Getting Started on an Automation Project - Part 2: Data Flows

A lot of business owners have heard business automation can dramatically save time and money, but they don’t know where to start. In the first post, I discussed creating an application inventory. In this post, I’ll talk about ways to help you identify automations using your application inventory.

Recall from our first post, that you should now have an application inventory. It might look like this:

App Used For Cloud? API? Notifications? Notes
QuickBooks Online Accounting
Autotask Operations and Project Management we already use the QBO/Autotask invoice automation
Toggl Time Tracking
Hubspot Marketing and Sales implemented Hubspot in last 6 months
MS Excel Sales Quotes maybe (see notes) Each rep has their own spreadsheet folder. Stored on Google Drive

You should also have a list of “pain point” scenarios written in business terms. Like this:

  1. At the end of the month, Larry spends two days copying time entries from our time tracking app into our accounting system.

  2. Our sales staff gets frustrated when talking to existing customers because their CRM system doesn’t have the latest contact and contract information for the account.

How do we use this to identify automation projects? What we’re going to do determine the “flows” of data that could lead to better, more automated processes. By identifying the flows of data, you can understand:

  • which applications are involved

  • what kind of data is involved

  • how much data is involved

These three pieces of information help to establish a level-of-effort for each flow of data. They will also be used to establish whether a flow is “worth it” to automate.

Identifying a Data Flow

Let’s begin with the first scenario. First, Identify the applications being used in the scenario.

Next, identify the corresponding applications on your application inventory.

Now, draw an arrow from the application where data originates, ending on the application where the data finally resides.

You’ve just identified your first data flow! From here, we can start writing down additional notes for this flow about

  • which applications are involved

  • what kind of data is involved

  • how much data is involved

Since we already know which applications are involved, so now let’s consider the kind of data, and how much of it there is.

What Kind of Data?

Go back to the scenario you’re working on and try to identify what data is involved.

We can see that Larry is working with time entries in the time tracking system (Toggl). He is entering time entry data into QuickBooks. At this point, you may want to ask “why”? This can shed further light on the process and identify additional important information.

It turns out, in our hypothetical business, Larry is entering time entry data into QuickBooks so it can be invoiced. Therefore, in addition to entering time entry data, he is actually also creating invoices in QuickBooks. You can’t create an invoice in QuickBooks without a customer, so that is also a kind of data we need to note. We should revise our scenario description:

How Much Data?

Now let’s progress to the “how much”. This can be a little difficult to determine, but take heart; for planning an automation, generally you only need order-of-magnitude estimates - not exact numbers. To come up with an estimate, it can be helpful to try to count transactions based on an individual user, and then extrapolate to the entire business.

To start with, you should talk to the people involved with the scenario directly to understand how much data they enter. They will be intimately familiar with how much work there is, and are the best sources of estimates.

After talking with Larry we find on average, each of our business’ ten consultants create about three time entries per day over the month. We also know that Larry typically performs his current process once a month, over the course of several days. Our estimate:

3 entries per day x 20 business days per month x 10 consultants


600 time entries per month

What about the data volume of invoices? Larry estimates this varies, but typically there are 4-6 customer projects a month, and customers are always invoiced monthly. Therefore, we can just use his estimate:

4-6 invoices per month

Add this to your notes on the flow.

Great! We’ve now identified the flow, and drilled down to get quite a bit of additional useful information about the potential automation.

In my next and final article in this series, I’ll discuss how to use this information to estimate the value an automation to the business and identify the kinds of automation options available for the data flow.