, who have some skills in machine learning.
As a recommendation, LinkedIn advertising tool wants to optimise ads based on the ad performance – better performing ads are shown more. Usually I’d use that selection but in this case I wanted to show the different ads equally to avoid being biased by the ad optimisation.
It is now 26th of April and we are seeing the results. I will write about the results and next steps soon! 😎