In my prior life as a technology consultant. I implemented complicated technology projects for government agencies. The job included a selling component, which was essentially preparing technical responses to government RFPs (requests for proposal).

Here’s how government technology procurement works:

1. The government wants to build or buy a new piece of technology
2. It gets all of the stakeholders in a room and asks them what they need from a new system
3. They write down all of the requirements
4. Requirements are compiled into an RFP
5. The RFP is published to the vendor community
6. A bunch of vendors respond with their proposals of how their technology is the best fit for the given set of requirements. The response includes a high-level overview followed by a line-by-line response to every single requirement that the government asked for. If you assume that vendors “stretch the truth” in these yes/no responses, you’d be spot on.

If you’re at all unclear about how well this process typically performs, just refresh yourself on the royal cluster**** that was the healthcare.gov launch. And take a look at some of these stats. One of my favorite is a government executive from the UK: “Only 30% of our projects and programs are successful.”

So why does this type of software procurement go so poorly?

It turns out that large-scale software development is insanely complicated. Building a piece of software from a frankenstein set of requirements, then flipping a switch and hoping that it will come alive with a jolt of electricity is not a realistic approach to a such a complex problem.

Outside of the government, the past 10-15 years of software development best practices actually encourage short product cycle times, releasing early and often, and writing automated tests as you write code. Entire methodologies (lean, agile, scrum, etc.) have been created with these principles. The commonality between all of them is this: start small and then build as you go. Don’t try to boil the ocean right from the start.

Wait, wasn’t this article supposed to be about business intelligence software?

Right, let’s get back on topic.

I recently answered a question on Quora: What are most suitable business intelligence solutions for a high growth SaaS company?

The original question went into some detail. The asker wanted a huge list of functionality:

• ad-hoc and static reporting
• data warehousing & ETL
• statistical testing
• large number of built-in connectors
• advanced visualizations

Quite a list! It also went on to say that they wanted to do most of their dashboarding in Excel and knew most of their core KPIs already.

The person who posed this question was tackling their problem just like a government agency would and, while well-intentioned, will likely get similar results.

Historically, BI software is considered an “enterprise tool” and sometimes at RJMetrics we run into potential customers who have long feature checklists, similar to (but longer than) the one above. When this happens, our goal is to reframe the conversation. We’re proud of our capabilities and are confident when going feature-for-feature with any of our competitors, but this buying process is flawed. We want our customers and potential customers to think like the lean-mean-growth machines they are, and a little less like the US government.

Here’s the right way to think about purchasing business intelligence software.

How you should think about your upcoming BI purchase

If you’re a fast-growing online business, your goal is to learn as quickly as possible. You want to A/B test your website, A/B test your email marketing, test different promotions, ad campaigns, remarketing strategies, product prices, merchandise strategies, etc. You get the idea. You’re building a growth machine and your business intelligence tool is the single biggest asset in this process.

Your BI tool is the way you’re going to gather all of the data to figure out what works and what doesn’t. It’s going to show you how behavior changes by cohort of user, what the lifetime value is of different marketing channels and tactics, and the result of changes you made to your pricing. Your goal isn’t to pick a tool that will make everyone happy, but one that will enable you to build that growth machine.

So how do you pick the best business intelligence tool for your company? Pick the one that:

• …allows you to quickly and easily measure the results of whatever experiment you want to run.
• …is the easiest for business users to use. Business users need to be able to do the work themselves or else it will slow down the entire test-and-validate process!
• …can be implemented the fastest. The longer an implementation takes, the more time you’re taking away from learning.
• …requires the fewest human hours. When you’re growing, you’re always understaffed. If gathering data is takes a long time, your staff will avoid it.

That’s it. Those are the core requirements. Anything else is just a nice-to-have.

RJMetrics is the business intelligence tool of choice for an ever-growing number of clients because our platform is built for lean, intelligent, data-driven teams. But by all means, do the research yourself. Find the tool that’s the best fit for your business, just please don’t select your tool using the same antiquated process the US Government uses.