Which of these options actually fits us?
Choosing a Growth Partner
Managed engine, agency, fractional executive, platform or internal hire. What each is genuinely good at, and how to tell which one your situation calls for.
What should you look for in a growth partner?
Four things worth checking before anything else. Whether they define the ICP or inherit it from you, because a partner executing your unexamined assumptions will scale the wrong targeting. Whether they own an outcome or an activity. What you keep when the engagement ends, specifically the contact data and the CRM. And whether the people who sold you are the people who will run it, which is where most agency relationships quietly degrade in month four.
Why it matters
The fourth one is the least discussed and the most predictive. Senior attention during the sale followed by junior execution afterwards is the standard shape of agency disappointment.
Data ownership is the second. Many arrangements retain the contact database and sequence history, which turns leaving into a loss rather than a decision.
What to look at
- Who defines the ICP
- Outcome accountability versus activity reporting
- What you keep at exit
- Who actually runs the account
Worked example: four questions to ask on the first call
Ask who defines the ICP. If the answer is that they will work from yours, understand that you are buying execution against assumptions nobody has tested, and that execution will scale whatever is wrong with them.
Ask what you keep at the end. Specifically the contact database and the sequence history, by name, and get the answer in writing rather than in principle.
Ask who runs the account after signature, and whether that person is on the call. Senior attention during the sale followed by junior execution afterwards is the standard shape of agency disappointment, and it is visible in month four.
Ask what would make them tell you not to proceed. A partner with no such answer has not thought about fit, which means they will find your engagement a fit regardless of whether it is one.
None of these four questions are about capability, and that is deliberate. Capability is the easiest thing to verify and the least likely to be the problem. Almost every partner relationship that disappoints was staffed by people who could do the work. It disappointed because the ICP was inherited unexamined, the reporting measured activity rather than pipeline, the senior person disappeared after signature, or the data stayed behind at the end. All four are visible before you sign, and none of them appear in a capabilities deck.
Ask them of us too. We have answered all four on this site, which was the point of writing them down rather than waiting to be asked.
The VGL perspective
We would rather lose a deal to these questions than win one despite them. A partner who cannot answer the fourth one clearly is telling you something.
Managed engine, SDR agency or building in-house: which fits?
It depends entirely on what you already have. If you have a proven motion and an existing RevOps function, hiring is usually right, because ramp risk is low and the seat is the only real cost. If you have neither, hiring buys you the ramp risk as well as the seat. An agency fits when you need volume quickly and the ICP is already settled. A managed engine fits when the ICP is the thing that still needs settling.
Why it matters
The question most companies ask is which option is cheapest. The more useful question is which risk you are best placed to carry, because all three carry a different one.
Hiring carries ramp and retention risk. An agency carries targeting risk, because they execute your assumptions at volume. A managed engine carries less of both and costs more than doing nothing, which is the fourth option nobody puts on the list.
What to look at
- Whether the motion is already proven
- Whether a RevOps function exists
- Which risk you are best placed to absorb
The VGL perspective
We turn away companies whose motion is proven and whose ops function is strong, because hiring will serve them better and they tend to work that out by month three anyway.
Fractional CRO, agency, or an internal hire?
A fractional CRO owns the revenue system: ICP, positioning, motion design and the pipeline number. An agency owns activity inside a system somebody else designed. An internal hire owns both eventually, at full cost and after a ramp. The common mistake is hiring a fractional CRO too early, as leverage on a motion that has not been found yet. That work is still the founder's and it does not delegate.
Why it matters
Sequencing matters more than selection here. Fractional leadership applied to an unproven motion produces a well-run version of the wrong thing.
The reverse error is just as common: an agency retained before anyone has defined the ICP, which scales an untested assumption very efficiently.
What to look at
- Is the motion found, or still being found
- Who owns the number afterwards
- Whether the engagement designs or executes
The VGL perspective
Win first, then scale applies to who you hire as much as to what you automate. Founder finds it, operator systematises it, team scales it. Out of order it gets expensive.
Data platform or managed engine: what is the real difference?
A data platform sells you access to contacts and firmographics. A managed engine sells you conversations. Between those two things sits everything that actually takes effort: scoring for fit, detecting timing, writing the sequences, handling deliverability, qualifying the replies and routing what turns out to be real. A platform leaves all of that with you, and renews whether or not anyone did it.
Why it matters
The measurable version: a data license consumed at 20 percent is still billed at 100 percent. Utilisation is the hidden cost line and almost nobody tracks it.
A managed engine also fails visibly. If conversations do not appear you know inside a month. An underused platform can stay invisible for a year, which is why it renews.
What to look at
- What share of the license is actually used
- Who does the work between data and conversation
- How quickly failure becomes visible
The VGL perspective
We use best-in-class data tools inside the engine rather than building a database. The tools are not the product. What happens to their output is the product.
Should we build it ourselves or buy a platform?
Buy whenever your process is genuinely standard, which is most of the time. Build at the point where the workarounds have become the process, when everyone knows the platform half fits and people are quietly absorbing the gaps. At that point you are already paying for custom software in salary rather than in a build, and unlike a build, that bill recurs every month and never appears on a budget line.
Why it matters
The three-year arithmetic is the part that changes minds. A platform at $3,000 a month is $108,000 over three years and still half fits. A build is paid once and owned outright.
The counter-argument is real. A build you own is a build you maintain. That cost is genuine and it is smaller than most people expect once the scope is one system rather than a platform.
What to look at
- Whether workarounds have become process
- Three-year cost, not the monthly figure
- Who maintains it afterwards
The VGL perspective
Owned beats rented when fit actually matters. When fit does not matter, rent it and spend the attention somewhere it does.
How do you evaluate an AI growth vendor's claims?
Ask what the AI actually does, in one sentence, without using the word leverage. Ask which part is a model and which part is a person, because most AI go-to-market offerings are a service with a model inside it. Ask for the arithmetic behind any percentage. Then ask what would make them tell you not to proceed, because a vendor without an answer to that has not thought about fit.
Why it matters
Percentages without line items are the most common tell. A 60 percent cost claim means nothing until you can see what it is measured against and at which end of which range.
The second tell is a demo of the tool rather than a description of the outcome. If the product is the interface, you are buying software and staffing the rest of it yourself.
What to look at
- A one-sentence description of what the model does
- Line items behind any percentage
- What they would decline and why
The VGL perspective
We published our own cost arithmetic because the claim was unfalsifiable without it. Ask us the same four questions you would ask anyone else on this list.
Related insights
Long-form on this topic
Insight pieces expanding on these answers publish twice a month from October 2026. They will appear here as they go live.
First pieces publishing October 2026. In the meantime, the four concepts set out the thinking these answers are built on.
Start here
Find out where your own growth engine stands
The AI Growth Readiness check takes about three minutes. It scores you across six dimensions, places you on the growth maturity model, and gives a recommendation you can act on, including when the honest answer is not yet.
What you get back
- A score across growth dependency, unit economics and systems
- Your position on the growth maturity model
- A recommendation, or an honest not yet
- A concrete next step either way