Assessment · 6 November 2026
Can the data be trusted? · Saoirse Manufacturing
Solid data infrastructure with minor issues to resolve. Well-positioned for most AI use cases (Band B).
What can the data actually run?
The finding that lands
Counts computed from the data2,799 customers exist as two different companies across Microsoft Dynamics and SAP.
€4,294,588 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.
Worth fixing before you scale AI: Any AI keyed on customer identity (pricing, churn, lifetime value, next-best-action) trains on a contradiction and ships confident, wrong answers.
7,237 of 124,200 customers in Microsoft Dynamics have no matching record in SAP.
€4,294,588 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.
1,987 internal employees from BambooHR are sitting in the Microsoft Dynamics contact list.
1,987 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.
6 of 6 relevant use cases ready · tuned to Manufacturing
The Opportunity · AI Value Creation
What this data could actually run to create value
Specific value-creation plays for this company's data, including moves beyond the standard use cases. Grounded and put through an adversarial check: every fact is cited and verifiable; the play is a hypothesis, not a promise; no figure is invented. Open any play for a full, downloadable proposal.
Reconcile the 2,799 duplicate customer records across Dynamics and SAP
An entity-resolution agent that fuses the 2,799 split customer records into a single golden customer master will surface duplicate-account credit exposure, fragmented spend, and pricing inconsistencies that are currently invisible because the same customer is quoted and invoiced as two companies.
Based on - verifiable in the assessment
- Microsoft Dynamics and SAP match at only 70% · crossSystem[0].matchRate
- 2,799 customers exist as two different companies across Microsoft Dynamics and SAP · crossSystem[0].context
- Accuracy score is 82 · strengths
- Consistency score is 79 · strengths
Gap: 70% match rate means ~30% of the Dynamics↔SAP customer pairs are unlinked; the golden-record match must be built and remaining fuzzy matches human-reviewed before consolidated spend/credit can be trusted.
Demand-sensing replenishment agent off the SAP transaction spine
Because the SAP order/production data is complete and accurate enough to be deployable now, a demand-sensing model trained on order arrival patterns per customer can forecast reorder timing and drive proactive replenishment outreach before customers issue POs.
Based on - verifiable in the assessment
- SAP is a live system of record · systems
- Completeness score is 80 · strengths
- Accuracy score is 82 · strengths
- 6 of 6 assets are deployable now with zero blocked · leverability.deployableNow / leverability.blocked
Gap: None blocking a pilot; forecast accuracy must be validated on held-out history before automating outreach.
Supplier-term leverage engine from consolidated demand data
Aggregating true consolidated purchase volume per input material (once demand is de-duplicated across systems) creates a fact base to renegotiate supplier tiers and consolidate spend that is currently fragmented across split records.
Based on - verifiable in the assessment
- SAP holds procurement/transaction data as a system of record · systems
- Completeness score is 80 · strengths
- Governance score is 84 (highest pillar) · strengths
- Dynamics↔SAP customers are split across two systems at 70% match · crossSystem[0]
Gap: Requires the customer/entity de-duplication (70% match) resolved first so demand-to-supply volumes are not understated by fragmented records; supplier-side spend must be confirmed as within SAP scope.
Cross-system customer LTV and cross-sell model on the golden master
Once customers are unified, a lifetime-value and next-product model built on combined Dynamics (relationship/pipeline) and SAP (actual order/production) signals can identify which customers buy only a subset of the product range and target them for cross-sell.
Based on - verifiable in the assessment
- 2,799 customers appear as two companies across Dynamics and SAP · crossSystem[0].context
- Dynamics↔SAP match rate is 70% · crossSystem[0].matchRate
- Accuracy 82 and Completeness 80 · strengths
Gap: Depends on the 70%-match customer consolidation being completed; model cannot join order history to relationship data until records are unified.
Exploratory · needs validation before it's shown to a client
Employee-to-customer relationship graph for account coverage risk
Linking BambooHR employees to the Dynamics accounts they own creates a coverage-and-continuity map that flags key accounts concentrated on single reps and reps who are flight-risk, letting the operator pre-empt revenue loss from departures.
Based on - verifiable in the assessment
- BambooHR (HRIS) links to Microsoft Dynamics (CRM) at 77% match · crossSystem[1].matchRate
- The link represents employee vs customer separation (HRIS to CRM) · crossSystem[1].context
- Governance score is 84 · strengths
Gap: 77% HRIS↔CRM match; the remaining ~23% of employee-to-account links must be resolved, and ownership fields in Dynamics confirmed populated, before coverage risk is reliable.
Benchmarked lead-time / reliability data product for key accounts
The strong-governance SAP order-to-delivery record can be productised into a customer-facing reliability/lead-time dashboard offered to key accounts, turning internal operational data into a stickiness and premium-justification asset.
Based on - verifiable in the assessment
- SAP is the transactional/production system of record · systems
- Governance score is 84 (strong) · strengths
- Consistency score is 79 · strengths
- 6/6 assets deployable, 0 blocked · leverability
Gap: Requires OTIF/lead-time fields validated for external reporting and a governance sign-off on what customer-facing metrics can be exposed; consistency at 79 must be lifted for shared-facing numbers.
Are you ready for the tools you're evaluating?
6 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 | Ready | 82 | Fully assessable | Ready on current data |
Celonis Process Mining and Execution Management | Ready | 82 | Fully assessable | Ready on current data |
Clari Revenue Intelligence | Ready | 66 | Partial | Ready on current data |
Gainsight Customer Success / Health Scoring | Ready | 66 | Partial | Ready on current data |
MadKudu Predictive Lead Scoring | Ready | 66 | Partial | Ready on current data |
Sift Digital Trust and Fraud Prevention | Ready | 66 | Partial | Ready on current data |
Gong Conversation Intelligence / Revenue Intelligence | Conditional | 54 | 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 | Conditional | 54 | 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)
€1.3-2.3M
The assessment is €20,000. The single largest line alone is worth €944,809. This is insurance, not a report.
How we get to this number
Revenue leakage from unreconciled systems
€944,809
Observed: €4,294,588 of revenue across 7,237 accounts has no match between Microsoft Dynamics and SAP
Rate: x 22% leakage (collection slippage, attribution loss, double-billing), the low end of the 20-35% range
AI initiative at risk of stalling
€530,640
Observed: Truth Score 78 (Band B), 22 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
€145,582
Observed: ~4% of 124,200 contacts are duplicates or internal records
Rate: x €7.40 cost-per-contact x 4 touch cycles a year
Manual reconciliation & rework
€111,780
Observed: 18-point accuracy gap across 690,000 records
Rate: x ~€0.90 per record of analyst time stitching systems by hand
Governance & compliance exposure
€9,600
Observed: 16-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 Manufacturing reference model - a calibrated cross-sector baseline. Live peer benchmarking activates as real assessments accumulate.
Top 3%
+24 vs reference median
Pillars vs sector median
Continuous monitoring
Every snapshot retained since onboarding · trend, not just point-in-time
▼ 11 pts - €99K of annual data-risk re-opened
89 → 78
What moved over the period
Truth Score fell 11 points over the last two quarters - monitoring flagged the regression for review.
The roadmap: what to fix first
Establish a shared customer key between SAP and Dynamics - 9% of accounts do not reconcile across the two.
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 Saoirse Manufacturing.