Datamining for Arts Majors

Don’t Let Data Intimidate You

How many times have you read a marketing article, heard a talk, or saw a post about using data to increase your sales, build your business, and skyrocket into the universe? How many times did you get that anxiety rise, and deep down felt less than, intimidated after reading it?

Stop it. Don’t Beat Yourself Up! 

It’s been years since we read the Target story about the pregnant teen, so we all need to just get over ourselves and realize no one has this mastered yet, and the thinking that it’s just too much an investment to even start is defeatist and just plain wrong. 

You can do this. You’ve got this. 

Along with a million excuses. 

I want to tell you a story about what you can do with the data you have. 

What Datamining Really Is

It was a regular evening on a regular day of the week in a regular business that isn’t a tech company. We had a regular small team and an IT department of one, who also served as the “all things with a plug” support for the entire company. I had an idea, and on my own time began to research it and pull it together. 

Normal small business. Everyone wearing many hats. Everyone wanting more …. and already pushed to the limits. With outdated systems and bloated customer data. 

That’s how we all saw it, but the data we had, even in the form it existed, was TWENTY YEARS of historical data for different audiences. 

Sure, we had a complicated buying process, with cross over for B2B and B2C, and the data wasn’t cleaned, scrubbed, and wasn’t likely tagged as associated clients … it was a mess, JUST LIKE MOST EVERYONE ELSES’S data. 

But twenty years worth of it. 

What can you do with that?


I’m just one person, with a liberal arts degree, and even I could figure this out. 

Here are the 7 things you need to do:

  1. Think about it, first. 
  2. Identify what you would LOVE to do if you had the right data.
  3. Review what you DO have. 
  4. Pare down the dream. 
  5. Pare down the next steps. 
  6. Call it a pilot (psychologically this helps get you started and able to take on the smaller chunks, putting the need for the big picture aside, for now, to test to see if it’ll work.)
  7. Get started. 

That was super helpful, right? 

Yeah, I know, I’ve read the same vanilla sales articles you have that whet your appetite for a solution, get you excited, and leave you with bland general info that makes you feel less equipped to do anything. So I’m not going to be that marketer. 

First off, this won’t be your path. Every business, need, goal, and personnel make up is different, so there isn’t a one size fits all, but you’ll hopefully get a spark to start to see how you can apply this to your unique situation. 

How to Use the Data: Trends

I decided I wanted to know who, when, and how to communicate to customers, what messages to send them and when, to increase sales to that audience. 

Here’s what I planned out:

  1. I looked at our biggest ROI customer types/personas, in our case, B2B.
  2. I acknowledged but essentially IGNORED the B2B/B2C cross over and further focused on one single segment of the B2B audience. (In our case, builders.)
  3. I pulled the data we had that included the contact and purchase history. 

I want to pause here and explain something. I was NOT looking for reporting level data, I was looking for trends to build on. The ACCURACY of the data was important but not as critical because it was the trends that mattered, not the results. 

Continuing on ….

Here’s what I knew I needed to understand

  1. How long between “successful” contact to sale. This is tricky, because not all contacts are successful. Often they’re just reminders that we exist. A B2B customer wasn’t going to buy our custom product unless they had a specific customer for it, so the generalized (yet targeted) marketing communications weren’t what I was looking at here. 
  2. What time of year did they actually purchase. 
  3. What was going on in the world (economy) during blocks of time that would affect those purchases (this was a “pin this” thing as it was critical specifically to our product. Many of the sales we made were opportunistic during this time period and while the sales themselves were not the objective of this exercise, this information *might* skew the time to purchase trends. I made sure to tag these for further review, because let’s be honest, the economy is also cyclical and the information is useful when similar situations arise again, and also helpful in better planning the baseline. 
  4. Chart it and normalize it — because life is about outliers that aren’t planned for, so they shouldn’t be accounted for (highs or lows) in your expectations. They are part of the contingency. This entire exercise is basically about planning for a more consistent sales funnel, to offset the outliers and account for seasonality.
  5. Marvel and what you just figured out. 

WTF did you just figure out?

  • I saw how long it took for buyers to buy in season, and it was consistent. I broke it into smaller groups, including things like region, small businesses, repeat buyers, and saw the time to sale was consistent across all segments. 
  • I saw regionally when the contacts sped up due to seasonality, and when they slowed down. 
  • I saw how many contacts between the outreach to purchase it took, along with the steps along the way. 
  • I saw who made the first contact, us or them. 
  • I saw the trends. 

Cool. Trends are Awesome. 

Still just an idea though. 

Now what?

Reminder, I’m a liberal arts major. 

I was wearing more hats than Bartholomew

So rather than forge ahead, I knew I’d need help, additional resources and support to put together what could be a veritable controllable pachinko machine to pull the levers and normalize consistent sales across the year, despite varied audiences and seasonality. Using this data on one segment, we could pull for all segments, and stage out marketing planning and sales efforts to reach whoever was buying when, and plan all schedules around full on customer behavior. 

And this was all with cruddy, old, messy, unreliable data. But it showed trends. 

So that’s what you can do with data, even limited, messy data. 

What did we do with it? 


Well, not nothing, I definitely used it in my own marketing planning. I developed timelines and customer workflows based on what I had found out, and based on that timing scheduled B2C marketing to coincide and convince THEIR customers when to buy, and set workflows internally about when to contact and with what messages. And sales increased, time to purchase decreased, and AOVs increased. So not nothing. 

But the machine never got approved.

The powers that be didn’t “agree” with the data, said, I shit you not, “Data is Defensive”. 

Apparently, data insecurity isn’t just about the data.