KPIs, reporting obligations, and claims to verify

Commercially sensitive. Contains retainer economics and the breakeven argument used in pitches. Internal use only.

Source: ../source/JSGlobtech_GEO_Agency_Playbook.pdf §3 (tier deliverables) and §4 (pitch script), extracted 2026-07-30.

Client-facing reporting commitments

These are contractual-feeling promises embedded in the pricing tiers. Every tier sold creates recurring work — see ../operations/client-engagement-sop.md.

Commitment	Cadence	Tiers	Source
Citation reports — where the client appears across ChatGPT / Gemini / Perplexity	Monthly	1, 2, 3	§3 Tier 1 deliverables
Hosting & monitoring of the deployed MCP server	24/7 (continuous)	2, 3	§3 Tier 2 deliverables
Competitor displacement strategy	Bi-weekly	3 only	§3 Tier 3 deliverables

Implications worth stating plainly:

• Monthly citation reporting is the core recurring deliverable that justifies the retainer on every tier. Without it, Tier 1's retainer has no visible output.
• Tier 3's bi-weekly cadence means two touchpoints a month on displacement alone, on top of the monthly citation report. That is roughly a 3× reporting load versus Tier 1 for a 4× retainer — check that the margin actually holds before selling it.
• "24/7 hosting & monitoring" is sold on Tiers 2 and 3, but this repository has no monitoring, alerting, uptime checks, analytics SDK, or CI (verified absent, see ../architecture/README.md). Selling a monitoring SLA without tooling is an operational risk. Resolve before the first Tier 2 close.

Measurement problem

The primary KPI — "does the AI cite the client?" — is inherently unstable:

• LLM answers are non-deterministic; the same prompt yields different citations across runs and sessions.
• Results vary by model, model version, personalisation, and whether web browsing is active.
• There is no vendor-provided rank API equivalent to Search Console.

So a citation report needs a fixed, written methodology to be defensible: a frozen prompt set per client, the specific engines and modes tested, a fixed number of runs per prompt, and a citation-frequency count rather than a binary yes/no. Without that, month-to-month deltas are noise and the client will eventually notice.

No such methodology exists in the playbook or in this repo. It is the largest missing operational artifact.

Breakeven framing

The pitch's economic close (§4):

"If this brings you just one extra client a month, the entire system pays for itself. Everything beyond that is pure profit."

Checked against ./target-markets.md deal values and ../sales/pricing-tiers.md retainers:

Tier	Monthly retainer	One-client value (lowest targeted vertical)	Holds?
Tier 1	$1,000/mo	$3,000 (one car rental)	Yes, comfortably
Tier 2	$2,000/mo	$3,000	Yes
Tier 3	$4,000+/mo	$3,000	No at the low end — needs a clinic or property close

The argument is sound for Tiers 1 and 2 across all four verticals. For Tier 3 it only works in cosmetic surgery, real estate, and dental. Do not use the one-client-a-month line verbatim when pitching Tier 3 to a car-rental prospect.

Also note the framing ignores the one-time setup fee. "Pays for itself" is true against the retainer, not against setup + retainer in month one. It is fair as a steady-state argument; be ready to defend it if the buyer does the arithmetic.

⚠ Unsourced claim — verify before reuse

The playbook asserts, twice, that:

"Over 40% of high-intent clients now research treatments using AI search engines instead of Google" (§4 pitch script and §5 email template 1)

No source, date, sample, geography, or methodology is given. It appears in the pitch script and in cold outreach that goes to doctors and business owners — i.e. it is the most externally-visible number in the whole playbook and the most likely to be challenged.

Before this is used again:

• Find a citable source, or replace it with one that exists (published surveys on AI-search adoption exist; pick one with a date and a method).
• Confirm the scope matches the claim — UAE? high-net-worth? healthcare specifically? "Skipping Google completely" is a much stronger claim than "also using AI search".
• Add the citation inline in ../sales/pitch-script.md and ../sales/outreach-templates.md once found, or soften the wording to something defensible.

Until then, treat it as unverified. Sending an unsourced statistic to a clinic director is a credibility risk that costs more than the statistic buys.

Internal KPIs (not in the playbook — proposed)

The source has no metrics for JSGlobtech's own funnel. Worth tracking from the first outreach batch, since the numbers determine whether this GTM motion is viable at all:

• Cold emails sent → demo booked (the templates promise a 5–10 minute demo, so this is the real conversion step)
• Demo held → closed
• Which of the four verticals actually replies
• Retainer churn after month 3, when the novelty of the first citation report wears off
• Delivery hours per tier per month, versus the retainer — the only way to catch Tier 3 margin erosion early