Assessment · 6 November 2026
Can the data be trusted? · Harbour Financial
Core data capabilities in place but significant work is needed. Targeted remediation required before AI adoption (Band C).
What can the data actually run?
The finding that lands
Counts computed from the data9,355 customers exist as two different companies across Microsoft Dynamics and Oracle.
€11,811,448 of revenue can't be reconciled
The same account is recorded under different names, IDs or addresses in each system. Each one is clean in isolation, so every single-system audit passes them.
Broken for AI: Any AI keyed on customer identity (pricing, churn, lifetime value, next-best-action) trains on a contradiction and ships confident, wrong answers.
16,683 of 73,800 customers in Microsoft Dynamics have no matching record in Oracle.
€11,811,448 of revenue sits in a cross-system blind spot
Forecasting and revenue AI can only see half the picture, so projections are built on a partial book.
3,690 internal employees from ADP are sitting in the Microsoft Dynamics contact list.
3,690 records mislabelled as customers
Segmentation, scoring and outreach AI treat employees as prospects, polluting every model trained on the list.
Use Case Readiness
Whether the data can run each standard AI use case, and the gap to close.
0 of 6 relevant use cases ready · tuned to Financial Services
Are you ready for the tools you're evaluating?
0 of 8 ready on today's dataName the AI tools on your shortlist and we score whether your data can actually feed each one. Independent by design: we sell none of them. Tools marked External data depend on inputs we do not read (call recordings, web or intent events), so we score the structured prerequisites and say so.
| Tool | Verdict | Readiness | Assessability | Fix first |
|---|---|---|---|---|
HubSpot AI Predictive Lead Scoring and Breeze CRM / Predictive Lead Scoring | Conditional | 45 | Fully assessable | Close the flagged data gaps first |
Celonis Process Mining and Execution Management | Conditional | 45 | Fully assessable | Close the flagged data gaps first |
Clari Revenue Intelligence | Not ready | 29 | Partial | Close the flagged data gaps first |
Gainsight Customer Success / Health Scoring | Not ready | 29 | Partial | Close the flagged data gaps first |
MadKudu Predictive Lead Scoring | Not ready | 29 | Partial | Close the flagged data gaps first |
Sift Digital Trust and Fraud Prevention | Not ready | 29 | Partial | Close the flagged data gaps first |
Gong Conversation Intelligence / Revenue Intelligence | Not ready | 17 | External data | Connect Core input is recorded and transcribed sales calls, meetings and emails, captured through telephony/video/email integrations, not from any CRM/ERP/HRIS export DataTruth reads |
6sense ABM / Predictive Intent | Not ready | 17 | External data | Connect Core signal is third-party intent and anonymous web behaviour from the 6sense Signalverse, plus a website tracking tag, none of which appears in a CRM/ERP export |
In a live assessment these verdicts are measured from the company's own systems, across 30+ tools including the ones on your list.
What it costs to wait
Conservative estimate, built line by line from observed dataCapital at risk, this year (estimate)
€2.9-5.4M
The assessment is €20,000. The single largest line alone is worth €2,598,519. This is insurance, not a report.
How we get to this number
Revenue leakage from unreconciled systems
€2,598,519
Observed: €11,811,448 of revenue across 16,683 accounts has no match between Microsoft Dynamics and Oracle
Rate: x 22% leakage (collection slippage, attribution loss, double-billing), the low end of the 20-35% range
AI initiative at risk of stalling
€1,008,900
Observed: Truth Score 41 (Band C), 59 points below AI-ready
Rate: x 1.8% of revenue, scaled by the readiness gap (most stalled GenAI fails first on data quality)
Wasted spend on duplicate / contaminated records
€249,905
Observed: ~11% of 73,800 contacts are duplicates or internal records
Rate: x €7.40 cost-per-contact x 4 touch cycles a year
Manual reconciliation & rework
€191,880
Observed: 52-point accuracy gap across 410,000 records
Rate: x ~€0.90 per record of analyst time stitching systems by hand
Governance & compliance exposure
€30,000
Observed: 50-point governance gap: ownership, lineage and audit trail
Rate: data-governance & audit-trail readiness for AI regulation, capped conservatively at €60K
Every rate is deliberately conservative and yours to change. Dial them to your own numbers and the total moves with you. The point is the observed finding, not the decimal.
The findings above are observed facts in the data. The euro figures are modelled estimates from those facts. On a real engagement every number is computed from your actual records, with the working shown. Demo figures are illustrative.
The evidence behind the verdict
Six-pillar breakdown
Click a pillar to open the engineHow it compares
Positioned on the Financial Services reference model - a calibrated cross-sector baseline. Live peer benchmarking activates as real assessments accumulate.
11th percentile
-17 vs reference median
Pillars vs sector median
Continuous monitoring
Every snapshot retained since onboarding · trend, not just point-in-time
▲ 17 pts since Jan - ~€153K annual data-risk reduced
24 → 41
What moved over the period
Value attributed: a 17-point improvement reduced estimated annual data-risk exposure by ~€153K. This is what continuous monitoring proves - and what justifies the recurring engagement.
The roadmap: what to fix first
De-duplicate the client master - 18% duplicate rate on key fields makes any risk model unreliable.
Board-ready deliverable
Download a sample IC board pack built from this assessment.
Buying this company? See the M&A diligence verdict
Proceed, fix, or walk, with the time-to-AI-ready and remediation cost, the buyer's view of Harbour Financial.