If you follow us on Twitter, you know we tweet a lot about data and creativity. A couple of weeks ago we began a poll to see why data analysis is so difficult in SMBs. 

The results were close, but one came out on top as the biggest reason.

Not having enough data keeps SMBs from being creative with machine learning. With tools like Obviously AI at their disposal to get predictions, not having enough data to make valuable predictions can be frustrating. 

SMBs need data in order to build and train accurate ML models to make predictions and be competitive with large companies that have a data science team. Imagine scheduling a paint night with your friends. You buy drinks, easels, paint brushes, but when you sit down to begin a masterpiece, you realize you forgot to buy the paint. You can't create a painting without paint. You can't create a machine learning algorithm without data.

This post will help you get off the ground with collecting free data and using it to make predictions.

Here are some reasons why you may not have enough clean data and what you can do to get it. 

You Aren't Utilizing a POS System

A point-of-sale system (POS) can funnel data that you can use to predict customer behavior. An example of a POS system is Square, a device you’ve probably seen at a food truck or bar. Small businesses can use a hardware device to swipe cards with a software interface to collect data on a purchase. 

Square says, “Unless you have a solid system for recording each sale electronically, it’s hard to have an accurate understanding of what you’re selling. Basically, the more data you have about your business, the better able you are to make informed decisions.” With a POS system like Square you can make predictions about inventory, dynamic pricing, and what time of day sees the least and most demand. 

While Square is particularly useful for retail, restaurants, or pop-up events, establishing some kind of POS system can easily collect data to use in a data prediction. A POS system connects software to brick and mortar services to collect data on material goods and customer actions. 

If you’re low on sign-up data, also consider creating a sign-up process like this that requires very basic, but useful information. Be careful not to make it too time consuming. 

Read more about making predictions for retail and SaaS here.

You’re Overlooking the Data You Receive From SMB Tools

If you’re a SaaS or some type of tech company, you probably have some sort of analytics from your signup page or app analytics. Like us, you’ve probably gotten option paralysis on what to use to make data predictions. 

As you know there are tons of tools to use—but the essential ones for running a startup provide a lot of data.

We start with Google Analytics. GA is a pretty obvious tool to get website and app analytics. Few know you can actually export the data and use it in Obviously AI. 

You can predict top traffic sources, days where you can expect to see the most traffic, and more. Google Analytics is a great, free tool to be creative with and use in tandem with Obviously AI. If you don’t have Google Analytics set up, you’re missing out on a great opportunity for free data. 

Another great resource is Typeform. We use Typeform to get to know our audience better. While this is more qualitative data, we can still use it and take action from it. 

A couple of weeks ago, we made a survey asking our blog readers what kind of data analyst they are. While it gave them insight, they also had a conversation with us about what kind of data expertise they had. This informs us on future content that would be useful to our readers. 

From this kind of data, we can create a persona of a typical blog reader. We can see that those that come to our blog are typically no-coders with some technical background in ML coding languages. 

Other tools you can export data from include Ahrefs (what we use for SEO), Mailjet, Intercom, and more. 

Bottom line: If you’re using a tool that collects data, see if you can export into a CSV file. From there, you can upload it into Obviously AI and make predictions. 

You’re Not Taking Advantage of Look-A-Like or Competitive Data

On our blog, we’ve been providing use cases on how to be creative with your data with links to datasets you can use. We get most of these datasets from Kaggle or Google Dataset Search.

With data sources like these, you can easily find sample data or competitive data to get an insight. Make predictions about the app store or an eCommerce marketplace to see patterns in app features vs positive reviews or predict the average price of an app in your industry.

For example, in this blog post, Tectonic gained direction for building a client app without any owned data. Dan, the founder of the digital agency was able to answer questions on potential user behavior using all publicly available data.

You can also perform tons of healthcare or demographic research when trying to get insights from publicly available data. There are endless possibilities with the datasets already available to you.

Inside Obviously AI, we even have pre-loaded sample data to get off the ground with. Login here.

We Will Add to This List as We Grow and Find More Ways to Help SMBs

We plan to make this post evergreen and add to it as we find ways to help small businesses find and use data.

As always, feel free to reach out to us on Twitter or our customer support chat on the bottom right of our site. 

We would love to talk!