There’s a cultural assumption that the best experts have an answer for everything, instantly. Walk into a meeting, get hit with a question, fire back the right answer in under ten seconds. That’s the dream. That’s what people seem to want.
It’s also one of the most expensive habits in mid-market growth.
When you’re building the growth engine for a company doing somewhere between $1M and $100M in revenue, the systems you touch don’t sit politely in their own lane. A “quick” fix to one automation cascades into a downstream billing or fulfillment flow. A small permission change silently breaks a sync nobody owns. A field someone added two years ago turns out to be the load-bearing wall holding up half the reporting. The systems are denser than they look, and the blast radius of a confident-but-wrong answer is bigger than most people realize.
This isn’t a systems specific problem. It’s a growth stack problem. The more your CRM, marketing automation, billing, and ops tools are wired together, the less any single change is actually a single change.
So here’s the case I want to make: the experts worth keeping around aren’t the ones who answer fastest. They’re the ones who say “let me verify that, I’ll come back with an actionable answer” and then actually come back with something you can act on.
The AI parallel
You can see this playing out in plain view with AI right now. Push a language model for an instant answer on something it didn’t take the compute to verify, and you get hallucination. Fluent, confident, structurally plausible, and wrong. The fluency is the problem, not the solution. It’s the same trap humans fall into when they’re under pressure to perform expertise they don’t have in the moment.
The models that are honest about uncertainty – that pause, that check, that say “let me look that up” produce better answers. The models that pattern-match to confidence produce a mess that someone has to clean up later.
People are no different. A consultant who blurts out “yeah just turn that automation off” without looking is doing exactly what a hallucinating model does. They’re predicting the shape of a plausible answer rather than producing a real one.
A Salesforce example
Salesforce is a good example here because almost nothing in it is local, and because it’s where I see this play out most. The same dynamic applies in HubSpot, NetSuite, Monday, or any other system with enough integrations to matter.
Ask “can we change this picklist value?” and the honest answer is: it depends on what flows reference it, what reports filter on it, what integrations are mapped to it, what the marketing automation sync is doing with it, and whether anyone wrote a formula field that hardcoded the old label two years ago.
That’s not five seconds of work. That’s ten minutes of work, minimum, and sometimes it’s an hour. Do it wrong and you find out three weeks later when a sync starts silently dropping records, or when a flow throws a DUPLICATES_DETECTED error nobody can trace back to your change.
The cost of getting it wrong isn’t symmetric with the cost of slowing down. Slowing down costs you ten minutes. Getting it wrong costs you a weekend of cleanup, an awkward call with the client, and sometimes a damaged record set that’s nontrivial to rebuild.
What a real answer actually looks like
When a client asks a complex question, the answer has three parts, and none of them are “I think so.”
- What they actually asked. Restated back, in writing, so we both know what we’re solving. The number of times a one-line Slack question hides three different unstated assumptions is genuinely funny once you start counting.
- What’s affected. What records, what processes, what downstream systems. This is the verification step. In a Salesforce context this usually means actually opening the org, looking at the dependency graph, and tracing the change.
- The recommended action. Not “you could do X or Y or Z” — pick one, say why, and tell them what to watch for.
That third part is where a lot of consultants lose people. Verifying without committing is just another form of stalling. The whole point of slowing down is to come back with a better recommendation, not a longer list of caveats.
Reframing the customer relationship
Here’s the part that matters to anyone hiring this kind of work: the expert who pauses to verify is doing more for you, not less. The instant answer is cheaper to deliver and more expensive to live with. The verified answer is the opposite.
That doesn’t mean every question deserves a multi-hour deep dive. Most questions don’t. A good integrator develops a sense for which questions are five-second answers (“yes, that field is on the Lead page layout”) and which ones are five-minute answers (“let me check what references this before we change it”). The skill isn’t refusing to answer quickly — it’s knowing which category you’re actually in.
What we should be pushing back on, as an industry and as customers, is the pressure to fake the first category when you’re actually in the second.
The simple playbook
If you’re on the expert side, when you hit a question that needs verification, the script is short:
“Good question — let me verify that and come back with an actionable answer. I’ll have it to you by [specific time].”
Three things are doing work in that sentence. You acknowledged the question. You committed to a verified answer instead of a guess. You gave a deadline, which is what separates “let me check” from “let me stall.”
If you’re on the customer side, when an expert gives you that response, recognize it as a feature, not a bug. The ones who always have an instant answer for everything are usually the ones leaving the biggest messes behind them.
The point
Speed matters. Nobody wants a consultant who takes three days to answer a question that needed ten minutes. But the pressure to compress every answer into the same window — instant, confident, definitive — is producing a generation of decisions that look fast in the moment and cost real money over the year.
The teams that win the long game are the ones who built a culture where “let me verify that” is a normal sentence.
