Orchard Analytics provides technical data team services on flexible contracts. Reach out to Phil to learn more.

The software abundance problem created the need for data tools, and now -- surprise -- there’s an abundance of data tools.

It can be hard to judge the more granular pros and cons of each data warehousing, event tracking, ELT, and BI solution, let alone decide whether you need to invest company dollars on each one. There’s plenty of good writing on the tools out there.

Equally important as “what do I need” is  “when do I need it?” What are the actual signs that your team is approaching the need to invest in the coveted modern data stack? We’d like to spend some time on this question.

The right time is when you can’t answer questions on time

It might be time to move to a modern data stack when the effort required to answer business questions starts to drag down productivity.

Recently we met with a client’s accounting team to review their month-end process. “We pull from three main sources,” the associate said as she navigated into each portal and downloaded each software’s respective CSV.

We were excited to dive into her process, an endeavor we slowly realized was not going to happen because the files wouldn’t open. We sat there in silence as it buffered. The three files accounted for nearly half a gigabyte of space.

“This can happen,” she said. She minimized the window and we talked of other things. This same eight-figure business also has a month-end Google Sheet for performance and spend with hundreds of rows. Their team spends hours each month inputting revenue, COGS, and marketing spend/ROI just to see if what they are doing is working.

This is basic reporting, I thought to myself. If anyone wanted to run a more complex analysis through this method, most reasonable people would eventually throw their hands up, ask “why bother,” and find the closest happy hour as quickly as possible. And who could blame them? Downloading, saving, merging, over and over again; this kind of work is deadening.

Most growing teams will arrive at the point where they can *feel* they need BI. This might be because you’re consistently hitting row limits and generating a slew of CSVs that sit in some forsaken folder.

Tristan Handy, founder of Fishtown Analytics, the creators of dbt, wrote a great piece about the role of analytics tech at each stage of a company’s growth. Consider it carefully.

You have analyses worth doing

The decision to invest in a data stack should be motivated by a set of specific questions related to the customer journey that are prohibitively time-expensive to answer today. Think about what these questions might be.

Muse Operations has an excellent post making the point that it can be better to “start with the decision you intend to inform,” rather than occupy yourself with complex analyses just to dig up “stuff an agent could have told [you] anecdotally.”

An example question: what brings first-time customers back?

This broad question might lead to a tree of smaller explorations. Are there certain products that lead first-time customers to buy again? Are there products that deter? Are there certain ads that produce more repeat customers?  

Without any data technology in place, searching for these answers would be exhausting for an analyst and impossible for a business user. Why? Because they require pieces of information about a customer that are scattered across different systems that were acquired by each business unit.

I recall a very potent segment from Rob Giglio, CEO of Docusign, during a CMO Office Hours event back in May when everybody was attending webinars.

“As marketers,” he said, “we can't just measure orders or revenue or subscriptions. We are accountable for the entire customer experience. Total addressable market, awareness, some kind of engagement or intent, touchpoint, trials, meetings, buying process, a closed sale, and everything after that. Was the product used? Are clients happy? Are they loyal?”

A thoughtfully configured modern data stack would allow you, with relative ease, to find out what first-time shoppers purchase, whether those shoppers return, how long it takes them to return, and what they buy on their second, third, and fourth visits. It could also be saved, repeated, and filtered for product and customer segments or time series. Grant Winship and Sanjana Sen did a fantastic job visualizing this exact question at the recent dbt Coalesce conference.

Get your team together and find out what questions you want to answer, and use that to inform when is the right time to upgrade from manual sheets-based reporting, replete with repetitive downloads and merges, toward a more robust data stack.