The Two-Tier Grid Separates on Named-Human Judgment, Not on Counts — and the Offer It Prices Was Killed by the Founder on 2026-05-11

The question

Verbatim: "What scope dimensions (report cadence, # of data sources, # of dashboards, refresh frequency, analysis depth) most cleanly separate a $2-3k tier from a $4-5k tier for Client Reporting specifically? Build the actual two-tier scope grid."

Direct follow-up to [[2026-06-13-managed-ai-data-services-pricing-models]], which settled the structure (flat scope-tiered retainer, $2-5k/mo, no usage metering) and explicitly deferred the scope grid to this brief. That parent brief is not restated here.

Read the status caveat before the grid. The grid is built and it is usable. It is also a contingent artifact: it prices a service line the founder personally archived on 2026-05-11, behind a contract gate that has been open since 2026-04-16 and was still unread as of 2026-07-31.

What we already know (from the vault)

• The founder killed "Client Reporting" as a productized RDCO service line on 2026-05-11, via an explicit HQ click-back (ARCHIVE + typed rationale, 13:14 ET). Two reasons: retainer work doesn't scale with his hours, and — the load-bearing one — his phData day job is mid-market data-platform consulting, so selling the same shape to the same buyer is a conflict. The decision-log states the re-entry condition in the founder's own terms: if it returns "it would need to be re-framed (e.g. as a SaaS / self-serve dashboard product, not consulting)." [[2026-05-11-mac-pivot-retainer-to-info-product]]
• The blocking gate is a document nobody has read. The 2026-04-16 one-pager flagged "Moonlighting clause check — phData's employment agreement may restrict outside consulting work. Need to read it before any client pitch goes out." The 2026-07-31 brief found that item still open 3.5 months later and named it the gate on all paid RDCO client work, cash or equity — while separately retiring a "phData single-client-concentration guardrail" that four briefs had been citing as fact and that had no source at all. [[2026-07-31-fde-equity-kicker-retainer-pricing-shape]]
• Much of the grid this question asks for already existed in the vault before the question was posed. The April one-pager carries a three-row price table: Pilot (free 30 days), Standard $2,500/mo — "one report cadence, up to 5 data sources, MAC validation, snapshot archive," Multi-account $1,500/mo per extra account, and Founder advisory $7,500/mo = Standard + one 90-minute strategy call. The separator between $2,500 and $7,500 in the founder's own first draft was one live human conversation — nothing else changed. [[2026-04-16-client-reporting-automation-one-pager]]
• The unit economics that constrain any grid are already measured. Per-client ongoing labor <2 hrs/mo (≈30 min narrative QA, ≈15 min threshold tuning, plus rare "this number looks wrong" follow-ups); one-time setup 8-15 hrs. [[2026-04-16-client-reporting-automation-one-pager]]
• The differentiator to protect is trust and accountability, not narrative generation. The last-mile narrative gap is real but contested — vendors already market "AI drafts your monthly management report." The durable half is "a number you can defend in front of a board," i.e. the MAC validation stamp + reproducible lineage. Scope on that axis or the wedge commoditizes within a few quarters. [[2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge]]

What the web says

• Nobody publishes a scope grid at this price point — that absence is itself the finding. Three targeted searches and two fetches for two-tier managed-reporting scope tables returned zero published service grids in the $2-5k band. The buyer's guide that ought to contain one is explicit that "the price tag matters less than the pricing structure" and then supplies no numbers, thresholds, or tiers at all (sweent, How to Price Data Analytics Services). Every tiered grid that is published belongs to the software layer, not the service layer.
• Concrete market anchors for the adjacent shapes: a single dashboard on "existing, mostly clean data" is $3,000-$12,000 one-time; a reporting suite of "several connected dashboards" is $12,000-$30,000; a warehouse build is $25,000-$60,000+; an ad-hoc analysis answering one question is $2,000-$10,000. Ongoing retainers land $3,000-$15,000/mo, where "a few days a month sits at the low end, near-continuous support at the high end" (sweent, How Much Do Data Analytics Services Cost).
• Source count is not the cost driver; source disagreement is. "Reconciling five systems that disagree with each other is where budgets disappear," and "most of the work happens upstream, getting the data trustworthy" — named as the most common budget leak (sweent, ibid.). This directly attacks the intuition that "# of data sources" is the natural price axis.
• Interactivity is a hard cost cliff, not a gradient. "Static monthly reports cost significantly less than live, filterable, self-serve" tools (sweent, ibid.) — meaning self-serve is a different product with a different price, not a higher tier of the same one.
• The software layer already gates on counts, and it is two orders of magnitude cheaper. AgencyAnalytics runs four tiers from ~$79/mo (5-client minimum) with 130+ native integrations, gating API access, advanced dashboards, and custom metrics behind higher tiers (Databox comparison, ReportingNinja pricing breakdown). TapClicks runs four AI-reporting tiers with no published prices.
• Cadence tiers in the agent-written-reporting literature stack at four levels, and only two of them produce a client-facing artifact: daily anomaly checks (agent-only, no output), weekly internal digest, monthly client-facing narrative (the primary artifact), quarterly QBR sections. The same piece puts manual reporting at 5-10 hrs/client/mo vs. ~20 minutes of human review with an agent pipeline (Digital Applied) — corroborating RDCO's own <2 hr/mo figure and confirming the labor is in the review loop, which cadence multiplies.
• Fixed-price scope failure has a named mechanism: "A fixed price on a vague request quietly becomes a change-order machine," and outcome pricing fails specifically where the result "doesn't depend on whether you act on what the data shows" (sweent, How to Price). The tier boundary must therefore be countable or calendared, never adjectival.

Convergences and contradictions

• Convergence, and it inverts the question's own premise. The question offers five candidate axes; the market evidence says three of them (sources, dashboards, refresh) are exactly the axes the $79/mo software layer already gates on. Tiering a $2,500 service on the same counts makes the offer legible as a tool and hands the buyer a reference price 30x below yours. Meanwhile the founder's first draft grid separated $2,500 from $7,500 on one live conversation and nothing else — the vault's own instinct and the market's cost structure agree that the price separator is human judgment, not inventory.
• Convergence on where labor actually lives. RDCO's measured <2 hrs/mo and Digital Applied's ~20 min/client both put the recurring cost in the human review loop, not in ingestion or rendering. Cadence multiplies that loop linearly; source count does not touch it after setup. So cadence is a legitimate secondary lever and source count is not a lever at all.
• Contradiction between the research chain and the decision record. [[2026-06-13-managed-ai-data-services-pricing-models]] and [[2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge]] both treat Client Reporting as a live offer to be priced and sharpened. It was archived by founder decision five weeks before the first of them. Neither brief cites [[2026-05-11-mac-pivot-retainer-to-info-product]]; the 06-15 brief cites the April one-pager, whose status: field still reads drafted-for-founder-review. This is the status: proposal laundering pattern from feedback_workflow_agent_output_integrity, and this question is the third generation of it. A direct founder decision outranks a research-brief inference; the kill stands until he reverses it.

Synthesis for RDCO

The primary price separator is named-human judgment and accountability — whether a person commits to a recommendation and shows up live to defend it. Everything else is either a secondary multiplier or a guardrail that should never appear as a price lever. Three independent lines converge on this. The cost structure: after setup, marginal source, dashboard, and refresh cost round to zero, while every hour of synchronous or interpretive human time is a real draw on the binding constraint (founder attention). The competitive structure: counts are what the software layer sells, so gating on counts re-anchors the buyer to a $79-$500/mo reference price. And the differentiation structure: the only half of the wedge that doesn't commoditize is "a number you can defend in front of a board" — which is a human accountability claim, so it is the correct thing to charge more for. Report cadence is the clean secondary lever because it is the one dimension that multiplies the recurring human review loop, it is calendared (not adjectival) so it can't become a change-order machine, and the buyer feels it directly. Refresh frequency should be a fixed included spec in both tiers, never a lever — a point-in-time monthly artifact does not improve when the pipe refreshes hourly, so gating it charges for something the buyer cannot perceive and signals "tool."

The grid

Two tiers only. Both prices are all-in monthly, 90-day minimum, entry via a paid onboarding fee credited 100% into the first months.

Three design notes on the grid. First, the price doubles and the labor roughly doubles ($2,500/2.5 hrs → $5,000/6 hrs); the effective rate dips ~17% at Tier 2, which is the deliberate value-exchange the buyer must be able to see, and is affordable only because the live session is also the retention mechanism. Second, the source overage rate is identical in both tiers ($250/mo) — this is what stops sources functioning as a covert price lever and keeps the tier decision on the judgment axis where you want the conversation. Third, there is no third tier. Below $2,500 the honest answer is a reporting tool at $79-$500/mo; above $5,000 the buyer is asking for an embedded engagement, which is a different offer at a different number, not Tier 3.

What puts a client out of scope entirely

These are disqualifications, not upsells. Each one breaks the model rather than stretching it.

1. No system of record to validate against. The MAC stamp is the entire differentiator; without an authoritative source to reconcile to, the product does not exist and you are selling a prettier dashboard.
2. Sources that can't be pulled by API or MCP connector — screen-scraping, on-prem behind a VPN, or a human emailing a CSV each month. This converts one-time setup labor into unbounded recurring labor and destroys the <2.5 hr/mo economics. Hard test: authenticated programmatic pull, or no.
3. The systems disagree and reconciling them is the actual job. This is the market's named budget-disappearing act. It's a $12,000-$30,000 data-engineering project, not a $2-5k/mo reporting retainer. Sell it separately or decline; never absorb it into a tier.
4. The buyer wants live, filterable, self-serve BI. Interactivity is a cost cliff, not a gradient — that's a $3,000-$12,000 dashboard build plus a tool subscription.
5. Unlimited ad-hoc questions. Structurally an hours model, which RDCO refuses. Tier 2 caps at one bounded request; a buyer who won't accept a cap is buying a fractional analyst.
6. Public-company disclosure work (earnings materials, SEC-filed figures). Solo disclosure liability. The durable target is private companies doing board decks and investor updates.
7. More than three distinct stakeholder audiences wanting bespoke cuts. Past three, a fixed artifact silently becomes consulting.

The status problem, stated plainly

Do not treat this grid as a live offer. It prices a service line the founder archived on 2026-05-11 for a reason that has not changed — his phData role sells into the same buyer with the same shape — and the contract clause that gates all paid RDCO client work has been flagged as unread since April and was confirmed still unread on 2026-07-31. This question reached the queue because two June briefs priced and sharpened the offer without ever citing the archive decision, and it is now the third generation of that error. The correct sequencing is: founder reads the five clauses named in [[2026-07-31-fde-equity-kicker-retainer-pricing-shape]] (Outside Activities, Conflict of Interest, PIIA, non-solicitation, corporate opportunity) → founder explicitly re-opens or confirms the archive → then the grid ships. Worth flagging as the more interesting reading of his own re-entry condition: he said Client Reporting could return as self-serve product, not consulting. A two-tier retainer grid is consulting by construction. If the offer is ever revived, the higher-leverage build may be the one that separates on judgment packaged as software rather than judgment sold by the hour of a founder who doesn't have hours.

Why this is in the vault

It supplies the scope grid that [[2026-06-13-managed-ai-data-services-pricing-models]] deferred as its first open follow-up, and simultaneously breaks the citation chain that carried the Client Reporting offer forward as live through two deep-research briefs after the founder archived it — making the next action a founder clause-read, not another pricing brief.

Open follow-ups

• Does the founder want to reverse the 2026-05-11 archive at all, given that he named phData conflict as load-bearing and that constraint is unchanged? Founder decision, not research — but the answer determines whether this grid is an asset or a museum piece.
• If Client Reporting returns as the self-serve / SaaS shape he specified, what does the tier grid look like when the primary separator (named-human judgment) is the one thing a self-serve product structurally cannot deliver? That is the genuinely hard version of this question and it is unanswered.
• Should the two June briefs ([[2026-06-13-managed-ai-data-services-pricing-models]], [[2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge]]) be annotated in place with a pointer to the archive decision, and should /deep-research gain a pre-flight check that greps the decision-log for an archive/kill record on the offer a question presupposes? Vault hygiene plus a concrete harness fix.
• Is the ≈17% effective-rate dip from Tier 1 to Tier 2 actually recovered in retention, or does Tier 2 just buy worse margin for the same churn? Needs one real client-month to test; unanswerable from desk research.
• No published $2-5k/mo managed-reporting scope grid was found anywhere in this run. Is that because the segment prices bespoke by default, or because it is genuinely thin — and would a published grid therefore be a differentiator (transparency as positioning) rather than a leak?

Related

• [[2026-06-13-managed-ai-data-services-pricing-models]] — the parent brief; settled the retainer structure and deferred this grid as its first open follow-up.
• [[2026-05-11-mac-pivot-retainer-to-info-product]] — the founder decision that archived Client Reporting as a service line, and named the re-entry condition (self-serve, not consulting).
• [[2026-04-16-client-reporting-automation-one-pager]] — the original offer, its $2,500/$7,500 price table, its measured labor figures, and the moonlighting-clause gate open since April.
• [[2026-07-31-fde-equity-kicker-retainer-pricing-shape]] — names the unread clause as the gate on all paid RDCO client work, and diagnoses the same brief-to-brief laundering pattern this brief found again.
• [[2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge]] — the last-mile wedge framing; why trust and accountability, not narrative generation, is the thing to charge for.
• [[2026-06-07-solo-fde-contract-structures]] — the paid-pilot-credited-into-retainer entry mechanic the onboarding fee row implements.
• [[2026-08-01-above-the-platform-tier-corroboration]] — the parallel finding that RDCO's operative number and its cited tier have been drifting apart (read via search snippet only, not in full).

Sources

Vault:

• [[2026-06-13-managed-ai-data-services-pricing-models]] — 06-reference/research/2026-06-13-managed-ai-data-services-pricing-models.md
• [[2026-05-11-mac-pivot-retainer-to-info-product]] — 01-projects/mac/2026-05-11-mac-pivot-retainer-to-info-product.md
• [[2026-04-16-client-reporting-automation-one-pager]] — 01-projects/services-offering/2026-04-16-client-reporting-automation-one-pager.md
• [[2026-07-31-fde-equity-kicker-retainer-pricing-shape]] — 06-reference/research/2026-07-31-fde-equity-kicker-retainer-pricing-shape.md
• [[2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge]] — 06-reference/research/2026-06-15-fpa-ir-tooling-gaps-managed-reporting-wedge.md
• [[2026-06-07-solo-fde-contract-structures]] — 06-reference/research/2026-06-07-solo-fde-contract-structures.md
• [[2026-08-01-above-the-platform-tier-corroboration]] — 06-reference/research/2026-08-01-above-the-platform-tier-corroboration.md (search snippet only)
• Memory: feedback_workflow_agent_output_integrity (the status: proposal laundering failure mode this brief re-identified)

Web (accessed 2026-08-03):

• sweent — How to Price Data Analytics Services: A Buyer's Guide (scope drivers named; "fixed price on a vague request quietly becomes a change-order machine"; no numeric tiers published): https://sweent.com/guides/how-to-price-data-analytics-services
• sweent — How Much Do Data Analytics Services Cost? (single dashboard $3k-$12k; reporting suite $12k-$30k; warehouse $25k-$60k+; ad-hoc $2k-$10k; retainers $3k-$15k/mo; reconciliation as the budget sink; static vs. self-serve cost cliff): https://sweent.com/guides/how-much-do-data-analytics-services-cost
• Digital Applied — Client Reporting in 2026: Agent-Written, Not Dashboards (4-level cadence stack; 5-10 hrs/client/mo manual vs. ~20 min with agent review): https://www.digitalapplied.com/blog/agency-client-reporting-automation-2026-agent-written
• Databox — 9 Best AgencyAnalytics Alternatives in 2026 (software-layer tiering; 130+ integrations; API/advanced dashboards gated by tier): https://databox.com/9-best-agencyanalytics-alternatives-in-2026-ranked-for-agencies
• ReportingNinja — AgencyAnalytics Pricing 2026 (4 tiers, ~$79/mo entry, 5-client minimum): https://www.reportingninja.com/blog/agency-analytics-pricing