“I’m paying all this money for marketing but how do I know how effective it is?”

Travel Agency owner

In case I haven’t said it before, I love Disney. I’ve loved it as a kid and after college I had the privilege to work there as an intern and full-time. Though I’ve since moved on from Disney my family and I still visit regularly. While researching one of our trips we came across a travel agency which focused on all destinations, not just Disney. 

The Problem Statement

After using one of their agents to book our trip I got to talking with the owner of the agency about her self-grown business. (Side Note: After booking with and using them, I would always use a travel agent when planning a trip, it’s free for the trip taker and ends up saving you a lot of money; travel agents know all the deals!) Like many small business owners she was/is the CEO (Chief Everything Officer) and trying to work in and on the business at the same time. One of the biggest questions she was wrestling with was, “I’m paying all this money for marketing but how do I know how effective it is.”

She was very inventive and was trying many different avenues for marketing which included, but weren’t limited to: encouraging her agents to get agency-branded car wraps, taking out ads in movie theaters which show prior to movies playing, and of course social media ads. 

The Problem

With each of these, besides social media, there was no follow-up. If anyone called the agency or booked a trip because of seeing the car wrap or the ad at the theater, there would be no way of knowing unless they were asked or volunteered the information. Then on the small chance the agency did receive the information, they didn’t have any location to record the conversion to the specific marketing effort. 

Because there was no information or data for us to reference, data-analysis was impossible. So, we decided to change our focus and get creative. 

Rerouting

First, we wanted to prepare for the future by capturing those marketing conversions. To do so, we added a simple field on their ‘Contact Us’ page to include a “How did you hear about us?” line item. We then added a drop down of all of their marketing efforts. Now, anyone who reaches out for more information via their website is able to be tracked. They don’t even have to make a purchase it’s all right there.

Second, I suggested creating different URLs for each marketing effort; each link would automatically redirect to the homepage. Using the analytics of the page we could see how many hits each got. For example, the link visible at the theater could be www.<TravelAgency>.com/movies. Or for the car wraps it could display: www.<TravelAgency>.com/onTheGo. Something simple to remember for the customer.

Third, I decided to look at the path of a potential customer. I put myself in their shoes to see how easy it would be to get information from the agency. Clicks add friction to a transaction. Friction causes people to stop moving forward towards a transaction. With so many businesses clamoring for people’s attention we wanted to make it easy for the potential travelers to find info or book a trip. 2 – 4 clicks is a good rule of thumb from start to finish.

Putting myself in the shoes of a customer I first went to the agency’s website. It took me 2 clicks to transact (1- navigate to their ‘Contact Us’ page, 2- to fill out and submit the form; this was good because typically 2-3 steps is all you want customers to try and take to get where they want. However, then I was just left hanging. There wasn’t any follow-up, no automated email saying they were going to reach out soon, no message on their website saying I completed the form correctly. I think I submitted multiple times to ensure it went through. 

Social Media was my next investigation. My hypothesis was: customers are more likely going to find the agency via their friend’s suggestions and the previously mentioned paid ads. While the agency posted a lot of information, frequently, the follow-up on each post was the same: “Contact Us via [email]”, with the email itself being text and no URL. 

The lack of URL did two things: 1) It added extra friction to the transaction process (i’d have to perform at least 5 steps just to email them) and 2) it eliminated any data linkage we would be able to use as association between the marketing efforts. My suggestion was to remove as many steps as possible by using URLs where possible.  

Fourth, we started to focus on their Social Media messaging. Since we couldn’t look at specific marketing avenues we focused on their posts, since that is a form of marketing as well. They were using almost all of the big platforms but Facebook was far and above the site with the highest investment and engagement. To help them understand their marketing and what their voice was I looked at what they were saying. Literally the actual words used in each post.

Texting

Within Machine Learning there is a discipline called ‘Text Analytics’ which uses statistics to discover insights from text, things such as sentiment (positive or negative tone), Keyword Analysis, Topic Analysis and much more. 

Without going into too much detail I specifically wanted to get an idea of what they were objectively saying/posting to their customers, show it to them and see if it was in agreement to their marketing-messaging-brand. 

Using Topic Analysis for their previous quarter I was able to discover 6  topics they were discussing in their business page posts. Each topic is actually just a cluster of key words but put together they paint an interesting and insightful display.

 

As you can see in the image, each color is a different topic with the top 10 words per topic listed. The bars display the words’ contribution to each topic. Some topics discussed promotions they were advertising, others were about experiences their clients could discover and one of them, topic number five, was explicitly about how to contact their agency. I will note here that topic number does not have any significance, it’s just the order the machine learned them in.

Since we have trained the machine on each topic we can actually point the machine to each post and have it guess or predict which topic each post is mostly likely associated with. 

Let’s inventory real quick on what we have: 1) all the posts of the previous quarter for the business’ Facebook page, 2) the six topics discussed during that quarter, and 3) the topic each post is talking about. Can you believe that such a small amount of data gave of us all of this wealth of information? 

Well we took it one step further, next we counted the number of times each topic was discussed and, again because it was linked to each post, we were able to get engagement for each topic. I gave each interaction some points (a ‘Like’ got one point because it only takes a click to give, a ‘Comment’ got three points due to the fact it takes more effort on the part of the giver, and a ‘Share’ got 4 points because it takes a lot for someone to put something they see on their own feed). 

Now with the tally of frequency discussed and the engagement points for each topic we can visualize it. Check out the graph below.

The height of the color bars is the count of the frequency each topic is discussed, the red bar is the engagement for said topic. There are many insights to be gleaned but what we found the most interesting is in the third and fifth topics. The third topic (green) was about fun, exciting and interesting vacations their clients could take. It had the most engagement, but was the second-to-last leased discussed topic. Then there is the fifth topic (dark blue), it had the least engagement, by far, but was in the top three topics being discussed. What was topic number five you ask? That’s right, the ‘Contact Us’ topic where they asked potential-clients to do a six-step-friction-encouraging copy of the agency’s email address, paste it in their email client, write a message and send. 

Final Thoughts

Now, no judgement is passed to this agency. In fact, this is a great find for them because we were able to find co-witnessing facts and points to help their marketing. All from a small amount of data. Imagine what more insights could be found with more data? What questions do you have that could be answered with your data?

Key Takeaways

1. Focus on your customer’s journey. Work to reduce as many steps or friction as possible to get your customers where you want them to go.
2. Blaze your trails. Where possible, create links for your customers to follow so you can track their progress later.
3. Follow-up. Think about ways to capture how people hear about you.
4. Introspection. Look at what you’re saying and see if it is engaging with your audience the way you want.
5. Measure. Whatever result you want to track, think about how you can capture the steps or progress.