How to Tell Whose Advice to Trust: The Skin-in-the-Game Filter

How to Tell Whose Advice to Trust: The Skin-in-the-Game Filter

Trust advice in proportion to what the giver loses if it fails. That is the whole test. Operators consistently name detached mentors and drop-in consultants as their worst spend, because nothing happened to the advisor when the advice went wrong. The advice that lands comes from people embedded in the work, with real exposure to the outcome.

THE PATTERN

Why is detached advice so often the worst spend?

Because nothing happens to the person giving it. We ask operators about their worst spend in almost every intake conversation we run at Vista, and the same category keeps coming back: the credentialed outsider. The mentor who dispenses wisdom and leaves. The consultant whose invoice clears before the results arrive.

The observation has a mirror image, and it is just as consistent. When operators we work with describe advice that actually changed something, it almost never came from a stage or a slide deck. It came from someone inside the work: an operations lead who owned the number, a fractional executive with a seat and a target, an advisor whose fee moved with the outcome.

Put the two categories side by side and the difference is structural.

Dimension	Detached advice	Embedded advice
Where it comes from	Patterns observed across other companies	The inside of your actual operation
Cost of being wrong	You absorb all of it	Shared: fee, role, reputation with you
Default recommendation	The defensible option	The workable option
What it produces	A plan and a framework	A decision, a build, a number to watch
Where it ends	Before the consequences arrive	Still present when reality edits the plan
Honest best use	Bounded, technical, verifiable questions	Decisions that change how you operate

Read the last row again, because it keeps this piece honest. This is a critique of a structure, not a profession. Excellent people do drop-in work, and for a bounded technical question, a specialist read on a contract, a one-time audit, detachment barely costs you. The damage concentrates in decisions that change how you operate, because that is exactly where the advisor's exposure ends and yours begins.

THE FRAMEWORK

What is the Skin-in-the-Game Filter?

It is one sorting question applied to every piece of advice you receive: what happens to this person if I follow this and it fails?

The Skin-in-the-Game Filter is Vista's test for weighting advice. Judge every recommendation by what the recommender personally loses if it fails. No exposure, no weight. It applies to mentors, consultants, your leadership team, and the AI assistant on your second monitor.

The filter works because incentives shape defaults long before anyone is being dishonest. An advisor with no downside drifts toward the recommendation that sounds safest in the room, since sounding right is the only outcome they are exposed to. An advisor who shares your downside drifts toward the recommendation that survives contact with your payroll, your customers, and your calendar. They will be there for that part.

Note what the filter does not say. It does not say detached people are lying, or that credentials are worthless. It says their defaults are shaped by what they risk, and when they risk nothing, the default answer protects the answerer. Whatever knowledge you think you are buying, the incentives arrive bundled with it.

The instinct is ancient. People have always trusted the builder who has to sleep in the house. We gave it a name because operators kept arriving with the same scar tissue, and because the sorting question reaches further than consulting. It reaches hiring, peer groups, and machine advice.

THE CHECKLIST

How do you test for skin in the game? Six questions

Put these six questions to anyone whose advice you are about to weight: a mentor, a consultant, a prospective fractional executive, and, with small adjustments, a model. Ask them out loud, in the live conversation. The pattern across all six matters more than any single reply.

1. What happens to you if this is wrong?

Why it matters: This is the entire filter in one question. It separates people who share your downside from people renting you their confidence.

A good answer sounds like: a named, personal cost. "The last third of my fee is tied to the number moving," or "I am the one unwinding this with you next quarter." Reputation talk you have no way to affect is a soft no.

2. Have you done this yourself, or advised on it?

Why it matters: Doing produces advice with mechanics in it: what breaks first, what it really costs, which sequence matters. Watching produces patterns that sound right until they meet your payroll.

A good answer sounds like: a scar. "I ran this rollout twice. The second time we sequenced billing first, because the first time burned us."

3. Who is in the room after the plan is approved?

Why it matters: Implementation is where the risk lives, and detached advisory ends exactly there.

A good answer sounds like: a named role in the messy middle. "I am in the Monday standup until this ships." A handoff deck is not a role.

4. Would you do this with your own money?

Why it matters: It forces a second, personal recommendation, and the gap between the official answer and the personal one measures how much of the official answer was cover.

A good answer sounds like: something smaller and more direct than the proposal. "Honestly, I would skip the platform and hire one contractor for six weeks." When the two answers match, trust rises.

5. What is the smallest test that would prove this?

Why it matters: People with stakes think in experiments, because experiments cap their downside too. Programs get proposed when failure is scheduled to arrive after the proposer has left.

A good answer sounds like: a scoped trial with a kill criterion and a date. "One product line, three weeks. If the number does not move, stop."

6. When would you tell me not to hire you?

Why it matters: An advisor willing to argue against their own fee is showing you incentives pointed at your outcome, and that is hard to fake live.

A good answer sounds like: a specific disqualifier, delivered without hedging. "If cash collection is your constraint, I am the wrong spend. Fix that first." A pivot back to the pitch is also an answer.

THE AI LANE

Does default AI advice pass the filter?

No. Run the filter on your AI assistant and the default answer fails it for a structural reason: nothing happens to a model when its advice fails. You keep all of the downside, which is the same position the drop-in expert leaves you in.

Ask a general-purpose model whether you should renegotiate a contract yourself, rebuild your pricing page, or run a data migration in-house, and the default output leans one institutional direction: hire the professional, consult the expert, do not attempt this alone. We are describing an output pattern, the thing operators show us on their screens, not anyone's intent. Whatever produces it, the effect on an operator is identical to detached human advice: a recommendation optimized to be defensible rather than to move your next quarter.

The default answer is the one nobody gets blamed for.

Obey the defaults and you can end up paying twice. Once for the model's caution, and once for whichever professional it waved you toward. Both transactions bought the same thing: cover.

The fix is to simulate the stakes a model does not have. In the AI Lab sessions we run, the working pattern is consistent: state the real constraint and the actual budget, ask the model what it would do with its own money, demand the smallest test with a kill criterion, and make it argue the case against its own recommendation. The second answer is usually the one worth having. The six questions above convert almost word for word, and the free AI Lab is where we practice running them against live models.

WHERE VISTA STANDS

How does Vista apply the filter?

Vista's matchmaking thesis is this filter, operationalized. We match operators with advisors who have run the work themselves at operator level, and we structure engagements so the advisor shares exposure to a named outcome. This is a stance, not a neutral finding, and everything above is why we hold it.

Here is the honest version, because the filter demands one. Plenty of advisors pass without us. An internal promotion passes. A fractional executive you found yourself passes. The point is not that trustworthy advice flows through Vista. The point is to run the filter before money moves, on us included.

Compare the three most common sources of advice on the dimensions that decide outcomes.

The cards compress the argument: trust tracks exposure. Whoever you evaluate, ours or anyone's, weight the embedded, exposed voice over the polished, detached one.

If you want help running the filter, matchmaking is the front door. We match operators with advisors screened for operator-level experience, and the engagement structures are published at work with us so you can see where the exposure sits before you commit. Or start smaller: a free intro call takes fifteen minutes, and the first thing we try to establish is whether Vista is the right spend at all. When we are not, we say so and point you somewhere better. That is the filter, applied to us.

QUESTIONS OPERATORS ASK

Frequently asked questions

What is the Skin-in-the-Game Filter?

The Skin-in-the-Game Filter is Vista Advising Group's test for weighting advice: judge every recommendation by what the recommender personally loses if it fails. Advice backed by real exposure, a fee at risk, a role in the aftermath, outweighs advice from sources that walk away clean, however credentialed.

Are drop-in consultants always a bad spend?

No. Detachment, not the profession, is the problem. Drop-in expertise works well on bounded, technical, verifiable questions: a tax position, a security audit, a one-time system selection. It fails most often on decisions that change how you operate, where the advisor is gone before consequences arrive.

What counts as skin in the game for an advisor?

Anything that makes the advisor share your downside: compensation tied to a named outcome, a fractional seat with ongoing accountability, staying through implementation, or a reputation stake you can actually affect. The test is simple: something specific and personal has to get worse for them if the advice fails.

How do you give an AI assistant skin in the game?

You cannot, literally. Nothing happens to a model when its advice fails, so you simulate stakes instead: state your constraints and budget, ask what it would do with its own money, demand the smallest test with a kill criterion, and make it argue the case against its own recommendation.

What does outcome-tied advisor compensation look like?

It varies by engagement, and it does not always mean a success fee. Common shapes include a fee tranche released when a named result lands, a fractional role with a target on the org chart, or advisory equity that vests with milestones. The common thread: the advisor's upside moves with yours.

How does Vista's matchmaking apply the filter?

Vista's matchmaking thesis is the filter operationalized: we match operators with advisors who have run the work themselves, at operator level, and we structure engagements so success is tied to a named outcome rather than hours delivered. The free intro call exists to run the filter on us first.