14 minute read
The AI consulting market has a structural problem: the deliverable that's easiest to produce — a strategy deck — is also the deliverable that produces the least value. And because it's the easiest to produce, it's what most consultants default to.
We've talked to dozens of mid-market operators who went through an AI "transformation" engagement and came out the other side with a beautifully formatted PowerPoint, a vendor shortlist, and zero running systems. The consultants got paid. The AI didn't get deployed. The next consultant got called.
Here are five signals that you're about to fund another deck.
1. The proposal doesn't mention production
Read the scope of work carefully. If the final deliverable is described as a "roadmap," "strategy document," "executive presentation," or "recommendations framework" — with no mention of deployed systems, operational handoff, or production-readiness criteria — you're buying a deck.
Good consulting engagements define what will be running at the end of the engagement, not what will be documented. There should be a system, a workflow, or an integration that didn't exist before and does now.
2. Their team has never operated what they're advising on
Ask directly: "Can you show me an AI system your team has running in production — not for a client, but internally?"
The gap between theoretical AI strategy and operational AI deployment is enormous. Consultants who have never built systems that run in production under real conditions — with real edge cases, real data quality issues, and real maintenance requirements — will give you advice calibrated to the demo environment, not the production environment.
Firms that operate their own AI systems have a fundamentally different relationship with the constraints. They've been on-call when something breaks. They know what the maintenance tail looks like.
3. The timeline doesn't include a stabilization phase
Most AI consulting timelines look like this: discovery → design → build → deploy → handoff. The engagement ends at deployment.
This is a mistake. The first 60–90 days in production are when you learn everything the demo didn't show you. Data behaves differently at scale. Edge cases appear. The team that uses the system finds workflows the consultants didn't anticipate.
If the scope of work ends when the system goes live, the consultant has structured the engagement to avoid accountability for whether the system actually works in production.
4. Success is defined by deliverable quality, not business outcomes
Ask: "How will we know if this engagement was successful?"
If the answer involves words like "comprehensive," "actionable," "well-received," or references to the quality of the documentation — the consultant is measuring success by how good the deck is.
Production success looks different. It looks like: "The invoice processing time drops from 4 days to 6 hours within 90 days of deployment." Or: "The SDR team generates 40% more qualified pipeline with the same headcount by end of Q3."
Consultants who can't answer the success question in business outcome terms have never had to answer for their work after the check clears.
5. The payment structure doesn't survive go-live
The strongest alignment signal in any consulting engagement is payment structure. If the full fee is due before the system is in production, the consultant's incentives end at delivery.
Firms that are confident their systems actually work are willing to tie a portion of payment to production milestones. A structure like 30% at engagement start, 40% at system go-live, and 30% at 90-day post-production review creates alignment through the full cycle — not just through the presentation.
If a consultant won't accept milestone-gated payment, ask why.
The Question to Ask Before You Sign
Before you approve any AI consulting proposal, ask for a reference call with a client whose system is in production — not a client who went through the engagement, but a client whose system is actively running today. Ask that client: "Is the system still running? Who owns it? Has it required significant maintenance?"
The answers will tell you more than any proposal document.
At Perpetual Quest, every engagement produces something running in production. If you're evaluating AI partners, we'd be glad to take that reference call.