Controlling SaaS Spend Growth with a Dust Agent
SaaS spend grows 15%+ per year without anyone really deciding it. A Dust agent identifies in one prompt which tools doubled in cost and which are no longer being used.
The problem
SaaS is the fastest-growing spend category in tech companies — and often the least monitored. It grows through three discrete mechanisms: seat creep when teams scale without optimising licences, annual price increases that vendors apply automatically, and shadow IT where tools are subscribed to locally without central approval.
The result is predictable: each quarter, the SaaS bill is a little higher. Nobody consciously decided it. The individual increases are too small to trigger an alert (Figma goes from €1,200 to €1,440, a mere €240 difference), but cumulated across 50 tools, they represent tens of thousands of euros in annual drift.
Yet the quarterly SaaS spend review is rarely done. It requires isolating the SaaS category from exports, finding functional duplicates, identifying tools that are no longer being invoiced (and therefore perhaps no longer used, or cancelled without telling Finance). It is an hour of manual work that few FP&A analysts find time to do systematically.
What a Dust agent connected to the Spendesk MCP changes
Once the workspace admin has connected the Spendesk MCP, SaaS analysis becomes a 30-second question that any member of the finance team can ask via a shared Dust agent. The agent aggregates data by category, calculates quarterly variations, and automatically flags anomalies — sharp increases, potential duplicates, ghost subscriptions.
Analyse SaaS spend growth between Q1 and Q2 2026. Which tools had the biggest increases?
SaaS Analysis — Q1 2026 vs Q2 2026
| Tool | Q1 2026 | Q2 2026 | Change | Growth |
|---|---|---|---|---|
| Salesforce | €18,400 | €27,600 | +€9,200 | +50.0% 🔴 |
| AWS | €42,100 | €58,300 | +€16,200 | +38.5% 🔴 |
| Datadog | €8,900 | €12,200 | +€3,300 | +37.1% 🔴 |
| HubSpot | €6,200 | €7,800 | +€1,600 | +25.8% 🟡 |
| Notion | €1,800 | €2,100 | +€300 | +16.7% 🟡 |
| GitHub Enterprise | €4,400 | €4,900 | +€500 | +11.4% ✅ |
| Figma | €3,600 | €3,800 | +€200 | +5.6% ✅ |
| Zoom | €2,100 | €2,200 | +€100 | +4.8% ✅ |
| Slack | €5,200 | €5,200 | €0 | 0% ✅ |
| Intercom | €4,800 | €3,600 | -€1,200 | -25.0% ⬇️ |
| TOTAL SaaS | €97,500 | €127,700 | +€30,200 | +31.0% |
Would you like me to break down individual Salesforce invoices in Q2 to understand the source of that 50% increase?
An agent the whole finance team can query
This analysis doesn't need to be re-run by a single analyst every quarter: once the Dust agent is published for the finance team, an FP&A controller, a CFO, or a business partner can ask the same question directly — in Dust's chat or by mentioning @Dust in Slack — without depending on an export or a report prepared in advance.
Going further: 2 additional use cases
Identify functional SaaS duplicates
Functional duplicates — two tools doing the same thing — are common in fast-growing companies. Teams adopt solutions locally without central coordination.
Identify SaaS tools where we have functional duplicates — two tools doing the same job.
Based on active suppliers and their declared categories, I identified 4 areas of potential duplication:
Video conferencing:
Project management / collaboration:
Email marketing:
Application monitoring:
Would you like me to prepare a summary of these duplicates for the next SaaS review with IT leadership?
SaaS subscriptions with no recent invoice
Subscriptions that have not generated an invoice in more than 90 days may indicate an undocumented cancellation, a billing cycle change, or a ghost tool that is still active but no longer used.
Which SaaS subscriptions have not been renewed or invoiced in more than 90 days?
I cross-referenced active SaaS suppliers against invoices received in the last 90 days (10/04/2026–09/07/2026).
5 SaaS suppliers with no invoice in the past 90 days:
| Supplier | Last Invoice | Days Without Invoice | Last Payment |
|---|---|---|---|
| Hotjar | 28/02/2026 | 131 days | €420 |
| Segment (Twilio) | 15/03/2026 | 116 days | €1,200 |
| Typeform | 02/04/2026 | 98 days | €290 |
| Loom | 05/04/2026 | 95 days | €340 |
| Mixpanel | 08/04/2026 | 92 days | €780 |
Would you like me to identify who on the team submitted the last invoice for each of these tools?
What to remember
- SaaS spend grows structurally through seat creep, automatic price increases, and shadow IT — a quarterly review is essential.
spendesk_analyze_spendenables supplier-by-supplier quarter comparison in a single question, without any export.- Functional duplicates typically represent €20,000–€30,000/year in easily identified savings.
- A shared Dust agent lets the whole finance team re-run this analysis without depending on a report prepared by a single analyst.
- To go further on using the agent conversationally, see the guide on the Dust agent as a thinking partner.
Related skills
Map all SaaS subscriptions, identify price increases, detect new unauthorized tools, and calculate total annualized software cost.
Apply Pareto analysis to supplier spend, identify the long tail of micro-suppliers, and surface consolidation opportunities.