Acceptable Failure Rates in SMB Processes
Guesswork Unveiled, Issue #20
Guesswork Unveiled: Acceptable Failure Rates in SMB Processes
June 25, 2025 | Issue #20
Normally these Guesswork Unveiled posts for SMB operators are paywalled, but I’ve left this one non-paywalled. Please consider signing up to support my writing and to dig more in-depth into being a strong, intentional business owner.
At Blooma Tree, we’ve been thinking a lot about what our systems should look like.
As I’ve crossed the 3-year mark in my ownership period, and the business is now entering a size of business that feels like it can support true SOPs.
No one would accuse me of being a perfectionist, but there’s something about designing SOPs that feels like perfectionist bait for even the most lazy 80/20er out there.
The logician in me wants to map out an SOP as a perfect flow chart, capturing every potential eventuality.
If this, then that.
What is the perfect # of follow-up emails to clients to ensure nothing falls through the cracks?
What is the perfect van inventory checklist to make sure no equipment ever gets lost or stolen?
It’s easy to optimize to the nth degree, resulting in layers and layers of systems that become clunky. Anyone who has worked at a large corporation knows what the end result of that looks like — endless forms, endless procedures, every small item being run up the chain for approvals.
Many searchers are leaving corporate America behind, partially because they’re exhausted of inane rules & policies that seem to get in the way more than they support the organization.
This boils down to a core, underlying issue — every SOP has two potential failures, not one. Generally, system designers tend to focus on the more obvious failure point.
For example, think about a process to qualify a lead before sending out a salesperson.
If the lead qualifier person declines a good lead, no one will ever hear about it or even realize.
But if they accept a bad lead, the salesperson will complain to the sales manager, the sales manager to the operations manager, and so on. The failure outcome of accepting a bad lead will be far more obvious to the entire organization.
I think of this as a false negative (declining a good lead) versus a false positive (accepting a bad lead). The false negative failure has a real cost to the business, but system designers will focus on minimizing false positive failures. They often do not balance the relative costs of the two failure outcomes while putting together a process.
This is the underlying tension that leads to employees complaining about clunky systems or policies getting in their way more than supporting them. They’re speaking to an underlying failure outcome that is hard to define or quantify, but exists.
In other words, each time something goes obviously wrong in the business, you don’t have to take it as a signal that you need to build a new process to ensure it never happens again. It may just be an acceptable false positive.
So let’s run through a couple examples of how to quantify these acceptable failure rates.
Two Examples
Let’s take two different processes:
Hiring a new arborist for our sales team
Qualifying a new lead before sending a sales arborist out to give a quote
Let’s start by figuring out the cost of a success & failure of these systems.
When making a new hire, the potential teamwide impact is significant — one new person in a small business can meaningfully impact culture. In our company, if a sales arborist does a poor job of writing specs or estimating time, the tree crews end up in a tough situation trying to complete their job. The clients end up unhappy as well.
In other words, a sales arborist hire has significant downstream impacts that go beyond the usual hiring considerations of cost weighed against production.
On the other hand, when qualifying a new lead, the downstream impact is generally limited to one hour of one sales arborist’s time meeting with the bad lead.
In that scenario, the cost of an error is fairly low (to a point — send a sales arborist on a lot of bad leads and you’re burning morale).
But declining a lead that could’ve become a closed job is fairly expensive — we wasted the marking spend upfront on the generating the lead. We lost potential downstream business from their neighbors seeing our trucks, or the client themselves coming back to us. The total lifetime value of one good lead is high.
Structuring Systems
As a result, we have different acceptable failure tolerances for these two processes. Below is a way to structure the thinking outlined above.
Process: Hiring a Sales Arborist
Outcomes Possible:
True Positive: Good sales arborist on the team
True Negative: Bad sales arborist not hired (to be clear, I use bad as a simplifying word. This really means poor fit, not the right time, or any other reason a hire should not be made. It’s not a moral judgement)
False Positive: Bad sales arborist hired
False Negative: Good sales arborist not hired
In this process, the cost of a false negative is more palatable than the cost of a false positive. As a result, we will gear the system design more heavily towards avoiding false negatives.
In practice, that may mean a longer interview process, with more reference checks, or higher prior experience thresholds.
Process: Qualifying A Lead
Outcomes Possible:
True Positive: Our office team schedules a sales arborist for a good lead
True Negative: Our office team declines a bad lead
False Positive: Our office team schedules a sales arborist for a bad lead
False Negative: Our office team declines a good lead
In our business, one good lead can generate significant positive lifetime value. The cost of a false negative is high.
But sending a sales arborist out to a bad lead has a limited cost — the cost of a false positive is low. (Side note — if your sales arborist capacity is your #1 bottleneck in the business, then the cost of a false positive rises.)
As a result, you may design a system where your customer service reps are a bit more lax on letting bad leads slip through to ensure you’re not declining any good leads.
Conclusion
It’s tempting to build systems that endlessly guard against one failure mode, usually the most obvious. This leads to a “one-way ratchet” — processes that keep guarding against obvious failures, while silently strangling the company by all the invisible failures it is generating instead.
Take a look inside your company today — is there a process your team grumbles about? If so, can you define clearly what the false positive and the false negative failure outcomes are? And can you associate costs with each?
If so, you should be able to understand if the current policy is balancing those costs poorly, which allows you to make an informed decision as to whether it needs to be changed.
I hope this is helpful — I do my best to write when I have something useful to say, but that does lead to odd variances in my publishing cadence. But I really appreciate the Guesswork Unveiled subscribers for sticking around and supporting my writing! Please consider subscribing as well if you don’t already!
Feel free to reply to this email with any thoughts or questions.
Best,
Kaustubh // Guesswork Investing
(There’s no content below this, but I had to put the paywall somewhere, so it’s at the bottom.)