June 5, 2024

Where are Marky's Customers Coming From?

Author Profile Image
Benjamin Coad
Freelancing data and computer scientist

This graph for the way that customers found Marky is laid out by the percentage of people who found Marky via a specific discovery method. In this case, there are 8 ways, spanning from January 2024 to March 2024. The taller the area, the more people found Marky that way.

Collecting the Data

Marky ( is an up-and-coming AI marketing platform. They launched in September 2023 via a TikTok influencer video and have since mostly invested in Facebook advertising. When a user signs up for their site, they ask, "How did you hear about us?". They also have an "Other" field where people can type their own answers. This leads to some messy data. I was hired by Marky to clean and analyze the data that they've collected from more than 20,000 answers. Below I present my approach and results. Enjoy!

My Approach to Cleaning the Data

At first, the raw data showed that a lot more people found Marky through other methods than the ones provided. Taking one look at the custom responses added when people selected “other” shows that most people didn’t see one of the above-provided methods. To clean this data so that all the "custom responses" fell into one of the eight categories, I wrote an algorithm. I created a cleaning function that looked for certain keywords in the custom responses and used to that map it to one of the eight predefined categories. For example, "ig" maps to Instagram. I then looped though all of the responses; and if the response contained one of the provided keywords, I added that response under the corresponding group. This included misspellings and synonyms that pertain to that group. Some of the examples in the data are that someone spelled Instagram “Instragram”, and a lot of people called it “ig”. 

Assumptions of My Approach

The result after looping through the data showed that most people found Marky through the likes of Tiktok, Facebook, other people, publications, search engines, and LinkedIn. This is great; however, there are some caveats to taking the liberty of cleaning up the data as I did. Responses containing the selected keywords provided may not be under the group they were assigned to. For example, if someone said “twig”, it would fall under the category “Instagram”. Also, it’s easy to misinterpret certain responses or completely miss some altogether. I wrote a program to clean up the data. A human didn’t go through every single response, so it’s completely possible that some responses that would have fallen under one of the categories were missed, as none of the keywords applied to it. To fix the biggest problem of responses being misinterpreted by the check because it had the keyword in it, I made sure to limit the amount of short, broad keywords such as “ig”.

Interpretation of the Results

We can see from the graph that most people found Marky through TikTok and Facebook, indicating that a big percentage of the marketing budget is going into those areas. This also indicates that a lot of people needing Marky’s services are using Facebook and TikTok, which is great. Because of this, it would be good to keep in mind that these are the main discovery methods and therefore it would be wise to fund advertisements on these social media platforms accordingly.

Right behind those, we can see “friend”, which means that Marky was recommended to people by friends, which is word-of-mouth marketing. Word-of-mouth is a powerful aspect of marketing, so it would be great to increase the amount of people finding Marky via this discovery method. 

Behind “friend”, “blog, article, or publication” is trailing right behind. This is great news, as it means that people find Marky interesting and unique. Blogs, articles, and publications usually only cover things that stand out from the crowd or do something different than alternatives. The fact that a lot of people are finding Marky this way indicates that people find the product important and innovative.

Lastly, we can see that other people found Marky through the likes of Instagram, search engines, LinkedIn, and other methods. This indicates that most people looking for ways of easily marketing their business are generally not looking through Instagram or LinkedIn, and do not know that such an option for marketing exists, leading them not to look Marky up on search engines. To increase the number of customers coming from search engines, we can increase the amount of spending on Google Ads or other alternatives. We can also add more key tags associated with Marky so that the first thing people see when they look up easy marketing is Marky. Increasing word-of-mouth will lead more people to look Marky up on search engines.

Recommendations to Business Stakeholders

One way we could grow our customer acquisition is by prompting customers to share their experiences with others, which leads me to my next suggestion. Another way we can increase our word-of-mouth marketing is to provide testimonials from people in Marky’s marketing and advertisements. People are more likely to trust other people who have used the product rather than facts provided by the product itself. Lastly, a way we could increase word-of-mouth is by engaging and interacting with user-generated content. Doing things like commenting on TikTok, Facebook posts, or other forms of user-generated content regarding Marky encourages people to make the most of the product and gets them excited about it.


  • Most people are finding Marky through TikTok and Facebook, so increasing funding to those might be a good idea.
  • Word-of-mouth marketing is a great way to help even more people find Marky, and we can increase it by engaging with customers and user-generated content, prompting people to share their experiences, and showcasing testimonials.
  • We can increase the amount of people finding Marky through search engines by increasing word-of-mouth and putting more money into Google Ads or other alternatives. We can also add more key tags associated with Marky.

Benjamin Coad
Freelancing data and computer scientist

Keep reading

No additional items at the moment.