01The engagement in brief
Futura Tec GmbH · data-center, IoT and solar technology · Frankfurt am Main
Three businesses on one view: export risk screened before every quote, vendor rebates protected, margin known per deal
Futura-Tec aimed to see how its three businesses, data-center projects, NB-IoT design and solar energy, really ran from opportunity to cash, and to be sure no export order or vendor deal left the company without the right checks.
Client issue: Futura-Tec, a Frankfurt technology company of about 50 people, builds data centers as a reseller of HPE, Dell EMC, Fujitsu, Cisco, Corning, Red Hat, Oracle, Panduit and Nexans and as a partner of Quanta Cloud Technology in Turkey, Armenia and Oman; it also designs NB-IoT chips and modules and runs solar projects. Each line quoted, bought and invoiced its own way; export screening was done case by case and often after the quote; 45% of vendor deals were registered too late for rebates; and deal margin was known only when a project closed.
Solution: In eight weeks AIx2 mapped the workflow and the data flow from opportunity to cash for all three lines in parallel across CRM, ERP, vendor partner portals, the screening log, project files and e-mail metadata, translated them into one canonical text stream on-premise, and fused deal, destination, vendor, bill of materials and invoice into an organization model. The model was checked against Futura-Tec’s own procedures, the vendors’ partner-program rules, ISO 9001, and EU export-control law: the dual-use regulation (EU) 2021/821 and its export-authorisation rules for its export markets.
Impact: The due diligence delivered 24 findings with evidence and owners and a control specification for each. Futura-Tec’s teams built the controls into their own CRM: every export order is screened before the quote, 95% of vendor deals are registered before quotation, and margin is visible from the first quote. Recurring value is about €0.36M a year, 4.2 times the fee, and no order has shipped without a recorded screening since March 2026.
02Results, measured by the client
Seven months after the due diligence, screening, registration and margin control are routine
Before: the 12 months to November 2025. After: March – September 2026, once Futura-Tec’s own controls were live (n ≈ 140 quotes, ≈ 60 orders, ≈ 35 export orders).
All figures from Futura-Tec’s CRM, ERP and screening log, compiled by its finance lead.
03How the results were verified
Counts from Futura-Tec’s own CRM, ERP, vendor portals and screening records
Same three business lines and team size over the period
The mechanism is traceable: every quote now carries a screening ID, a deal-registration ID and a margin field
AIx2 has not been involved since February 2026
04What Futura-Tec built itself
A screening step and a deal-registration step before quotation in the CRM, a margin field and floor on every quote, invoice release on delivery confirmation, and a monthly R&D portfolio review.
05Financial case, one-time due diligence
A due diligence returning about €0.36M a year, 4.2 times the fee
One time, eight weeks, mid-December 2025 – mid-February 2026; no further fee. Values annualised from March – September 2026, using Futura-Tec’s own margins and financing cost.
Vendor rebates no longer lost to late registration — Annual value: ≈ €180k; Basis: registration 55% → 95% on rebate-eligible data-center deals
Margin leakage stopped on quotes — Annual value: ≈ €120k; Basis: quotes below target margin 18% → 4%, caught before sending
Faster invoicing — Annual value: ≈ €60k; Basis: financing cost of 14 fewer days on data-center receivables
Total recurring value — Annual value: ≈ €360k a year; Basis: 4.2 times the fee each year
Payback: the fee was recovered in about three months of the controls running (June 2026). Not priced in: the cost of an export-control breach, which can mean losing export authorisations and vendor partner status.
Conclusion: Futura-Tec paid once. The controls it built from the findings return about €360k a year, so the fee was back by June 2026 and returns 4.2 times every year; over three years about €1.0M. The larger value, avoiding an export-control breach in markets such as Armenia and Oman, is not priced.
06How the work ran: three phases
Eight weeks inside Futura-Tec’s network, each phase signed off by the managing director
Weeks 0–3 · Connect and map: Read-only links to CRM, ERP, vendor partner portals, the screening log and project files; three order-to-cash flows mapped in parallel with their data.
Weeks 3–6 · Check: Against Futura-Tec’s procedures, vendor partner-program rules, ISO 9001, EU export-control rules (Regulation (EU) 2021/821). Each finding with evidence, a count and an owner.
Weeks 6–8 · Hand over: 24 findings, a control specification per finding and a 90-day plan for Futura-Tec’s own teams.
Human sign-off at every step: every finding and the final report were approved by the managing director before release.
07Data stayed on site
Ran on Futura-Tec’s own servers: no cloud
Read-only connectors; personal data masked
Customer and end-user names pseudonymised in all outputs
08How AIx2 accessed it
Jump host controlled by Futura-Tec, sessions recorded
No inbound connections
All AIx2 access removed at the end of week 8
09Controls on Atlas
Role-based access, encryption, tamper-evident audit log
Software bill of materials
Deletion of all client data at exit, confirmed in writing
10What the work required
Five capabilities the due diligence depended on, and how common tool categories cover them
Process-analytics tools and data platforms are strong in their own scope, and Atlas reads from and feeds both. The right-hand column shows what each capability produced at Futura-Tec.
Compare three order-to-cash flows from event data — Process-analytics tool: Core strength; Data platform: After pipelines are built; AIx2 Atlas: Yes; Evidence at Futura-Tec: 3 ways of quoting, buying and invoicing found
Join deals, destinations, screening and vendor registration — Process-analytics tool: Process only; Data platform: Data only; AIx2 Atlas: Yes, the FUSE stage; Evidence at Futura-Tec: screening after quote on 60% of non-EU orders
Check against export-control and partner-program rules — Process-analytics tool: No; Data platform: After custom build; AIx2 Atlas: Yes, clause extractor + rule compiler; Evidence at Futura-Tec: EU 2021/821 and rebate-rule gaps with counts
Read vendor price lists and BOMs as text — Process-analytics tool: No; Data platform: After custom build; AIx2 Atlas: Yes, tabular serializer + OCR; Evidence at Futura-Tec: margin computed per quote line
Run entirely on-premise on CPU — Process-analytics tool: Cloud first; Data platform: Mostly cloud; AIx2 Atlas: Yes; Evidence at Futura-Tec: no customer or end-user data left Futura-Tec
11What this means for other clients
For resellers and integrators, the value sits where risk and margin are decided: before the quote
What transfers from Futura-Tec to similar companies.
Export orders screened before the quote, 40% → 100% → EU export-control checks for any exporter of IT, OT or electronics
Vendor deals registered before quotation, 55% → 95% → partner-program rebates at any reseller
Margin known at quotation, 30% → 90% → quote discipline in project businesses
Delivery to invoice, 21 → 7 days → cash conversion in project delivery
Why one due diligence was enough: Futura-Tec had several years of CRM and ERP data, an engineering team, and a sponsor who decided fast, so it could build the controls itself.
What a less mature client would need: Without system data and owners per business line, the same findings need a managed service to stay in place.
12Timeline
Eight weeks of due diligence, then seven months run by Futura-Tec’s own teams
Connect · Dec; Map & check · Jan; Hand over · Feb; Futura-Tec runs the controls · Mar – Oct 2026
13Where it stands, October 2026
Screening and deal registration are CRM steps before every quote
Margin floor on every quote
Invoice released on delivery confirmation
Monthly R&D portfolio review for the NB-IoT programme
14Appendix: the engagement report
APPENDIX · ENGAGEMENT REPORT
AIx2 Atlas at Futura-Tec
The detail behind the deck: scope, how the due diligence ran, findings, measured results and the financial calculation. Company facts from futura-tec.com; size and engagement terms from Futura-Tec.
15Executive summary
Engagement: mid-December 2025 – mid-February 2026 · one 8-week due diligence · since then run by Futura-Tec
Futura Tec GmbH is a Frankfurt technology company of about 50 people with three business lines that grew side by side: data-center solutions, where it designs and builds data centers as a reseller of HPE, Dell EMC, Fujitsu, Cisco, Corning, Red Hat, Oracle, Panduit and Nexans and as Quanta Cloud Technology’s partner in Turkey, Armenia and Oman; NB-IoT, where it designs and produces chips and modules; and solar energy projects.
Each line had built its own way of quoting, buying and invoicing. Export screening for non-EU orders was done by finance when an order arrived, usually after the quote had gone out; vendor deal registrations, which decide rebates, were left to individual sales staff; and project margin was known only when a project closed, because bill-of-materials costs lived in spreadsheets outside the CRM.
In eight weeks AIx2 mapped the workflow and the data flow from opportunity to cash for all three lines in parallel, fused deal, destination, vendor, bill of materials and invoice into one organization model, and checked it against Futura-Tec’s own procedures (Δ1), ISO 9001 and the vendors’ partner-program rules (Δ2), and EU export-control law, Regulation (EU) 2021/821 (Δ3). It delivered 24 findings, each with evidence, a count and an owner, a control specification per finding and a 90-day plan.
Measured March – September 2026: export orders screened before the quote 40% → 100%; vendor deals registered before quotation 55% → 95%; quotes with margin known at quotation 30% → 90%; quotes below target margin 18% → 4%; delivery to invoice 21 → 7 days. Annual value: about €360k, 4.2 times the one-time fee; payback in June 2026.
16Client profile
Facts from futura-tec.com (imprint, Datacenter, Solar and IoT pages, visited October 2026); size and engagement terms from Futura-Tec
Legal entity: Futura Tec GmbH, Frankfurt am Main, Germany
Positioning: “a high-tech research and development company specializing in design, development and implementation of advanced technology solutions”
Data center: designs and builds data centers; reseller of HPE, Dell EMC, Fujitsu, Cisco, Corning, Red Hat, Oracle, Panduit and Nexans; partner of Quanta Cloud Technology (QCT) in Turkey, Armenia and Oman
IoT: design and production of NB-IoT chips and modules; IoT solutions
Solar energy: solar energy projects
Markets: Germany and export markets including Turkey, Armenia and Oman
Size: ≈ 50 employees
17Operating model and processes mapped
Three business lines, one value chain
Lead & opportunity → Vendor deal registration → Configuration & BOM → Export screening → Quote → Order & purchase → Delivery & installation → Acceptance → Invoice & collection → Rebate claim
Opportunity to quote — Sub-processes: qualification, vendor deal registration, solution design, BOM, pricing, export classification and screening, approval; Systems and traces: CRM, vendor partner portals, price lists, e-mail; Line: all
Order to delivery — Sub-processes: order confirmation, purchase from vendor or distributor, logistics and customs, installation; Systems and traces: ERP, project files, forwarder documents; Line: data center, solar
Delivery to cash — Sub-processes: acceptance, invoice, collection, rebate claim; Systems and traces: ERP, finance, vendor portals; Line: all
NB-IoT development — Sub-processes: specification, chip and module design, prototypes, certification, small-series production; Systems and traces: repositories, project files; Line: IoT
Solar projects — Sub-processes: site assessment, design, procurement, installation, handover; Systems and traces: project files, ERP; Line: solar
Compliance — Sub-processes: export classification, end-use checks, export authorisations, record keeping; Systems and traces: screening log, files; Line: all
18Pain points diagnosed
Ranked by impact; confirmed with evidence during weeks 2–6
1. Export screening after the quote (Critical) — Evidence found: 60% of non-EU data-center and IoT orders screened after quotation; 7 after the order; Root cause: screening a finance task triggered by the order, not a sales step; Business impact: exposure under EU export-control rules (Regulation (EU) 2021/821); risk to export authorisations and vendor status
2. Rebates lost to late deal registration (Critical) — Evidence found: 45% of rebate-eligible deals registered after the quote; Root cause: registration left to individual sales staff, never checked; Business impact: about €180k a year of rebates lost
3. Margin unknown until close (High) — Evidence found: 70% of quotes sent without a margin figure; Root cause: BOM costs in spreadsheets outside the CRM; Business impact: 18% of quotes below target margin
4. Slow invoicing (High) — Evidence found: 21 days median from delivery to invoice; Root cause: acceptance paperwork collected by hand; Business impact: cash tied up in receivables
5. Three ways of working (High) — Evidence found: each line quoted, bought and invoiced differently; Root cause: growth line by line, no shared process; Business impact: training, handovers and reporting slowed
6. Stale distributor costs (Medium) — Evidence found: price lists older than 60 days used on 1 in 5 quotes; Root cause: no owner for price-list data; Business impact: margin leakage
7. R&D without a portfolio view (Medium) — Evidence found: NB-IoT milestones tracked per project only; Root cause: no portfolio review; Business impact: engineering time pulled to urgent bids
8. Rebate outcomes not fed back (Medium) — Evidence found: sales never saw which rebates were lost; Root cause: no loop from vendor portal to CRM; Business impact: same mistakes repeated
19Scope and data access
All connections read-only, each with a data contract and a named owner
CRM — Data used: opportunities, quotes, stages, approvals; Method: API; Privacy: customer names pseudonymised; Owner: Sales lead
ERP — Data used: orders, purchases, deliveries, invoices, receivables; Method: connector; Privacy: aggregated finance; Owner: Finance lead
Vendor partner portals — Data used: deal registrations, rebate status; Method: export; Privacy: —; Owner: Sales lead
Screening log — Data used: export classification and export-control check records; Method: export; Privacy: end users pseudonymised; Owner: Finance lead
Price lists and BOMs — Data used: vendor and distributor costs (PDF, spreadsheets); Method: OCR + tabular serializer; Privacy: —; Owner: Sales lead
Project files — Data used: data-center, solar and R&D projects; Method: document modules; Privacy: access tags; Owner: Project leads
E-mail metadata — Data used: who, when, which deal; no message bodies; Method: Graph API; Privacy: metadata only; Owner: Managing director
Procedures and policies — Data used: sales, purchasing, export control, quality; Method: document modules; Privacy: —; Owner: Managing director
Canonical text records produced from a quote, the screening log and a vendor portal:
[2025-10-07 14:12] DEAL DC-25-118 · ACT quote-sent · DEST AM · VALUE 412k EUR · MARGIN n/a · SRC crm [2025-10-14 09:30] DEAL DC-25-118 · ACT screening · RESULT clear · SRC screening-log · GAP after quote: 7 days [2025-10-20 16:05] DEAL DC-25-118 · ACT deal-registration · VENDOR HPE · STATUS rejected (late) · SRC partner-portal
20How the work ran
The six steps, as executed
1 · Data hub — What we did: registered 8 sources with data contracts; read-only connections; price lists and BOMs read as text; Toolboxes and pipeline: connectors, OCR, tabular serializer, PII gateway; Output: one evidence hub per deal
2 · AI mapping — What we did: event logs per deal across CRM, portals, screening log and ERP; three order-to-cash variants discovered; Toolboxes and pipeline: event templater, case correlator, process mining; Output: as-built process map + data supply chain map
3 · Δ1 own procedures — What we did: compared with sales, purchasing and export-control procedures; Toolboxes and pipeline: rule compiler; Output: deviations with exact counts
4 · Δ2 best practice — What we did: ISO 9001 §8.2 and §8.4; vendor partner-program rules; Toolboxes and pipeline: control library with clause mapping; Output: distance to practice per level
5 · Δ3 regulation — What we did: EU 2021/821 dual-use regulation; EU export-control rules (Regulation (EU) 2021/821); Toolboxes and pipeline: clause & obligation extractor; Output: compliance matrix
6 · Hand over — What we did: control specifications and 90-day plan for Futura-Tec’s own teams; Toolboxes and pipeline: —; Output: report approved by the managing director
21Process and data findings by level
4 workflow levels · 5 data-governance levels
01 · Strategy — What we mapped: growth across three lines and export markets; Finding: no shared view of margin or risk across the lines
02 · Strategy execution — What we mapped: engineering time between data-center bids and NB-IoT R&D; Finding: time allocated by urgency, not by value
03 · Detailed processes — What we mapped: opportunity → quote → order → delivery → cash; Finding: 3 variants; screening and registration after the quote
04 · Outcomes — What we mapped: projects, invoices, rebates; Finding: margin and rebate outcome known only at close
L0 · Governance foundation — What we mapped: ownership of price lists, screening records, BOMs; Finding: no owner for price-list data
L1 · Data inputs — What we mapped: CRM, portals, spreadsheets; Finding: BOM costs outside the CRM on 70% of quotes
L2 · Data processing — What we mapped: margin and rebate calculations; Finding: done by hand per quarter
L3 · Data outputs — What we mapped: margin and sales reports; Finding: rebuilt by hand each quarter
L4 · Operations-generated data — What we mapped: rebate outcomes, screening results; Finding: never fed back to sales
22What the organization model answered
Where is export risk decided? — Signals joined: quote date, destination, screening date; Answer in the due diligence: at quotation, but screening ran on average 7 days later
Why are rebates lost? — Signals joined: quote date, registration date, portal status; Answer in the due diligence: registration after the quote on 45% of eligible deals
Which quotes leak margin? — Signals joined: quote lines, price-list age, distributor cost; Answer in the due diligence: 18% below target, mostly on price lists older than 60 days
Where does cash wait? — Signals joined: delivery, acceptance, invoice dates; Answer in the due diligence: between delivery and acceptance paperwork
Which line uses engineering time? — Signals joined: project files, timesheets; Answer in the due diligence: urgent data-center bids pulled time from NB-IoT milestones
23Check findings (summary)
Each finding came with exact counts, witness evidence and a named owner
Δ1 — Reference: own sales procedure; Finding: no screening or registration step before the quote; Severity: High; Owner: Sales lead
Δ1 — Reference: own purchasing procedure; Finding: price lists without an owner or validity date; Severity: Medium; Owner: Sales lead
Δ2 — Reference: ISO 9001 §8.2; Finding: customer requirements not reviewed before quotation on 1 in 3 quotes; Severity: Medium; Owner: Managing director
Δ2 — Reference: ISO 9001 §8.4; Finding: distributor costs not verified at quotation; Severity: Medium; Owner: Purchasing
Δ2 — Reference: vendor partner-program rules; Finding: 45% of eligible deals registered after the quote; Severity: High; Owner: Sales lead
Δ3 — Reference: EU 2021/821 (dual-use); Finding: classification and end-use checks not recorded before the quote; Severity: Critical; Owner: Finance lead
Δ3 — Reference: EU export authorisation rules (Regulation (EU) 2021/821, Art. 3 and 4); Finding: export check after the quote on 60% of non-EU orders; 7 after the order; Severity: Critical; Owner: Finance lead
24Deliverables
Handed over in February 2026
Process and data maps — Content: three order-to-cash flows + data supply chain with owners; Format: interactive map
Findings register — Content: 24 findings with evidence, count, owner and due date; Format: register
Control specifications — Content: screening step, deal-registration step, margin field and floor, invoice release on delivery; Format: specification per control
Price-list governance — Content: owner, validity date, update rule; Format: policy
R&D portfolio review — Content: milestones, resources, decisions for NB-IoT; Format: template
90-day plan — Content: order of work with owners; Format: plan
25Measured results
Measured by Futura-Tec: the 12 months to November 2025 vs March – September 2026
Export orders screened before the quote — Definition: screening record dated before the quote ÷ non-EU orders; Before: 40%; After: 100%; Sample / window: ≈ 35 export orders; Source: screening log, CRM
Vendor deals registered before quotation — Definition: registration before the quote ÷ rebate-eligible deals; Before: 55%; After: 95%; Sample / window: ≈ 60 deals; Source: vendor portals, CRM
Quotes with margin known at quotation — Definition: quotes with a margin figure ÷ all quotes; Before: 30%; After: 90%; Sample / window: ≈ 140 quotes; Source: CRM
Quotes below target margin sent — Definition: share of quotes; Before: 18%; After: 4%; Sample / window: same quotes; Source: CRM
Delivery to invoice — Definition: median days, data-center projects; Before: 21; After: 7; Sample / window: ≈ 20 projects; Source: ERP
Price lists older than 60 days in use — Definition: share of quotes; Before: 20%; After: 3%; Sample / window: same quotes; Source: CRM, price lists
26Financial calculation
Rebates protected — Calculation: registration 55% → 95% on rebate-eligible data-center deals, at Futura-Tec’s average rebate; Annual value: ≈ €180,000
Margin leakage stopped — Calculation: quotes below target 18% → 4%, at the average shortfall per quote; Annual value: ≈ €120,000
Faster invoicing — Calculation: 14 fewer days on data-center receivables, at Futura-Tec’s financing cost; Annual value: ≈ €60,000
Recurring value — Annual value: ≈ €360,000 a year
Payback: about three months of the controls running (June 2026)
Three years: ≈ €1.0M
Return: 4.2 times the fee each year
Conclusion. Futura-Tec paid once. The controls its own teams built from the findings return about €360k a year, so the fee was back by June 2026 and returns 4.2 times every year. Not priced: avoiding a breach of EU export-control rules, which can cost export authorisations and vendor partner status.
27Timeline and governance
Dec 2025 · weeks 0–2 — Phase: Connect; Work done: 8 sources, data contracts, access; Signed off when: every source has an owner
Dec 2025 – Jan 2026 · weeks 2–4 — Phase: Map; Work done: three flows and the data supply chain; Signed off when: maps approved by the managing director
Jan 2026 · weeks 4–6 — Phase: Check; Work done: three checks, findings with counts; Signed off when: findings with owners
Feb 2026 · weeks 6–8 — Phase: Hand over; Work done: report, control specifications, 90-day plan; Signed off when: report approved
Mar – Oct 2026 — Phase: Run by Futura-Tec; Work done: controls built into the CRM and ERP by its own teams; no AIx2 involvement; Signed off when: results measured by finance
Managing director: sponsor, scope, sign-off
Sales lead: quotes, deal registration, price lists
Finance lead: screening log, margin, invoicing
Project leads: data-center, solar and NB-IoT projects
AIx2: connections, mapping, checks, report
28What made it work
Starting from the quote, where risk and margin are decided
Joining the screening log with the CRM showed timing, not just coverage
Reading price lists and BOMs as text made margin computable per quote line
A sponsor who turned findings into CRM steps within weeks
Years of CRM and ERP data to map from
29How each result happened
HOW EACH RESULT HAPPENED
The story behind every number
For each measured result: what Atlas found by joining Futura-Tec’s systems, what it compared the as-built company with, what it recommended, what Futura-Tec’s own teams built, and how the outcome was verified. Atlas found, measured and proposed; Futura-Tec’s people decided and made the changes.
30The loop every finding followed
Understand → identify → trace data and dependencies → compare with the target → recommend → client executes → client verifies.
31What combining the data adds
Each system alone looked fine. Only joined did they show when screening happened relative to the quote, which rebates were lost and why, and where margin leaked.
32How each result happened: Export & compliance
Where export risk is decided, and why it was decided too late.
Screening before the quote · 40% → 100% · Deal × destination × screening log · timing analysis
We joined every CRM deal with its destination country and the screening log. Coverage looked fine: almost every order had been screened. Timing did not: the screening record was usually dated after the quote, and for seven orders after the order itself.
Under EU export-control rules (Regulation (EU) 2021/821) the check has to happen before Futura-Tec commits. We recommended moving it into the sales flow.
Atlas showed: For 60% of non-EU orders the screening record was dated after the quote; average gap 7 days. We found: Screening was a finance task triggered by an incoming order, not a sales step. What changed: Futura-Tec added a mandatory screening step in the CRM before a quote can be sent. Why it moved: A quote cannot leave the CRM without a screening ID; finance reviews exceptions weekly.
End-use and classification records · not recorded → recorded per deal · Clause & obligation extractor · EU 2021/821
The regulation requires knowing what is exported and to whom. Classification and end-use statements existed in e-mails but not as records.
Atlas showed: Classification and end-use evidence was missing from the file for most non-EU deals. We found: Information was gathered informally by sales. What changed: A classification and end-use field set in the CRM, filled before screening. Why it moved: The record now exists before the quote, not after a question from a customs officer.
33How each result happened: Vendors & margin
Where rebates and margin were lost before anyone noticed.
Deal registration before quotation · 55% → 95% · Vendor portal × CRM reconciliation
We reconciled registrations in the vendor partner portals with CRM quotes for every rebate-eligible deal. Late registrations were rejected or lost their rebate, and nobody in sales saw the outcome.
Atlas showed: 45% of rebate-eligible deals were registered after the quote. We found: Registration was left to individual sales staff and never checked. What changed: Registration became a CRM step before quotation, with the portal status shown on the deal. Why it moved: Rebate outcomes are visible per deal and reviewed monthly by the sales lead.
Quotes below target margin · 18% → 4% · Price list × BOM × quote line
Price lists and BOMs, held as PDFs and spreadsheets, were read as text and matched to every quote line.
Atlas showed: 18% of quotes went out below target margin. We found: Most used distributor costs from price lists older than 60 days. What changed: A margin field and floor on every quote, and an owner and validity date for every price list. Why it moved: Below-floor quotes need a named approval; stale price lists are flagged.
Margin known at quotation · 30% → 90% · BOM costs moved into the CRM
Margin was calculated only when a project closed, from spreadsheets outside the CRM.
Atlas showed: 70% of quotes had no margin figure when sent. We found: BOM costs were not part of the quote record. What changed: BOM costs imported into the quote; margin calculated automatically. Why it moved: Sales sees the margin before the customer sees the price.
34How each result happened: Cash & delivery
Where cash waited after the work was done.
Delivery to invoice · 21 → 7 days · Delivery × acceptance × invoice timeline
We rebuilt every data-center project from delivery to invoice. The work was finished long before the invoice went out.
Atlas showed: Invoices waited a median 21 days after delivery. We found: Acceptance paperwork was collected by hand before invoicing. What changed: Invoice released on delivery confirmation; acceptance tracked in parallel. Why it moved: The waiting step was removed from the flow.
Three ways of working · 3 variants → 1 flow · Process mining across the three lines
Each line quoted, bought and invoiced differently, which slowed handovers and reporting.
Atlas showed: Three order-to-cash variants with different steps and owners. We found: Growth line by line had never been followed by a shared process. What changed: One flow with line-specific steps owned by named people. Why it moved: Reports and handovers now use the same steps for all three lines.
35How each result happened: R&D
NB-IoT portfolio view · none → monthly review · Project files × timesheets
NB-IoT milestones were tracked per project; urgent data-center bids kept pulling engineers away.
Atlas showed: Engineering time shifted to bids in every month with a large tender. We found: No portfolio view to weigh R&D against bids. What changed: A monthly portfolio review with milestones, resources and decisions. Why it moved: Trade-offs between bids and R&D are now decided, not drifted into.