Naming conventions that survive
A field guide to naming your Novi artefacts so they’re still readable long after you’ve forgotten why you built them.
There’s a moment every modeller knows. You open a model you built eighteen months ago (maybe you built it, maybe someone who’s since left did) and you’re greeted by Table1, Copy of Revenue (2), a dictionary called New Dictionary, and a link named Link_final_FINAL. The logic still works. The numbers still tie. But nobody can read it anymore. The model survived. Its meaning didn’t.
Good naming is the difference between a model that’s an asset and a model that’s a liability waiting for the day its author goes on holiday.
Novi doesn’t make you do this. That’s the point
Some EPM platforms force a naming syntax on you: rigid codes, mandatory prefixes, structural rules baked into the tool. They do it because their models are otherwise unnavigable. Novi takes the opposite bet. With folder-based navigation, logical sorting, and a structure you can actually browse, Novi lets you name things in natural language, the way you’d say them out loud to a colleague.
“Operating Expenses.” “Products by Category.” “Allocate Overhead to Cost Centres.”
No cryptic prefixes. No DIM_ or tbl_. Just the business, named plainly. It is the same property that lets a model stay readable to a machine: a name that a colleague understands is a name an agent can understand too.
But freedom isn’t the same as a free-for-all. The fact that Novi lets you name a dictionary stuff doesn’t mean future-you will thank you for it. This is where naming becomes a craft, and, like a craft, it rewards a little discipline.
It’s a bit like writing code
Ask three programmers to name the same variable and you’ll get userCount, user_count, and numUsers. Each is defensible. What’s not defensible is using all three in the same project. Professional codebases pick a convention and hold the line, not because one style is objectively superior, but because consistency is itself the feature. A reader who’s learned your pattern can predict the next name before they see it.
Novi models work exactly the same way. The goal isn’t to find the One True Name. It’s to be the kind of modeller whose models a stranger can navigate, because every name keeps the promise the last one made.
The five habits of a model that survives
1. Name for the reader, not the writer
When you create an artefact, you have all the context in your head. Eighteen months later, that context is gone, and the only thing left is the name. Write it for the person who has nothing but the name. That person might be a colleague, an auditor, a successor, or you, with the memory wiped clean.
Adj → Manual FX Adjustment
Calc2 → Gross Margin %
2. Be specific: generic names are a debt
Data, Table1, Model, New Dictionary are not names. They’re placeholders you forgot to replace. Every generic name is a small loan against your future understanding, and the interest compounds every time the model grows. The fix costs ten seconds today and saves an afternoon later.
3. Don’t embed things that will rot
The most common way a name goes stale is that it captures something temporary as if it were permanent:
- Dates and versions:
Budget v3,Forecast_Mar. Today’s “v3” is next quarter’s confusion. Let Novi’s scenarios and versions carry time and iteration; let the name carry meaning. - Status words:
final,FINAL,temp,draft,old,do not use. These describe a moment, not a thing. If something is obsolete, archive or delete it. Don’t rename it into a tombstone. - Owner names:
Krzysztof’s model. People move on. Their models shouldn’t have to be renamed when they do.
4. Let folders do the heavy lifting
Novi’s folder structure is part of the name. If a dictionary already lives in Sales / Dimensions, calling it Sales Products Dimension is stutter: the path already said “Sales,” already said “Dimension.” Just call it Products. Good structure lets names get shorter, not longer, because the hierarchy supplies the context. Name the artefact for what it is within its folder, and let the path tell the rest of the story.
5. Pick a casing and a number, then never think about it again
Decide once: Title Case for artefacts (Operating Margin), natural sentence case for elements if that’s how the business writes them. Decide once: dictionaries, the collections, are plural (Products, Months, Cost Centres); the elements inside them are singular (Product A, January). Then stop deciding. A convention you don’t have to think about is one that survives being tired, being rushed, and being three minutes from a deadline.
A naming cheat-sheet for every Novi artefact
Novi gives you a rich vocabulary of artefacts, and each has a natural way to be named. Here’s a starting convention. Adapt it to your house style; just keep it consistent once you do.
| Artefact | Name it for… | Good | Avoid |
|---|---|---|---|
| Models | The business domain it represents | FY26 Operating Plan, Group Consolidation |
Model1, Test, KK working copy |
| Folders | The grouping, as a plain noun | Dimensions, P&L, Source Data |
Misc, Stuff, Various |
| Dictionaries | The dimension, plural | Products, Cost Centres, Months |
Dim1, New Dictionary, prod_dim |
| Elements | The real-world thing, singular | North America, Direct Labour |
NA, Item 4, xx_other |
| Hierarchies | The rollup perspective | Products by Category, Org by Region |
Hierarchy2, Rollup |
| Tables | What they hold or compute | Revenue by Product, Headcount Plan |
Table1, Copy of Sheet |
| Links | The flow, source → intent | Allocate Overhead to Products |
Link3, transfer |
| Matrices | The relationship they hold | Product × Channel Margins |
Matrix_new |
| Mapping Definitions | The translation they perform | Map GL Accounts to P&L Lines |
mapping1, convert |
| Stories | The narrative or question | Why Did Q3 Margin Fall? |
Story (2) |
| Dashboards | The audience and purpose | CFO Monthly Review |
Dashboard_final |
| Data Sources | The system of origin | SAP GL Extract, Salesforce Pipeline |
import, data2 |
Notice the pattern across the whole table: the good names read like business language, and the bad ones read like the software’s leftovers. Natural naming isn’t a style choice. It’s the whole philosophy.
You’re not labelling objects in a tool. You’re describing a business in a model that happens to run.
The eighteen-month test
Here’s a simple test you can apply to any name before you commit it:
If I came back to this in eighteen months, knowing nothing but this name, would I know what it is, and would I trust it?
If the answer is yes, ship it. If the answer is “well, I’d remember because…”, stop. You won’t remember. Name it again, for the reader who has only the name.
Novi gives you the freedom to name things naturally, and that freedom is worth spending well. Models that pass the eighteen-month test don’t just survive; they stay trusted. And a model you trust is a model people actually use to make decisions.
Consistent, natural naming isn’t bureaucracy. It’s the craftsmanship that turns a working model into a professional one, and the difference between a model that decays and one that lasts.
A model named in plain language
Novi lets you name every artefact the way you’d say it to a colleague: no prefixes, no codes, no syntax to memorise. Structure you can browse, names you can read, and a model that stays legible as it scales.