Analysing Cost Center Spend Trends with a Dust Agent
Cost center overruns are spotted too late with monthly exports. A Dust agent connected to the Spendesk MCP answers in one question — and can be shared with the whole FP&A team.
The problem
Cost center spend tracking still relies too often on static monthly exports. The FP&A analyst pulls data at month-end, builds a dashboard in Excel, and shares a report with department heads — 10 to 15 days after the period closes. By then, the overruns have already happened and it is too late to act.
The problem is structural: operational teams spend continuously, but visibility into their spending is discontinuous. An Engineering cost center that burns 40% of its July budget in two weeks only gets flagged in August during the monthly review. The overrun is locked in, explanations are painful, and corrective actions come too late to change the outcome.
Existing BI tools improve the situation but remain rigid: dashboards present predefined views that analysts cannot freely interrogate. Comparing two specific months, isolating a cost center, or understanding which suppliers are driving department-level growth requires either modifying the dashboard or exporting raw data.
What a Dust agent connected to the Spendesk MCP changes
Once the workspace admin has connected the Spendesk MCP, any Dust agent can use it — including a shared FP&A agent published for the whole finance team. Cost center trend analysis becomes conversational: the agent queries spendesk_analyze_spend directly and returns a structured comparison in seconds, in Dust's chat or in Slack, with no export and no dashboard to configure.
Compare spend by cost center in June 2026 vs May 2026. Which centers had the highest growth?
Here is the cost center spend comparison — June 2026 vs May 2026:
| Cost Center | May 2026 | June 2026 | Change | Growth |
|---|---|---|---|---|
| Engineering | €48,240 | €67,890 | +€19,650 | +40.7% 🔴 |
| Sales | €31,500 | €39,200 | +€7,700 | +24.4% 🟡 |
| Marketing | €22,100 | €26,800 | +€4,700 | +21.3% 🟡 |
| Product | €18,600 | €19,400 | +€800 | +4.3% ✅ |
| Finance & Legal | €12,400 | €11,900 | -€500 | -4.0% ✅ |
| Customer Success | €9,800 | €9,200 | -€600 | -6.1% ✅ |
| HR & People | €14,200 | €15,100 | +€900 | +6.3% ✅ |
| TOTAL | €156,840 | €189,490 | +€32,650 | +20.8% |
Would you like me to break down the top suppliers behind the Engineering increase, or project the annual run rate at this pace?
A shared agent, not a personal export
Unlike an Excel export that a single analyst builds and emails around, a Dust agent published for the FP&A team can be queried by anyone — CFO, FP&A controller, business partner — without depending on the availability of the one analyst who usually pulls the report. Everyone asks their own question, with their own filters.
Going further: 2 additional use cases
Cost centers above 110% of their monthly average
Identifying cost centers spending significantly above their usual run rate lets you alert department heads before overruns compound.
Which cost centers spent more than 110% of their monthly average over the last 3 months?
I calculated the monthly average for each cost center across April–June 2026 and compared June 2026 to that threshold.
Cost centers above 110% of their average (alert threshold):
| Cost Center | Avg April–June | June 2026 | % of Average | Status |
|---|---|---|---|---|
| Engineering | €52,180 | €67,890 | 130.1% | 🔴 Alert |
| Sales | €34,600 | €39,200 | 113.3% | 🟡 Watch |
Would you like me to set up a weekly Trigger to automatically monitor these two cost centers?
Monthly Engineering trend over 6 months with top suppliers
Understanding a cost center's trajectory over time and identifying the suppliers driving it is essential for anticipating contract renegotiations.
Show me the monthly spend trend for Engineering over the last 6 months, including the top 3 suppliers.
Engineering monthly spend — January to June 2026:
| Month | Spend | MoM Change |
|---|---|---|
| January 2026 | €41,200 | — |
| February 2026 | €43,800 | +6.3% |
| March 2026 | €47,100 | +7.5% |
| April 2026 | €49,600 | +5.3% |
| May 2026 | €48,240 | -2.7% |
| June 2026 | €67,890 | +40.7% 🔴 |
Top 3 Engineering suppliers — June 2026:
| Supplier | June Amount | Share of Center | vs May |
|---|---|---|---|
| AWS (cloud infra) | €28,400 | 41.8% | +€18,200 🔴 |
| GitHub Enterprise | €12,100 | 17.8% | +€1,200 |
| Datadog | €8,900 | 13.1% | +€2,100 |
Would you like me to compare the Engineering trajectory with Product over the same period?
What to remember
- Monthly exports catch overruns too late —
spendesk_analyze_spendenables real-time analysis in plain language, in Dust's chat or in Slack. - Month-over-month comparison by cost center surfaces anomalies that rolling averages conceal.
- A cost center at 130% of its average warrants immediate investigation — usually a single supplier explains most of the spike.
- A Dust agent published for the whole FP&A team removes the dependency on a single analyst for this data.
- To turn this check into a recurring automatic run, see the guide on Dust Triggers.
Related skills