Money and reporting
A dashboard that reconciles to the bank
A channel-attributed dashboard plus a weekly executive email, built from whichever platforms the owner actually uses. The pipeline is documented end to end, from extraction through attribution to delivery, so every number on the screen can be traced back to its source.
- Who it is for
- Owners who have outgrown platform-native dashboards and want one screen where marketing spend, pipeline and accounting agree with each other.
- Deployed for
- Multiple Prospectr client deployments across trades, franchise and services verticals, plus Prospectr internal reporting. Described by vertical. Naming permission is not on file.
- Status
- Live deployment
The goal
One place where the ad platforms, the CRM, and the accounting system tell the same story. Platform-native dashboards each tell you their own version and none of them reconcile.
Attribution that survives scrutiny, meaning revenue attached to channel through the accounting system rather than through platform-reported conversions.
A weekly executive email, because a dashboard nobody opens is a very expensive screensaver.
Cost, against the human alternative
Built as part of a deployment or as skill hours at the published rate. The build is platform-agnostic by design, so the cost scales with how many systems you connect rather than with a per-connector license.
Anchor title: General Manager, plus a per-seat BI platform subscription. A GM in this market carries a typical fully-loaded cost of $80,000 to $90,000 a year including payroll tax, benefits, paid time off, ramp time, and turnover, and a meaningful slice of that week currently goes to assembling numbers by hand from three systems that disagree. That is a market range for the title, not a quote of anyone actual salary. We have not benchmarked a specific competing BI platform price and will not quote one we did not verify. The goal is not to replace that person. It is to put them in their most productive role. A GM who stops building the report and starts acting on it is the entire return here.
The dashboard is a static hosted artifact rather than a seat-metered SaaS product, which is the structural reason its cost does not grow with headcount.
The goal is not to replace humans. It is to put them in their most productive role. Nobody gets cut here. Your billing manager stops doing data entry and starts working collections. Your GM stops chasing paperwork and starts selling. The agent absorbs the repetitive half of a job so the person keeps the half that actually needs a person.
This is a typical fully-loaded annual range for that title, meaning base pay plus payroll tax, benefits, paid time off, ramp time before the person is productive, and the cost of refilling the seat when they leave. It is a market range for the role, not a quote of anyone actual salary.
Security
- Runs in your cloud account against your credentials. The data does not transit a vendor multi-tenant warehouse.
- Client-level dashboards are separated per client. A build for one client cannot read another client data.
- Credentials are referenced from a secret store by location, never embedded in configuration that ends up in a repository.
- Freshness is audited rather than assumed. A dashboard silently serving three-week-old accounting data is worse than no dashboard, so staleness is a monitored condition.
Output
- A hosted multi-tab dashboard with channel-attributed metrics.
- A scheduled weekly executive email, drafted rather than auto-sent until the owner approves the format.
- A documented pipeline: extractors, normalization, storage, attribution, calculation, delivery. Every stage is inspectable.
- A per-client configuration file, so the build is reproducible rather than a one-off someone assembled by hand.
Ongoing cost to maintain
- Extraction and refresh are deterministic scheduled jobs at no per-run inference cost. Only the written commentary consumes paid model usage.
- Hosting is static. There is no per-seat charge and no viewer license, so showing the dashboard to your whole team costs the same as showing it to one person.
The framework that keeps it safe and keeps it cheap
- Platform-agnostic from the first line. The operator picks which platforms to connect; the skill provisions the extractors. Nothing about the build assumes a specific vendor, which is what keeps you from being locked to one.
- Deterministic work goes to plain scheduled scripts, reasoning work goes to a model. Mixing them is how recurring reporting quietly becomes the most expensive line item in an AI budget.
- Freshness auditing built in, because the failure mode of a dashboard is not an error message, it is a confident wrong number.
- The pipeline is documented, so if you take it in house, it is a system rather than a black box.
The procedures behind this
Each of these is a written document with a plain-English section for a person, a machine-readable section for the agent, and a dated changelog.
- BI_Build
- bi-multi-brand-dashboard
- bi-dashboard-qbo-freshness-audit
- bi-dashboard-debug
- metabase
Where every claim on this page comes from
- Platform-agnostic build flow, the four standing outputs per invocation, and the per-client configuration path are specified in skills/BI_Build/SKILL.md.
- Freshness auditing exists as its own dedicated skill, skills/bi-dashboard-qbo-freshness-audit, which is the evidence that stale data is treated as a first-class failure mode rather than an afterthought.
- The deterministic-versus-reasoning cost routing rule is the standing internal CRON versus SYSTEMD rule in AGENTS.md dated 2026-04-28.
Questions people ask about this
How is this different from the dashboard my ad platform already gives me?
Platform dashboards report on themselves. They each claim the conversions they can see, which is why the numbers never add up across platforms and never match the accounting system. This build attributes revenue through your books, so the total on the screen is the total in the bank.
Do we pay per user to view it?
No. It is a hosted artifact in your own cloud account, not a seat-metered product. Showing it to your whole team costs the same as showing it to one person.
What stops it from silently showing stale numbers?
Freshness auditing is a built-in monitored condition rather than something you notice three weeks later. The dangerous failure of a dashboard is not an error, it is a confident wrong number, so staleness is treated as a real alert.
Is there a version of this in your business?
The discovery call is a working conversation, not a demo. Bring the process that annoys you most.