The Three Questions I Ask Before Taking On a Service Business

I run a small client roster on purpose. A handful of operators at a time, each with a real engagement, each getting work I’d put my name on. That model only works if I’m selective about who I work with, not because I’m trying to be exclusive, but because taking the wrong engagement is expensive for both of us.

These are the three questions I use to figure out whether an engagement is likely to work.

Question one: what does your call data look like?

This question tells me more than any other single piece of information.

Operators who can answer it, even roughly, understand their business as a system. They know approximately how many calls they receive, roughly what percentage convert, and whether they have a coverage problem or a conversion problem. That foundation makes every subsequent conversation more productive. We’re talking about data, not impressions.

Operators who can’t answer it usually haven’t looked. That’s not disqualifying on its own. Most operators are busy running the business, not measuring it. But it means the first step of any engagement is establishing baselines before we can measure anything we do against them. That adds time and complexity to the front end.

What disqualifies: an operator who has been actively avoiding looking at the data because they’re worried about what they’ll find. That’s a management problem that no AI infrastructure or SEO work can fix.

What good looks like: you have 30 days of call records you can pull, a rough sense of your booked-to-called rate, and you’ve noticed a specific pattern that concerns you (high voicemail rate, low conversion on callbacks, seasonal peaks you can’t staff for). You have a problem that’s defined well enough to build a solution around.

Question two: are you in the business or above it?

This is the buy-in question. It sounds philosophical but it has a practical edge.

Operators who are in the business, doing dispatch, fielding calls, and handling customers, can tell me in 20 minutes what their call patterns look like, what their dispatchers struggle with, what questions customers ask most often, and where the process breaks down. That ground-level knowledge is the raw material for building a voice agent or a workflow that actually fits the business. Without it, we’re building against a model of the business rather than the business itself.

Operators who are above the business, having removed themselves from day-to-day operations and now managing managers, often have the right strategic goals (efficiency, margin improvement, lead cost reduction) but don’t have the operational texture to make the build specific. An AI voice agent calibrated to a business’s actual call patterns requires knowing those patterns. If the owner doesn’t know them and the dispatcher isn’t part of the process, the build will be generic.

This isn’t a judgment on management style. Removing yourself from day-to-day operations is often the right business move. It just means the engagement requires a different structure, usually involving the dispatcher or office manager as the primary point of contact during the build, with the owner in a review role rather than a primary role.

What disqualifies: an owner who wants AI to fix a process they haven’t understood and isn’t willing to spend time on during the build. Voice agents and agentic workflows are built around specific processes. If no one on the client side can describe the process in detail, there’s no process to build around.

Question three: can I anonymize you in writing?

Every engagement I take becomes a potential case study or data point in a blog post. Not always. Some builds aren’t interesting enough to write about, or the results are too early to share. But the work I publish is anchored to real situations: specific call volumes, specific before-and-after comparisons, specific numbers I can defend.

I don’t name clients in public writing. The case study format I use, “a Metro Vancouver pest control operator” or “a national pay-per-lead agency charging $700 setup + $219/month,” protects client identity while keeping the data real. No one who doesn’t already know the client would identify them from the description.

What I’m asking in this question is whether the operator is comfortable with that model. The content I publish builds the trust that brings in the next client. That content needs to be grounded in real situations. An operator who needs everything kept completely off the record, no data, no before-and-after, no description of the work, makes it harder to do the kind of writing that builds that trust.

This rarely disqualifies anyone outright. Most operators are comfortable with anonymized, data-anchored content. The conversation usually takes 5 minutes. But it’s worth having before the engagement starts rather than after.

Why I run this filter

Three-question filters aren’t unusual in professional services. What’s unusual is being explicit about them rather than discovering the problems mid-engagement.

The engagements that have gone badly, for any service provider and not just me, tend to share a common structure: unclear goals, misaligned expectations about process involvement, or a client who thought they wanted X and actually needed Y. Running this filter at the start of the conversation doesn’t eliminate those risks. It reduces them, and it surfaces the information I need to scope the engagement correctly.

If you’re reading this and wondering whether you’d qualify: the filter is less strict than it sounds. The questions are really asking whether you know your business well enough to build something specific for it, whether you have enough operational involvement to make the build accurate, and whether you’re comfortable with the way I document work publicly. Most operators who’ve been running a serious business for more than two years can answer all three.

Frequently asked questions

What size of service business does Clearslope work with?

Operators with enough call volume to make measurement meaningful - typically 30+ inbound calls per month - and ticket sizes that make AI voice agent or SEO economics work. Below those thresholds, the per-engagement ROI is thin for both sides. The filter isn’t about company size; it’s about whether there’s enough signal to build something specific and measurable.

How long does Clearslope’s client evaluation process take?

The initial conversation covers these three questions and takes 20–30 minutes. If the answers look right, the next step is a discovery call to scope the specific engagement. The goal is to surface misalignment early rather than discover it mid-build, when changing course is expensive.

What does “anonymized” mean in practice for case study content?

Published case studies describe engagements as “a Metro Vancouver pest control operator” or reference specific metrics without naming the business. No one who doesn’t already know the client would identify them from the description. The operator gets to review any content that references their data before it’s published. This model has worked for every client to date; it rarely disqualifies anyone.

For more context on how I approach work, see About. For the services where these questions come up most often, see After-Hours Booking & Lead Capture, Agentic Workflows, and Local SEO.