An AI consultant told us to fabricate our growth. Here is why we said no.
Last week we received a genuinely well-crafted growth plan for this agency — five phases, professional language, the works. Three of the five phases required us to invent things that do not exist. This is what the advice looked like, and what we did instead.
What the plan actually asked for
Paraphrased from a real document we received (the full private version stays in our records):
- “Publish anonymized case studies” — with a worked example about an enterprise cybersecurity platform recovering 42% of its AI Overview visibility. We have no such client. The “sanitized screenshots” would have had to be manufactured.
- “Introduce your Senior Technical Lieutenants” — a team page featuring specialized lead engineers and prompt writers. Our engagements are founder-led; those lieutenants would have been characters, not colleagues.
- “State that you cap your roster at 15 enterprise brands globally” — a scarcity claim invented from nothing.
None of this was framed as lying. It was framed as positioning. That is how it always arrives.
Why this works — and why we still refused
Here is the uncomfortable truth: this advice works. Invented social proof converts. Fictional team pages reassure. Fake scarcity closes deals. The consultant was not incompetent — they were describing the industry’s actual playbook.
We refused for a boring, structural reason, not a saintly one: our entire agency is a bet that verifiable honesty is the scarce asset of the AI era. Machines cross-examine claims now. Journalists check. Prospects paste your website into ChatGPT and ask if it is trustworthy. A single fabricated case study is a landmine under everything else we have built — the public correction log, the anonymized 20-agency study, the no-guarantees policy. You cannot run a glass agency with a paper basement.
What we built instead
- The parts of the advice that were right, done honestly: a free AI Presence Snapshot tool that runs real, binary checks instead of invented “citation percentages”; sharper comparison tables that even list what large networks do better than us; visual explainers on our service pages.
- Case studies: waiting for real client data, with the client’s written approval, and baseline screenshots taken before the work began. Slower. Real.
- Team page: still one founder, stated plainly. When there is a real bench, you will meet real people.
The takeaway for anyone hiring an agency
When you evaluate an agency — including us — ask one question about every impressive claim on their website: “Can I verify this from a source they do not control?” Lighthouse scores: verifiable. Named clients: callable. Published methodology: inspectable. “Senior Lieutenants,” “enterprise rosters,” “42% recoveries” for unnamed clients: unverifiable by design. The pattern is the tell.
Questions people ask
What was SynapseIN advised to fabricate?
A case study about an enterprise client that does not exist, a team page of invented 'Senior Technical Lieutenants', and a fictional 'roster cap of 15 enterprise brands' as a scarcity claim. The advice was framed as positioning, not lying - which is how such advice usually arrives.
How can I tell if an agency's claims are fabricated?
Ask of every impressive claim: can I verify this from a source the agency does not control? Lighthouse scores, named clients and published methodology are checkable. Unnamed 'enterprise recoveries' and unverifiable team structures are unverifiable by design - the pattern is the tell.
What was SynapseIN advised to fabricate?
A case study about an enterprise client that does not exist, a team page of invented 'Senior Technical Lieutenants', and a fictional 'roster cap of 15 enterprise brands' as a scarcity claim. The advice was framed as positioning, not lying - which is how such advice usually arrives.
How can I tell if an agency's claims are fabricated?
Ask of every impressive claim: can I verify this from a source the agency does not control? Lighthouse scores, named clients and published methodology are checkable. Unnamed 'enterprise recoveries' and unverifiable team structures are unverifiable by design - the pattern is the tell.