Go back to Market Research
Silvon Software develops Stratum, a business intelligence and planning platform for manufacturers and distributors. It consolidates operational data for sales, inventory, supplier, manufacturing, and financial analysis. Users work through browser views, dashboards, Excel and Power BI integrations, collaborative budgets, statistical forecasts, and inventory calculations. Silvon offers both hosted and customer-operated deployments. This review covers the publicly documented Stratum 7.7 environment and the associated planning applications. 1 4 9 13
Overall supply chain score: 4.4/10
| Dimension | Score |
|---|---|
| Supply chain depth | 3.4/10 |
| Decision and optimization substance | 3.2/10 |
| Product and architecture integrity | 5.4/10 |
| Technical transparency | 5.6/10 |
| Vendor seriousness | 4.4/10 |
Silvon has a coherent reporting product, unusually accessible administration manuals, and a maintained release history. The supply chain limitation is its decision model: assemble trusted figures, calculate conventional inventory policies, identify deviations, and ask people to investigate. Better information can eliminate substantial spreadsheet work. It does not establish that the resulting purchases, allocations, or production choices maximize economic returns.
The inventory offering extends beyond dashboards: its brochure describes automatic replenishment quantities. Yet the documented mechanisms remain service targets, safety stocks, and EOQ, with planners handling the broader trade-offs. The stronger customer evidence concerns report preparation and distribution. There is little public evidence of a continuously operated, economically validated decision engine that combines demand uncertainty, supplier uncertainty, and competing uses of capital. 2 6
Silvon versus Lokad
The principal difference is the work each product is intended to absorb. Stratum organizes information for business users and supports their planning process. Lokad’s supply chain optimization approach organizes software around the recurring decisions themselves: what to buy, where to allocate stock, when to produce, and which economic trade-offs to accept.
A shared data model is useful in either approach. It avoids reconciling different meanings of sales, customers, and inventory across departments. But consistency of representation does not resolve a conflict between two commercially attractive actions. Buying more of one product can consume the cash, warehouse space, or supplier capacity needed by another. Ranking those opportunities requires an economic objective and a representation of uncertain future outcomes.
Silvon’s recent AI argument preserves this distinction. Its September 2026 article presents Stratum as the foundation of reliable business definitions for AI applications, with salespeople using forecasts to investigate accounts and adjust expectations. That is a credible use of a data platform; it leaves much of the operating workload with the customer. 17
Lokad’s Supply Chain Scientists build and maintain the executable decision logic as part of its service. Its broader design perspective, explained in Introduction to Supply Chain, places uncertainty, financial consequences, and the automation of mundane decisions at the center. The comparison therefore turns on net economic performance and remaining labor, including maintenance. Producing more accessible reports is an intermediate result.
What the planning mechanisms actually do
Statistical forecasts followed by overrides
The forecasting brochure, whose document metadata dates its creation to 2016, describes automatic method selection, Holt-Winters, seasonal Box-Jenkins, event adjustments, Croston’s method, and discrete models for sparse demand. Forecasts reconcile across hierarchies, and users can override individual values or totals. It also advertises automatic outlier removal using a configurable threshold. 3
These are identifiable forecasting techniques. Their presence is more informative than an unexplained AI label, but it does not demonstrate a distinctive optimization method. Sparse-demand forecasting is useful; selecting an expected demand trajectory still leaves the purchasing decision unresolved. A distribution of demand during uncertain replenishment time is materially different from a forecast with a manually agreed adjustment.
Outlier removal also needs a business interpretation. A recording error, an exceptional promotion, a shortage-constrained sale, and an infrequent large order carry different information. Removing all four because they are unusual can erase precisely the events that determine inventory risk. The public brochure offers statistical cleaning, without explaining how that distinction reaches replenishment decisions. 3
The planning application starts from historical sales, entered targets, or imported forecasts. It spreads summary changes using historical proportions or user percentages, compares working and frozen budgets, and sends reminders when plans need attention. 5 Arithmetic consistency across a hierarchy helps budgeting. It does not establish that the allocated SKU quantities are commercially desirable or physically feasible.
Inventory policies rather than a common economic objective
Silvon’s inventory brochure specifies service-based safety stocks, a non-probabilistic weeks-of-supply alternative, EOQ, and projected balances using inventory, forecasts, customer orders, and production and supply plans. It describes what-if analysis for combining products on orders, carrying-cost analysis, and planned-versus-actual lead times. 2
The distinction matters. A service target constrains an inventory policy before the software has established whether the marginal service improvement is worth its cost. EOQ introduces purchase and holding economics, but its classical simplifications do not capture the full effects of shared capacity, price breaks, expiry, substitution, and correlated supply failures. A planner’s what-if comparison can explore some of those interactions; the documents do not specify a joint decision procedure for them.
Silvon does recognize supply uncertainty. Its inventory-planning essay identifies variable lead times alongside forecast errors and criticizes production runs that accumulate stock to absorb overhead. 7 Its supplier analytics also expose lead-time variability. 23 The unresolved step is converting these observations into a probabilistic projection of future supply and combining it with demand and economic losses. Measuring that a supplier has been unreliable is the beginning of the calculation.
What the customer evidence establishes
Wonder Meats provides a concrete operating example. Silvon loads its Dynamics NAV data three times daily and maintains the application. The main customer user prepares and broadcasts more than 300 recurring reports, including budgets, executive summaries, and broker or rebate statements. Imported sales projections are compared with actuals. At the time described, the deployment primarily analyzed sales revenue; costing and inventory analysis remained future extensions. 6
This is a plausible reduction in spreadsheet consolidation and distribution work. It also identifies the remaining human task: deciding what the reports imply and acting on it. The case supplies neither a before-and-after total labor account nor a measured replenishment or margin outcome. The number of reports is an output count, not an economic return.
The distinction is particularly important for food manufacturing. A reporting deployment at a meat producer does not demonstrate that an engine handles shelf life, uncertain yields, or the economics of perishable production. Those require decision-specific evidence.
Technology history and roadmap integrity
Silvon’s record is more consistent than the succession of unrelated technology slogans common in this market. Its historical forecasting and planning brochures describe recognizable statistical methods and collaborative budgeting. Its June 2022 discussion of spreadsheets advocates a governed data hub with controlled updates and shared reporting. The commercial center has remained business information and analysis. 3 5 18
The release record documents concrete delivery. Version 7.4 added hosted Microsoft identity integration in 2022. Version 7.5 and subsequent updates added a glossary and support for Windows Server and SQL Server 2022. Version 7.6 introduced further import sources, templates, and the Cloud Import API. Version 7.7 extends administration and information delivery; the broader redesign of administration belongs to the proposed version 8.0. 9
The AI turn also has a defined scope. The June 2026 training schedule describes AI connectors as complements to existing workflows. The September article argues for reliable data and business semantics beneath AI-assisted forecasting. The documented contribution remains data preparation and interpretation, with people inspecting the information and deciding how to act. 16 17
The forecasting alliance has continuity. Silvon lists Business Forecast Systems, the maker of Forecast Pro, as a solution partner. Forecast Pro still markets supported forecasting products and subscriptions in 2026. The partner listing does not identify the engine version deployed inside each Stratum installation. 19 20
The main roadmap concern is consequently strategic: improving a mature reporting workflow may preserve its usefulness while leaving the much larger opportunity—automating consequential supply chain decisions—outside the product’s core.
Bill of software materials and system boundaries
The February 2026 Connector manual specifies Stratum.Server on SQL Server or DB2 for i, Connector building an Analysis Services cube, and Viewer using a SQL Server database. Administrators configure dimensions, measures, partitions, and calculations before processing. Nightly batch refreshes are typical, with other processing modes available. 8
| Component or boundary | Documented role | Ownership implication |
|---|---|---|
| Stratum.Server | Source repository on SQL Server or DB2 for i | The chosen deployment determines database administration; these are alternatives. |
| Connector and Analysis Services | Build and refresh the analytical cube | Data changes must propagate through the processing chain. |
| Viewer | Browser analysis, configuration, and planning | Business definitions and permissions are application configuration. |
| Excel and Power BI | Additional report interfaces | Report owners must preserve consistent calculations and access rights. |
| Cloud Import API | Imports from SQL Server and Azure sources | Pipeline operation and mappings remain part of the implementation. |
This is a concentrated Microsoft-oriented architecture with an IBM database deployment option. Its principal maintenance burden is the layered data model and the customer-specific reporting estate.
Silvon’s undated Technical Consultant posting corroborates the work involved: configuring installations, building import interfaces, creating reports, and changing client data models. It requests SQL, SSIS, SSRS, and MDX experience. Its old Windows and SQL version references reflect the age of the posting; the newer release documentation explicitly supports 2022 versions. 12 9
Data export and familiar interfaces ease access to results. They do not make a Stratum implementation portable as executable business logic. Replacing it would also require recreating its dimensions, calculations, planning behavior, permissions, and distribution rules.
Configuration maintenance, security, and total ownership cost
Hosting transfers work; it does not eliminate it
Silvon’s hosted service includes server maintenance and software updates. Its implementation team consists of Silvon employees, and onboarding assigns project and customer-success responsibilities. 13 14 15 This gives customers a clearer service boundary than an open-ended chain of unrelated integrators.
The business model still needs upkeep. Version 7.7’s calendar tools generate Excel mapping files that administrators complete and import. Administrators can inspect usage and permissions through downloadable reports; the migration of all administration into Viewer remains future work. 9 These are useful improvements to recurring tasks, with people still responsible for correct mappings, definitions, and cleanup.
The relevant cost account includes subscription and hosting charges, data integration, model changes, report maintenance, training, and the labor needed to resolve the exceptions produced by the application. Vendor-hosted work belongs in that account even when it disappears from the customer’s IT timesheets. The public cases do not provide this combined cost.
A particularly revealing customer testimonial comes from Hoffmaster: introducing an item absent from the budget history required preparation and assistance to populate the relevant views. 21 New products are ordinary business changes. Their administrative cost matters when assessing how well a model survives daily commercial activity.
Access controls are documented; broader containment is less visible
Stratum’s role manual specifies restrictions by measure and dimension member, with separate read and update permissions. Only security administrators manage roles. The Connector setup also gates whether a measure group can support planning updates. 10 8 These are actual controls over which users can see or alter business information.
The Cloud Import API is restricted to administrators designated as Data Stewards and requires the relevant licenses. Its guide documents SQL and Azure sources, automatic or template-based mapping, and invocation from external services. 11 This provides a programmatic route for data preparation.
The reviewed material gives less detail on hosted tenant isolation, service-account privilege, secret handling, and containment of compromised imports or connectors. Those boundaries matter because an ingestion component can reach both source business systems and the analytical repository. An access-control matrix inside Viewer addresses only part of that exposure.
Planners operate through configured data operations. Security depends on how those interfaces enforce permissions and contain mistakes or compromised credentials.
Productivity and economic accountability
Automating report assembly is useful. The harder question is how much recurring decision labor remains after information reaches its recipient. Current training covers view design, drilling into data, budget updates, imports, and exception analysis. 16 These are productive skills for a reporting platform, but they preserve a substantial workload of inspection and intervention.
A conversational interface over the same reports would change access, without resolving the economic trade-off that produced an exception. Post-2022 agent technology raises the benchmark: the target should be fewer hours spent producing and supervising useful decisions, including the hours spent maintaining the automation.
Silvon can be evaluated against that benchmark without demanding that every reporting product become an optimizer. A deployment should specify the reports or manual reconciliation it replaces, measure the residual workload, and identify which decisions improve. If purchasing or production returns are promised, those returns need a separate baseline that includes inventory losses, service consequences, and ongoing service costs.
Onboarding identifies obligations for both Silvon and the customer. That is operationally useful, but the public material supplies no measured allocation of economic downside when the expected gains fail to appear. 15 Better evidence would follow a disappointing recommendation through a corrected economic formulation and a fresh test—the process discussed in Holimization. Updating a dashboard or tuning an alert threshold alone does not establish that correction.
Supply chain score
Supply chain depth
| Criterion | Score |
|---|---|
| Economic framing | 4/10 |
| Decision end-state | 3/10 |
| Conceptual sharpness on supply chain | 5/10 |
| Freedom from obsolete doctrinal centerpieces | 2/10 |
| Robustness against KPI theater | 3/10 |
Average: 3.4/10.
Economic framing — 4/10. Inventory investment and carrying costs are explicit, but no common objective reconciles the marginal value of stock across products and competing resources. Cost visibility stops short of economic prioritization. 2
Decision end-state — 3/10. The replenishment calculations produce quantities, while the wider planning process culminates in reviewed plans and alerts. Public evidence of routine execution without SKU-by-SKU supervision is thin. 2 4
Conceptual sharpness on supply chain — 5/10. The critique of overhead-driven production runs identifies a real conflict between local accounting incentives and inventory economics. The explanation of supplier variability also reaches beyond demand alone. 7
Freedom from obsolete doctrinal centerpieces — 2/10. Service targets and conventional inventory rules govern the documented policy framework. That structure leaves practitioners to compensate for interactions the rules do not capture. 2
Robustness against KPI theater — 3/10. Sales comparisons and inventory indicators can reveal problems. More indicators cannot establish which intervention creates the most value, and the cases do not connect their reporting output to a net economic result. 6 22
Decision and optimization substance
| Criterion | Score |
|---|---|
| Probabilistic modeling depth | 3/10 |
| Distinctive optimization or ML substance | 3/10 |
| Real-world constraint handling | 4/10 |
| Decision automation and labor productivity | 3/10 |
| Resilience under real operational complexity | 3/10 |
Average: 3.2/10.
Probabilistic modeling depth — 3/10. Sparse demand receives explicit statistical treatment. The missing mechanism is a joint representation of future demand and supply uncertainty feeding economic decisions. Historical variability measurements do not supply that mechanism. 3 23
Distinctive optimization or ML substance — 3/10. The identified forecasting methods are established techniques. Their integration into a shared repository is useful engineering, but the evidence does not expose a distinctive decision algorithm. 3
Real-world constraint handling — 4/10. Shared data can represent many business dimensions, and users can compare alternative plans. A dimension in a report is not an enforced constraint on a recommendation; coupled capacity and commercial restrictions remain poorly specified. 4 8
Decision automation and labor productivity — 3/10. Wonder Meats demonstrates recurring information distribution with vendor-operated loading. The downstream inspection, decision, and intervention workload remains unmeasured. 6
Resilience under real operational complexity — 3/10. Maintained models and editable plans support adaptation. The public evidence does not trace operating failures into revised economic assumptions and validated improvements in subsequent decisions. 5 12
Product and architecture integrity
| Criterion | Score |
|---|---|
| Architectural coherence | 6/10 |
| System-boundary clarity | 7/10 |
| Security seriousness | 5/10 |
| Software parsimony and ownership cost | 4/10 |
| Compatibility with programmatic and agent-assisted operations | 5/10 |
Average: 5.4/10.
Architectural coherence — 6/10. The repository, cube, and Viewer have identifiable roles. Several processing and administrative layers still need coordinated configuration. 8
System-boundary clarity — 7/10. The product explains how it obtains operational data, presents it, and accepts planning updates. Customer examples also identify vendor-operated work, making the actual operating boundary relatively intelligible. 6 8
Security seriousness — 5/10. Granular data access and update authority are documented mechanisms. The public account is less complete at the hosting and integration boundaries where privileged services operate. 10 11
Software parsimony and ownership cost — 4/10. Concentrated infrastructure avoids an obvious collection of unrelated engines. Customer-specific models and recurring administrative tasks still create costs that the published outcomes do not quantify. 9 12 13
Compatibility with programmatic and agent-assisted operations — 5/10. The import API supports automated data movement. It is an integration interface, without a documented economic decision contract through which agents could reliably operate the supply chain. 11
Technical transparency
| Criterion | Score |
|---|---|
| Public technical documentation | 7/10 |
| Inspectability without vendor mediation | 6/10 |
| Portability and lock-in visibility | 4/10 |
| Implementation-method transparency | 6/10 |
| Evidence density behind technical claims | 5/10 |
Average: 5.6/10.
Public technical documentation — 7/10. Versioned manuals disclose concrete configuration, permission, and API behavior. They are substantially more useful than a product demonstration for assessing administration. 8 10 11
Inspectability without vendor mediation — 6/10. A reader can understand much of the operating structure before a sales call. Reproducing the replenishment engine or inspecting its treatment of uncertainty remains beyond the published material. 2 8
Portability and lock-in visibility — 4/10. Standard reporting tools expose results, but migration of executable planning behavior and the full configured model is not documented as a portable artifact. 18
Implementation-method transparency — 6/10. Onboarding, internal delivery teams, and consultant responsibilities identify who builds and supports the implementation. Effort and ongoing costs remain quotation-dependent. 12 14 15
Evidence density behind technical claims — 5/10. The administration evidence is detailed and the customer workflow is concrete. Optimization performance and economic results have much weaker support. 6 9
Vendor seriousness
| Criterion | Score |
|---|---|
| Technical seriousness of public communication | 6/10 |
| Technology roadmap integrity | 6/10 |
| Conceptual sharpness | 4/10 |
| Economic accountability | 3/10 |
| Defensibility in an agentic-software world | 3/10 |
Average: 4.4/10.
Technical seriousness of public communication — 6/10. Silvon describes a recognizable data-management role for AI and publishes operational manuals. Broad inventory-optimization language remains stronger than the disclosed economic mechanism. 2 17
Technology roadmap integrity — 6/10. The release history supports continued investment in the existing product, with future administration work identified separately. The reviewed record shows incremental delivery rather than an abandoned sequence of spectacular promises. 9
Conceptual sharpness — 4/10. The importance of business definitions is well explained. The wider value proposition still relies heavily on informed people turning visibility into better decisions, without specifying how conflicting choices are ranked. 17
Economic accountability — 3/10. Named delivery roles improve responsibility for implementation. Public results do not establish net returns after customer labor, vendor services, and maintenance, or describe correction when financial expectations disappoint. 6 15
Defensibility in an agentic-software world — 3/10. Governed business data retains value. Report navigation and interpretation face a much higher automation benchmark, and Silvon has not demonstrated that its current approach removes the resulting decision workload. 16 17
Sources
Stratum platform. Publication date not specified. Accessed November 4, 2026.
[2] Inventory Performance Management
Inventory Performance Management. Publication date not specified. Accessed November 4, 2026.
[3] Stratum Forecasting: A Solution Overview
Stratum Forecasting: A Solution Overview. Published or revised: 2016 (document metadata; modified 2022). Accessed November 4, 2026.
Planning and Forecasting. Publication date not specified. Accessed November 4, 2026.
[5] Planning and Budgeting with Stratum
Planning and Budgeting with Stratum. Published or revised: 2018 (document metadata; modified 2022). Accessed November 4, 2026.
Wonder Meats case study. Publication date not specified. Accessed November 4, 2026.
[7] Creating and Executing an Inventory Plan
Creating and Executing an Inventory Plan. Publication date not specified. Accessed November 4, 2026.
[8] Initial Setup of Application and Data
Initial Setup of Application and Data. Published or revised: 2026-02. Accessed November 4, 2026.
What’s New in Stratum 7.7. Published or revised: 2026. Accessed November 4, 2026.
Working with Roles. Published or revised: 2023-03. Accessed November 4, 2026.
[11] Working with Stratum Cloud Import API
Working with Stratum Cloud Import API. Published or revised: 2024-10. Accessed November 4, 2026.
Silvon Careers. Publication date not specified. Accessed November 4, 2026.
Silvon Hosted Services. Publication date not specified. Accessed November 4, 2026.
[14] Stratum Professional Services
Stratum Professional Services. Publication date not specified. Accessed November 4, 2026.
[15] Stratum Onboarding Services
Stratum Onboarding Services. Publication date not specified. Accessed November 4, 2026.
INSIGHT eLearning. Published or revised: 2026 schedule. Accessed November 4, 2026.
[17] How AI Is Improving the Sales Forecast
How AI Is Improving the Sales Forecast. Published or revised: 2026-09-30. Accessed November 4, 2026.
[18] Why Fortify Spreadsheets with a Data Hub and Analytics Strategy — Part 2
Why Fortify Spreadsheets with a Data Hub and Analytics Strategy — Part 2. Published or revised: 2022-06-29. Accessed November 4, 2026.
Silvon’s Partners. Publication date not specified. Accessed November 4, 2026.
Forecast Pro pricing. Published or revised: 2026-09-17. Accessed November 4, 2026.
[21] Silvon customer testimonials
Silvon customer testimonials. Publication date not specified. Accessed November 4, 2026.
Inventory performance. Publication date not specified. Accessed November 4, 2026.
Supplier performance. Publication date not specified. Accessed November 4, 2026.
[24] Flexible information delivery
Flexible information delivery. Publication date not specified. Accessed November 4, 2026.