How (not) to grow your startup: optimize the wrong thing
Focusing on what you like to do instead of attacking your growth bottleneck can distract you to death
In April 2018, I announced a new feature for users of the StackSource commercial mortgage marketplace. For “borrower” users (real estate investors looking for a loan on a property) that received multiple loan quotes in the marketplace, the portal would now allow them to run fancy Loan Quote Analysis to project their investment returns taking one loan quote vs another.
A lot of product and engineering work went into it (compared to our total budget and attention), because commercial mortgage quotes outline several factors that impact an investor’s cash flow.
Here’s the problem: selecting the optimal quote received from our marketplace of lenders was not the bottleneck for the business at that time. Our bottleneck was further upstream: reaching borrowers and removing the friction to receive quotes from the marketplace at all.
Why did my co-founder and I spend so much time on this feature then?
Bundles of bias
If the only tool you have is a hammer, every problem looks like a nail.
- Maslow’s Hammer
Let’s say you have a couple hours of unstructured time on your calendar, and you are motivated to advance the venture you’re working on. How do you wish to spend those two hours?
Your goals, your checklist, or an urgent request from an important stakeholder may dictate what you actually spend the time on, but sooner or later that bias for your favorite flavor of work will get its chance to take over.
Our work preferences are an inherently good thing to be explored and harnessed. We never do better work than when we are leaning into our strengths and getting into a flow state. But the problem with our bias comes in when we let our preferences steer the ship.
I’ve seen this in myself too many times, and I’ve seen it in others.
An engineer chooses to code when they should be stepping back and determining if they’d be better to leverage a simple plug-in.
A visionary product manager builds an ambitious and elaborate roadmap, without digging into today’s operational challenges holding users back on the current product.
A networker books the next conference when their messaging isn’t yet strong enough to build a lasting impression.
These all happen when we are led by our bias for the type of work we prefer to perform.
But there’s an antidote.
The right dose of analysis
Somewhere in every business, there’s the greatest bottleneck.
It could be the number of leads generated, the sales team, the onboarding experience, fulfillment issues… areas where bottlenecks can be hiding are particular to each business.
The job of a founder is not to find and fix any bottleneck at all, but the greatest one. The more complex the business, the more analysis it may require identifying it.
Finding bottlenecks usually starts by digging into the easily measurable stuff and the high-level metrics based on the type of business. The key metrics vary based on whether revenue is recurring or transactional. For a SaaS startup, top level metrics like churn rate and Customer Acquisition Cost (“CAC”) are typical starting points, while for a marketplace that will tend to be Gross Merchandise Volume (“GMV”), and transaction success rates. All sorts of businesses, not only tech startups, can also leverage a Net Promoter Score (“NPS”) to determine if their users/customers are happy with the experience of transacting with them.
The top-level metrics would then be broken down into their component pieces. For instance, Customer Acquisition Cost could be broken down into a Cost per Lead (“CPL”) and a Conversion Rate. Each of those components can be broken down further, so on and so forth. Analyzing goes beyond calculation, though. At a certain point, you need to judge whether these metrics are where they “should” be.
Coulds and Shoulds
When I ran StackSource, we viewed revenue as a pipeline of loan requests. With limited resources (vs competitors), we needed to find a few channels to drive commercial mortgage leads fairly efficiently. But the conversion rate was never where I hoped it would be.
How should that conversion rate have been addressed?
Well, first I needed to develop ideas of how it could be addressed.
There were many factors impacting Conversion, but let’s take a few key components for our example:
Clickers - Some percentage would click on and start filling out a commercial mortgage request in our web portal.
Requesters - A smaller percentage would finish filling out that commercial mortgage request.
Responders - A smaller percentage still would respond when one of our Capital Advisors reached out to discuss it.
There are more steps in the user funnel to get them to delivered value (and revenue for us), but you get the picture. Each of those three steps, and others, had their own mini conversion rate, which together determined our overall conversion rate.
We could spend our time and resources addressing any of the above. What we needed next was a concept of how much each mini metric is expected to improve based on
Rank by expected impact and cost
Maybe a better website, or hiring a Sales Development Rep would lead to an increase of Clickers. Maybe there was a simple feature we could build to increase the number of Requesters. Maybe an entirely new feature set would enable more Requesters to become Responders.
Knowing our current pipeline and conversion rates can help establish how much value could be achieved if we succeed in increasing the conversion rate at one of those pinch points. It’s just a bit of modeling in a spreadsheet to build that case, and a conversation as a team to determine a budget that would give the project a likelihood of success. The “budget” should consider days spent by employees, not just outside costs.
However, it would be a mistake to simply leave the potential value as the ranking criteria for which projects to spend time and resources pursuing. Just like a loan in our pipeline had a probability of closing, so a project to improve our user funnel had a probability of success. Well, even “probability” is too strong a word, as if it’s some roll of the dice. Each project should be multiplied by a Confidence Interval for how likely success may be. The growth ideas can then be laid out as a table like this:
or visualized like this:
Projects that have a high probability of success and a low budget end up getting priority in this model.
Fresh eyes
Despite our best intentions, sometimes we bake our biases right into the analysis itself. Typing a Confidence Interval into a spreadsheet is an all too tempting moment for Confirmation Bias (or blind optimism!).
That’s why we need outside perspectives.
I’ve found there are different voices I need to seek out, depending on the type of bottleneck I’m solving.
Is it a problem with my product? I’ll seek out founders who have solved similar problems.
Is it a problem with a person on my team? I try to talk with an advisor with both leadership experience and experience with me, because maybe I’m the problem!
Some have a bias to spend too long in analysis mode, of course, which leads to analysis paralysis. Not everything is knowable, and certainly not everything is measurable. Founders typically need to become comfortable making decisions and executing with imperfect information. While a larger incumbent may slow down to over-analyze or balk at risks, founders take the leap. Or, they’ve already taken the leap, and now they need to do their best to adjust mid-air.
When I’m stuck on this type of analysis, I like to tap on the shoulders of one or two key advisors/consultants adept in the type of project I’m trying to scope, or hold a “two pizza” size meeting of the most qualified members of the team to speak to an initiative. As the ancient biblical proverb says:
Without counsel plans fail, but with many advisers they succeed.
- Proverbs 15:22 (ESV)