Use case · Variance analysis
Every variance explained, all the way down to the transaction.
Cube runs variance analysis on live actuals: it finds what moved, explains why, and traces every figure to the GL transaction behind it.
Cloud & software drove 53% of the opex variance: AWS usage plus a Datadog annual true-up. Trace to source →
The problem
The close ends, and the hunt for the why begins.
Every finance leader knows the feeling of presenting a number they can't fully explain.
A week of forensic work
Actuals land, and the team spends days pulling GL exports and rebuilding pivots just to find what moved.
Answers that don't tie
The deck, the model, and the dashboard each show a different variance, and no one is sure which is right.
The why arrives too late
By the time the explanation is ready, leadership has moved on and the forecast is already stale.
One connection
How variance analysis works in Cube
Actuals flow in from every source system, Cube compares them to plan across every dimension, and the explained variance shows up where your team already works.
Cube
Compares actuals to plan across every dimension and finds the transactions behind the change.
Live data
Budget vs actual, on live actuals.
Fetch actuals straight into your own BvA model, and the variance columns stay current without a single export.
| $K | Mar · Actuals | Mar · Budget | Variance | Var % |
|---|---|---|---|---|
| Revenue | 4,820 | 4,650 | +170 | +3.7% |
| COGS | 1,210 | 1,180 | +30 | +2.5% |
| Gross profit | 3,610 | 3,470 | +140 | +4.0% |
| Payroll & benefits | 1,940 | 1,905 | +35 | +1.8% |
| Cloud & software | 486 | 402 | +84 | +20.9% |
| Travel & entertainment | 189 | 148 | +41 | +27.7% |
| Operating expenses | 2,615 | 2,455 | +160 | +6.5% |
| EBITDA | 995 | 1,015 | -20 | -2.0% |
The bridge
Budget to actual, one bar at a time.
The same variance, drawn as a bridge. Each bar is a driver of the move, and each drills to the transactions behind it.
The payoff
From what moved to why, without the forensic work.
The why, found for you
Smart Analysis runs root-cause on every material variance, across department, vendor, and entity, before you start digging.
Analysis where you work
Explanations land in Excel, Slack, and the board deck through direct, bi-directional connectivity, with nothing new for stakeholders to log into.
Traceable to the transaction
Every explained variance maps back to specific GL transactions, with cell-level access control deciding who sees what.
Adoption
Up and running without ripping anything out.
Finance-led, no-code setup with guided onboarding. Your current stack stays.
Finance-led setup
Configured by your team, guided by ours.
Rip-and-replace
Deploys alongside the tools you run today.
Onboarding
Hands-on setup and a named contact.
Cube in action
The agents run the analysis. You run the meeting.
Three FP&Agents cover variance work end to end, on clean, decision-ready data.
Smart Analysis
Instant variance and root-cause analysis, with drillable insights on every material movement.
AI Analyst
Ask why in plain language, in the browser, your spreadsheet, or chat, and get a traceable answer.
The Business Partner
Turns the explained variance into board-ready narrative in PowerPoint or Google Slides.
Where you work
One model, on every surface.
Cube delivers the same governed numbers wherever the work happens. Explore the full story on Where You Work.
Cube Workspace
The browser app where finance builds the model, sets permissions, and publishes.
Excel & Google Sheets
Bi-directional sync: fetch live actuals into the sheet, publish plans back.
AI assistants via MCP
Claude, ChatGPT, and Gemini answer from your governed numbers.
Slack & Teams
Ask a question in chat and get the governed figure back, with the trace.
PowerPoint & Google Slides
Decks with figures bound to Cube that refresh to the current numbers.
BI tools
Tableau, Looker, and Power BI read from the same model as the plan.
Where it fits
One motion, every variance conversation.
Monthly BvA package
Close lands, and the budget-vs-actual package assembles itself with commentary.
Board variance commentary
Explained movements, ready for the deck, with every figure defensible.
Department reviews
Cost-center owners see their own variances, inside their access rights.
Forecast vs actuals
Track how the latest reforecast held up, and feed what you learn into the rolling forecast.
Margin & COGS analysis
Cost, price, volume, and mix effects untangled so margin shifts make sense.
Ad hoc follow-ups
When the answer raises a new question, ad hoc reporting picks it up.
See it on your numbers.
Bring last month's toughest variance, and we'll show you how the answer gets found.
Security & governance
Numbers you can defend in front of anyone.
Cell-level access control
The CFO sees everything, a department head sees their cost center, an auditor sees read-only.
Read-only source connections
Cube reads actuals from your systems without writing back to your ERP.
Full audit trail
Every fetch, edit, and publish is logged and reviewable.
Region & retention
Data residency and retention policies configured to your requirements.
Trace to the source
Every number opens.
The +$84K on cloud and software is one click from the invoices behind it.
| Mar 03 | AWS | usage, prod & data | NetSuite | $186,400 |
| Mar 11 | Snowflake | compute + storage | NetSuite | $58,900 |
| Mar 27 | Datadog | annual true-up | NetSuite | $31,200 |
| + 14 more transactions | ||||
Already convinced?
Book a demo and skip the homework. Still researching variance analysis? The rest of this page is the deep end.
The definition
What is variance analysis?
Variance analysis
Variance analysis is the practice of comparing actual financial results to budgeted or forecasted figures, quantifying the differences, and explaining why they occurred so leadership can act on them.
Done well, it runs continuously on live actuals, and every explanation traces to the transactions behind it. That is the standard Cube holds every number to.
What goes in
What comes out
The hard part
Why is variance analysis difficult?
Six failure points show up in almost every finance team, and each one has a structural fix.
Scattered actuals
The GL, CRM, and HRIS each hold a piece of the answer. Cube unifies them into one clean, decision-ready layer.
Stale exports
A BvA built on last Tuesday's export is wrong by Friday. Cube's sync keeps actuals live in the model.
Inconsistent mappings
When accounts map differently across reports, variances appear that don't exist. Cube maps once, everywhere.
Surface-level answers
"Opex is up 6.5%" states the what. Smart Analysis finds the drivers underneath it.
The post-close crunch
The window between close and the board meeting is short. FP&Agents run the analysis the moment actuals land.
Commentary that evaporates
Explanations live in email threads and die there. Cube captures the why alongside the number, everywhere work happens.
AI without governance is a faster way to get the wrong answer.
Two ways to run it
Manual variance analysis vs Cube
| Manual process | With Cube | |
|---|---|---|
| Data collection | ✕Exports from each system, rekeyed into a master file | ✓Actuals sync automatically from hundreds of sources |
| Building the BvA | ✕Rebuilt every month from pivots and lookups | ✓The model stays connected, and the columns refresh in place |
| Finding the why | ✕Days of drilling GL detail by hand | ✓Smart Analysis surfaces drivers and materiality instantly |
| Explaining it | ✕Commentary rewritten for the deck, the email, and the dashboard | ✓One explanation flows to every surface, consistent |
| Trusting it | ✕Depends on whoever built the file | ✓Every figure traces to the GL transaction |
The landscape
Types of variance analysis
Most teams run several of these at once, and all of them get easier on one clean layer.
Budget vs actual
The monthly staple: how results tracked against the approved plan.
Forecast vs actual
How the latest reforecast held up, and where assumptions drifted.
Revenue variance
Price, volume, and mix effects separated across products and segments.
Spend variance
Vendor, department, and category movements against plan.
Headcount & comp
Hires, attrition, and comp changes reconciled to the workforce plan.
Period over period
Sequential and year-over-year movement, independent of plan.
Field guide
Variance analysis best practices
Start from one source of truth
If the deck and the model disagree, the meeting becomes about the data. Reconcile upstream, once.
Set materiality thresholds
Define what counts as material by account and percentage, and let everything below it stay quiet.
Pair every number with a why
A variance without an explanation invites the same question again next month.
Trace before you publish
Verify a sample of figures to source transactions before the number leaves finance.
Make it continuous
Teams that check variances weekly walk into close already knowing the story.
Buying guide
What to look for in variance analysis software
Live source connections
Analysis is only as current as the actuals underneath it.
Dimension-level depth
Department, entity, vendor, product: the why usually hides one level down.
Drill to the transaction
If a figure can't open to its source, it can't be defended.
Spreadsheet-native delivery
Your team's models should get smarter without being abandoned.
AI that explains
Flagging a variance is table stakes. Look for root-cause analysis grounded in your own data.
Governance built in
Role-based access and audit trails, so the analysis is safe to share.
Know why the number moved, every time.
See variance analysis run end to end on the demo, from live actuals to explained, traceable movements.
Prefer to explore first? See the product.
