Domestic, international, spreadsheets — and where NG One stands
Companies in Serbia choose between three things: a domestic system that knows the regulation, an international one that knows the business, and a spreadsheet that knows whatever someone typed into it. Each option solves one half of the problem and leaves the other. This page describes what each category does well and where it stops — including NG One, which is built on the assumption that the line need not exist, and which does not claim to be the answer to every case.
- We compare categories, not companies
- Every category has real strengths
- Regulation and modern modules in one core
Why the choice always feels like a compromise
Because it is one. The Serbian ERP market is split along a line that is not accidental, and both sides of it have good reasons to be where they are.
Domestic systems grew up around Serbian regulation. VAT, POPDV, payroll, the retail ledger, the cash desk, travel orders, agricultural purchase — those are not modules they added, they are the reason the products exist. When a rule changes in November, they follow it, because that is their only job. The price of that specialization shows the moment you step outside accounting: a real warehouse with bin locations and directed picking, QMS, equipment maintenance, subscriptions, rental, an SLA helpdesk, a full CRM and a self-service portal generally do not exist in their core. Neither the API nor the screens are where a serious company would expect them today.
International systems grew up around a business model, not around one country. Odoo, Business Central and SAP Business One have breadth, an ecosystem, dimensions in the ledger, and a partner network that knows how to implement them. Then Serbian payroll arrives and everything stops. Payroll moves out into a separate domestic program or to an accounting firm, the cash desk and travel orders get solved by customizations that break on every upgrade, and localization is a package maintained by a third party that lags behind the law. A system that is globally mature becomes locally half-finished.
The third option is rarely called an option, and it is the most common one: one program for goods, another for finance, an accounting firm for payroll, and a spreadsheet holding them together. None of that is stupid — each piece does its job. The problem is that there is no source of truth. Every management report starts with a reconciliation, every question about what a given customer actually earned takes half a day, and nobody can explain why two numbers differ, because the difference appeared in a step nobody recorded.
NG One covers both sides of that line on one platform: Serbian regulation in the core, modern modules in the same core, dimensional posting from the first line, an API over everything, configuration without a developer, and automation that leaves a trace. Nine business spaces, some sixty technical domains and eight end-to-end processes — Lead-to-Cash, Procure-to-Pay, Plan-to-Fulfill, Record-to-Report, Hire-to-Retire, Case-to-Resolution, Project-to-Cash and Acquire-to-Retire — sit over one database, one permission model and one API. That is the outline of the system, and every claim on this page is measured against it.
Four categories, four profiles
Every card has both strengths and gaps. Anyone showing you only one of the two is selling.
Domestic ERP
Pantheon, Calculus, BizniSoft, Minimax, Ines, SAOP
Systems built around the regulation of Serbia and the region. Decades in the field, thousands of installations, and people who know them by heart. For bookkeeping and statutory obligations this is still the safest choice on the market.
Does well
- Regulation is the core business: VAT, POPDV, statutory ledgers and legislative changes are followed quickly and reliably
- Serbian payroll, with PPP-PD, CROSO and the surrounding records
- Local documents that exist nowhere else: cash desk, travel orders, the retail ledger, agricultural purchase, offsets and assignments
- Maturity, references, and a network of accountants and consultants already using them
Where it stops
- Modern modules generally do not exist in the core: a real WMS with bin locations, QMS and laboratory, CMMS, subscriptions, rental, SLA helpdesk, full CRM
- A self-service portal for employees, customers and suppliers is usually missing
- The API is partial — some functionality is reachable only through its own screen, so integration means exporting a file
- Dimensional analysis is more often an agreement about account codes than a dimension in the posting model
- The UX is inherited from a time when density was the only criterion; onboarding a new person takes longer than it should
International ERP
Odoo, Microsoft Dynamics 365 Business Central, SAP Business One
Systems built around a business model, with global breadth and an ecosystem of apps and partners. What they do, they do maturely — and they do a great deal. The trouble starts at the Serbian border.
Does well
- Breadth and depth of modules: manufacturing, projects, service, CRM, subscriptions — in one product, with years behind them
- Dimensions in the ledger and serious analytics, particularly in Business Central and SAP Business One
- A mature API, documentation and an add-on ecosystem; integrating with global tools is a solved problem
- A partner network with implementation experience and a methodology
Where it stops
- Serbian payroll is the largest gap — it is routinely moved out to a separate domestic program or an accounting firm
- Local documents are missing: cash desk, travel orders, the retail ledger, agricultural purchase and its VAT compensation, offsets and assignments
- Localization is an add-on maintained by a third party; a change in the law lands when the package catches up, not when the law takes effect
- The customization that fills those gaps becomes a cost paid again at every upgrade
- Pricing and implementation complexity are calibrated for markets where a thirty-person company is small
Spreadsheets and separate systems
Excel or Google Sheets alongside one program for goods, another for finance, and an accounting firm for payroll
The most common configuration in Serbian companies, and the one least often admitted to be a choice. It works — until a question arrives that has to have the same answer twice in a row.
Does well
- Zero entry cost and zero training: everyone can use a spreadsheet, immediately, without a consultant
- Infinitely flexible — a new column appears in ten seconds and needs nobody's approval
- Each individual program does its own job, and does it well
- No implementation project, no migration, no cutover risk
Where it stops
- No source of truth: every report starts with a reconciliation, and every month reconciles slightly differently
- No record of who changed what and when — the gap between two numbers appears in a step nobody wrote down
- Stock levels are known after the count, not at the moment a customer asks
- The knowledge lives in one person's head and in one file; both leave together
- Automation is a macro nobody maintains, and AI over data like this has nothing to read
NG One
9 business spaces, ~60 domains, 8 end-to-end processes
Both sides of the line on one platform: Serbian regulation and modern modules in the same core, on an API-first platform with dimensional posting from the first line, configuration without a developer, and automation that leaves a trace. Everything named here shares one database, one permission model and one ledger.
Does well
- Regulation and modern modules in the same outline: SEF, the EEO and EPP VAT records, POPDV, PP PDV, e-fiscalization, PPP-PD, CROSO, e-sick-leave, e-delivery-note, CRF, the retail ledger and APR statements sit beside the WMS, QMS, subscriptions and service — none of it an add-on, none of it a second product
- Dimensional posting from the first line and effective-dated statutory parameters: a March correction made in November uses March's rates and bases
- REST with an OpenAPI spec over everything, a job queue and an MCP-ready layer; no module reachable only through its own screen
- Effective-dated configurability with a clean-core approach — a consultant changes rules without a developer and without a new build
- AI with evidence: the copilot answers with the source document and a drill-down, OCR carries confidence thresholds per field type, suggestions arrive as drafts a person confirms, and all of it lands in the AI audit log
- Multitenant with database-level isolation (RLS), Serbian and English and light and dark themes from day one
Where it stops
- The statutory depth is Serbian, not global: for a company running payroll in ten countries, an international system solves more of the problem than we do
- There is no marketplace of thousands of ecosystem add-ons — what works, works out of one core and under one contract
- Clean core is a boundary: configuration runs deep, but posting logic does not get forked per client
- The dimensional model is decided before go-live; we run that work with you, not instead of you
- Implementation runs through the vendor and a narrow circle of partners, so the number of parallel projects and the dates are agreed in advance
- For five people with one warehouse and no manufacturing, this is more system than the work requires
Row by row
The values are sentences rather than ticks, because none of these questions reduces to a yes and a no.
| Criterion | Domestic ERP | International ERP | Spreadsheets and separate systems | NG One |
|---|---|---|---|---|
| VAT, POPDV and statutory tax ledgers | A strength. It is the core business; changes in the law are followed reliably and fast. | Through a localization package. It works, but it lags the law and depends on who maintains the package. | By hand. The ledger is assembled from several sources, so every change means a new sheet. | A VAT engine in the core, with effective-dated parameters: a correction to a past period follows the rates in force then. The EEO and EPP records, POPDV and PP PDV come out of the same posting rather than out of an assembly job. |
| Payroll and HR under Serbian law | A strength, and the most common reason a domestic system is kept at all. | The largest gap. Payroll is routinely moved to a separate program or an accounting firm. | An accounting firm or a separate program; the link to the ledger is manual, month after month. | Payroll, PPP-PD, CROSO and e-sick-leave in the core, on the same effective-dated rules and with the dimension on the line: labour cost posts to its cost centre and project immediately, instead of being spread across a key afterwards. |
| Local documents: cash desk, travel orders, retail ledger, agricultural purchase, offsets | Covered. Those documents grew up alongside these systems. | Mostly missing, or solved by customization that breaks on upgrade. | The cash desk and travel orders often are the spreadsheet, or a paper book. | Cash desk, travel orders, the retail ledger, agricultural purchase with its VAT compensation, offsets and assignments — in the core, on the same ledger and the same document framework as everything else. Not an afterthought, and not something that survives an upgrade only if someone remembers it. |
| E-invoicing, CRF and fiscalization | Connected, often as one integration block sharing a single fate. | Through a localization package or an intermediary. | Typed into the portal by hand, copied from another system. | The SEF channel sends and receives invoices as UBL 2.1, CRF carries public-sector receivables, and fiscalization runs through LPFR/VPFR with an ESIR. Connected in one domain graph, split into adapters with their own lifecycles — a change on one channel does not take the others down. |
| Warehousing and advanced WMS | A warehouse exists. An advanced WMS — bin locations, directed picking, wave picking — generally does not exist in the core. | It exists, often as a separate module or product, at additional cost and with an integration. | Stock in a spreadsheet. The real figure is known after the count. | A real WMS in the core: bin locations, directed put-away and picking, waves, cycle counting. It switches on per warehouse — same database, no migration, no second product. |
| Manufacturing, MRP, QMS and CMMS | Manufacturing exists at varying depth. QMS, laboratory and CMMS usually do not. | Broadly and maturely covered. The strength of this category. | The plan in a sheet, consumption on paper, costing from memory. | Manufacturing and MRP, QMS with laboratory, and CMMS stand in the same core, over the same dimensions and the same ledger — with no integration between our own modules. |
| Subscriptions, rental, service and SLA helpdesk | Rare. Usually handled with a recurring invoice and a sheet of due dates. | Present through ecosystem apps; mature in Odoo and Business Central. | A calendar reminder and good intentions. | Recurring invoices, subscriptions, rental, service and an SLA helpdesk in the same core; SLA deadlines, escalations and approval chains are effective-dated configuration, not a development cycle. |
| CRM and Partner 360 | A weakness. Usually a contact list without a pipeline. | A strength, particularly in Odoo. | A sheet of opportunities, or nothing. | Opportunities, activities, scoring and campaigns, with a Partner 360 timeline holding one partner's quotes, invoices, debt, deliveries and cases in one place — out of the same database, with no sync against a separate CRM. |
| Dimensional posting and analytics | Analytics exist, but more often as an agreement about account codes than as a dimension in the model. | Business Central and SAP Business One have dimensions. A real advantage over the domestic systems. | Analytics come from a pivot table, slightly different every month. | The dimension is part of the posting model from the first line, not a field added later. That is why a report by project, cost centre or cost object comes out of the posting rather than out of a key in a spreadsheet. |
| Configurability without a developer | Settings exist. Larger changes need the vendor and the vendor's schedule. | A strong extension layer, but a serious change needs a partner and a development cycle. | Configuration is a new column. That is exactly why it survives. | Effective-dated settings, config sets that travel from test into production, a console with an audit trail, clean core. A consultant changes rules without a developer and without a new build. |
| API and integrations | Present, but partial; some functionality is reachable only through its own screen. | Mature, documented, with an ecosystem. A strength. | None. Exchange means exporting and importing a file, by hand. | REST with OpenAPI over everything, a job queue, notifications, an MCP-ready layer. No module is reachable only through its own screen — API-first is a requirement of the core, not a later layer. |
| Automation and AI | Individual automations. AI is mostly an announcement. | Increasingly built in; depends on licence, version and market. | A macro, if someone maintains it. AI over data this scattered has nothing to read. | A daily brief, a copilot that answers with evidence and a drill-down to the source record, OCR with confidence thresholds per field type, suggestions arriving as drafts a person confirms. The AI respects the user's permissions and tenant isolation, and every automatic action carries a time, a reason and a log entry. |
| Self-service for employees, customers and suppliers | Rare. Requests arrive by email and on paper. | Present, with an additional licence and configuration. | Does not exist. Everything moves by email and phone. | Employee self-service, a customer portal and a supplier portal — license granules over the same core, not a different edition and not a different product. |
| Delivery model | Cloud and on-premise, depending on the vendor. | Odoo both; Business Central mostly cloud; SAP Business One both. | A file on a disk or in a shared folder, and the hope that someone is taking copies. | Cloud, on-premise and hybrid. The on-premise license is a granule, not a different product. |
| Scope under one ledger and one contract | Bookkeeping and local documents, in depth. Modern modules mean a second product and an integration between them. | Broad and mature — right up to the border: payroll and local documents leave the system, so the ledger splits across two places. | Does not exist. The scope is the sum of programs someone joins by hand every month. | Nine business spaces, some sixty domains and eight end-to-end processes — Lead-to-Cash, Procure-to-Pay, Plan-to-Fulfill, Record-to-Report, Hire-to-Retire, Case-to-Resolution, Project-to-Cash, Acquire-to-Retire — over one database, one permission model and one API. |
Statements about other systems are general and describe the category, not a specific version, edition or localization package. Every product named has editions and partners that cover a given gap better than the row suggests — if that is your case, tell us and we will correct the row.
Where NG One is the wrong choice
If you read the NG One column this far and thought it sounds too good, here are six situations in which we would tell you to buy something else.
A system covering nine business spaces, some sixty domains and eight end-to-end processes is the result of a series of decisions, and every decision is an advantage to someone and a cost to someone else. A dimension on every line, a clean core, one core instead of a marketplace, Serbian regulation as depth rather than as a package — those are choices, and it is only fair to say who they do not suit.
NG One is designed for a company that operates under Serbian regulation and that, alongside its bookkeeping, runs a real warehouse, manufacturing, service, projects or subscriptions — that is, for the case where the line between domestic and international hurts most. The further your case sits from that picture, the weaker the reason to change systems at all.
If what you have does the work you do, that is an answer and a legitimate one. We have said that sentence before and we will say it again — a comparison that cannot bring itself to say 'buy the other one' is not a comparison, it is an ad.
The depth is here, not across forty countries
NG One's statutory depth is Serbian: SEF, the EEO and EPP records, POPDV, PP PDV, e-fiscalization with LPFR/VPFR and an ESIR, PPP-PD, CROSO, e-sick-leave, e-delivery-notes, CRF, the retail ledger, APR statements. If you run payroll in ten countries and consolidate a group across three continents, that is the job international systems were built for. We will not pretend otherwise.
One core, not a marketplace of add-ons
Odoo and Business Central have an ecosystem where somebody has already solved a narrow vertical case — a configurator for a print shop, a module for pharmacies, an add-on for one trade. Here, what works comes out of one core, under one contract, with one party answering for it. That is an advantage while you want what is inside the outline, and a cost the moment you want something outside it.
Clean core is a boundary
Configuration runs deep: document types, states, approval chains, SLA deadlines, rules and statutory parameters change in the console, effective-dated, with an audit trail, no developer involved. What we do not do is fork the core per client. If your edge genuinely depends on posting logic rewritten from scratch for you alone, we will say no — and that 'no' is the reason an upgrade here is not a project.
The dimensional model is your decision, made before go-live
Posting with a dimension on the first line means cost centres, cost objects, projects and channels get defined during implementation, not when the first report is needed. That is real work; we walk you through it, but we cannot do it for you. A company unwilling to make that decision gets the same thing from every system: a column filled in from a key in a spreadsheet.
Implementation runs through us, not through a thousand partners
Domestic systems have a network of accountants and consultants who know them by heart; that is a real advantage and should not be talked down. Here, implementation is led by the vendor and a narrow circle of partners. You get one party answering for the core, the regulation and the integrations — and, honestly, a limited number of parallel projects, so dates are agreed in advance.
For five people and one warehouse, a spreadsheet still makes sense
NG One is a system with a chart of accounts, dimensions, permissions and approval chains. That discipline pays off only once its absence starts costing you: a day a month on reconciliation, sales promising goods that are not there, a report that arrives after the decision. Until it costs you that, a spreadsheet is a rational choice, and that is what we will tell you.
Why compare categories instead of specific products, row by row?
Because any other comparison would be wrong the day it was published. Pantheon in its full edition is not Pantheon in the base package; Odoo with a good localization package and a good partner is not Odoo out of the box. Version-level comparisons go stale in a month, and the version chosen is always the one that suits whoever is comparing. Categories change slowly, and about them we can speak honestly.
Are you claiming NG One is better than Pantheon or Business Central?
No. We claim it is laid out differently, and we explain how. Pantheon knows Serbian regulation as well as anyone, with decades behind it and a network of people who know it by heart. Business Central has an ecosystem and a partner methodology we do not have. Our claim is narrower and checkable: the line between a system that knows the regulation and a system that knows the business does not have to exist, and NG One is built without it — regulation and modern modules in one core, a dimension on the line from the first posting, an API over everything. The better system for your company is the one that covers what actually hurts; if that is only bookkeeping, the domestic systems have been doing it for decades and we are not disputing that.
What does 'domestic systems lack modern modules' actually mean?
It means the core has no bin locations with directed put-away and picking, no wave or batch picking, no cycle counting, no QMS with laboratory analyses, no equipment maintenance plan, no subscription or rental management, no helpdesk with contractual SLAs, and no full CRM with a pipeline and forecast. Some of those exist as add-ons or get solved with a second product and an integration. They do not exist inside the same system, with the same partner and the same ledger.
Why is Serbian payroll such a problem for international systems?
Because it is not a calculation but a body of rules that change mid-year and attach to forms, filings and institutions that exist only here: PPP-PD, CROSO, e-sick-leave, minimum bases, employment incentives, garnishments. A global product solves that with a localization package built by a third party — so every change in the law has two dates: the one in the Official Gazette, and the one when the package catches up. For a company the second date is what matters, and neither the vendor nor you control it. In NG One that second date does not exist: payroll, posting and statutory parameters are the same product from the same vendor, and the parameters are effective-dated, so a March calculation run in November follows March's rules.
If spreadsheets work for us, why change?
Maybe you should not. If you are five people with one warehouse and no manufacturing, a spreadsheet is a rational choice and we will say so. Changing is worth it once you start paying the cost that never appears on an invoice: a day a month on reconciliation, sales promising goods that are not there, a report that arrives after the decision was made, and knowledge that lives in one person's head. Add those four up in hours and you get the figure a system's price can actually be compared against.
How do I check that what you write about NG One is not marketing?
By making us show it, on your own case, in one afternoon. Take a VAT correction for a period a year back and watch which rates it posts under. Take an inbound invoice from SEF and follow it from receipt to posting to its line in POPDV. Ask the copilot what one customer owes you and make it click through to the document the figure came from. Open the OpenAPI spec and find the endpoint behind the screen you just looked at. Open the console, change an approval chain without a developer, then see when it takes effect and where it is recorded who changed it. None of those five checks can be faked with a slide, and none depends on taking our word for it. The one thing that should not count as proof is a sentence on this page.
Tell us what you run today
The most useful conversation starts from your system, not ours. If what you have is good enough for what you do, that is an answer too.