Finance & FP&A

How to analyze vendor and category spend as Small IT and ITOps Teams

Finance & FP&AFor Small IT and ITOps Teams2 apps12 steps~24 min to set up

You have two people managing SaaS spend for 300 seats, and your visibility into what you're actually paying is scattered across a Torii dashboard that's six weeks stale, a Jira ticket queue asking why the Datadog bill jumped 40%, and a Notion doc with a vendor list nobody updates. Finance asks you quarterly to justify software costs. You pull exports from AWS Cost Explorer, screenshot a Zoom admin page, and stitch it together in a spreadsheet that's wrong before you send it. You're not overspending on purpose — you just can't see it happening in real time.

Build this on Starch → See the apps used

Finance & FP&AFor Small IT and ITOps Teams2 apps12 steps~24 min to set up

Outcome

What you'll set up

A live vendor and category spend dashboard pulling from your Plaid bank feeds so you see every SaaS charge the moment it hits — no spreadsheet, no waiting for month-end.

Automated anomaly alerts that flag when a recurring vendor charge spikes (e.g., your AWS bill doubles, or a new vendor nobody approved starts charging your card).

A custom spend breakdown by category — cloud infrastructure, SaaS tools, contractors, security vendors — that you can hand to finance without building a report from scratch every quarter.

The Starch recipe

Apps, data, and prompts

The combination of Starch apps, the data sources they pull from, and the prompts you use to drive them.

Apps used

Transaction Insights Runway Analysis

Data sources & config

Starch syncs your Plaid bank account data on a schedule — transactions, balances, and categorization update automatically so the dashboard is always current. For Stripe revenue context (useful when reporting net spend to finance), Starch also syncs your Stripe data on a schedule. Slack is connected from Starch's integration catalog; the agent queries it live to deliver your weekly spend digest.

Prompts to copy

Build me a vendor spend dashboard that pulls from our Plaid bank accounts, groups charges by vendor and category (cloud, SaaS, contractors, security), shows month-over-month trends, and flags any vendor charge that's more than 20% higher than its 3-month average.

Add a recurring subscriptions tracker that lists every vendor that's charged us in the last 90 days, the last charge date, the amount, and whether the charge has changed in the last 60 days.

Set up a weekly Slack alert that summarizes our top 10 vendor charges for the week and calls out any new vendors that haven't charged us before.

Run these in Starch → or paste them into your favorite agent

Walkthrough

Step-by-step

1 Connect your business bank accounts through Plaid — Starch syncs transactions on a schedule, so every SaaS charge, contractor payment, and cloud invoice lands in Starch automatically, no manual export required.

2 Start with the Transaction Insights app from the Starch App Store — it's pre-built to show vendor-level spend, month-over-month trends, and anomaly flags out of the box.

3 Customize the category groupings to match how your IT team thinks about spend: cloud infrastructure (AWS, GCP), SaaS tools (Okta, Jamf, Datadog, Zoom), contractors, and security vendors.

4 Tell Starch to flag any vendor charge that exceeds its 3-month rolling average by more than 20% — this catches the Datadog bill spike before finance does.

5 Build a recurring subscriptions tracker by describing it in plain language: 'Show me every vendor that's charged us in the last 90 days, sorted by total spend, with the last charge amount and whether it changed from the prior month.'

6 Connect Slack from Starch's integration catalog and set up a weekly automation that sends your team a digest of the top vendor charges and any new vendors that appeared on the account.

7 Add the Runway Analysis app alongside Transaction Insights — it layers Stripe revenue data against your Plaid expense data so you can show finance both gross spend and net burn in one view, useful when they ask whether the $8k Datadog bill is in or out of budget.

8 Build a 'new vendor alert' automation: 'Every Monday, check for any vendor that charged our Plaid accounts in the last 7 days but has no prior transaction history, and Slack me the vendor name, amount, and charge date.'

9 Create a quarterly finance export view — describe it to Starch as 'a category-level spend summary for Q1 showing total by category, top 5 vendors per category, and month-over-month delta' — and save it as a reusable dashboard.

10 Set a license reclaim trigger: if a SaaS vendor's charges drop to zero for 30 days (seat cancellation) but you're still seeing charges from a related vendor on the same contract, Starch surfaces it for review.

11 Review anomalies weekly in the dashboard — Starch highlights them automatically; your job is to triage, not hunt.

12 Share a read-only dashboard link with your finance contact so they stop asking you for spend reports mid-quarter.

See this running on Starch

Connect your tools, describe what you want, and the agent builds it. Closed beta is free.

Try it on Starch →

Worked example

Q1 2026 IT Spend Audit — 300-Seat Company

Sample numbers from a real run

AWS (cloud infrastructure)	14,200
Okta (identity)	3,600
Jamf (device management)	2,400
Datadog (observability)	6,800
Zoom (conferencing)	1,800
1Password (security)	720
Notion (docs)	480
Unknown vendor — flagged	1,200

During a routine Monday morning check in late February, the Transaction Insights dashboard flagged a $1,200 charge from a vendor with no prior transaction history — an AI tool someone on the engineering team had expensed directly to the company card without going through the approval process. Without the anomaly alert, that charge would have sat in the bank statement until the quarterly audit. Separately, Datadog came in at $6,800 for January against a $4,900 average over the prior three months — the dashboard surfaced the 38% spike automatically. The team traced it to a new APM integration rolled out in December that nobody had updated the budget estimate for. Total Q1 IT spend was $31,200. The category breakdown — $14,200 cloud, $9,800 SaaS tools, $4,200 security, $3,000 productivity — was ready as a dashboard view in Starch before the finance meeting, not assembled the night before in a spreadsheet.

Measurement

How you'll know it's working

Monthly SaaS spend per employee (target: under $105/seat for a 300-person company)

Number of vendor anomalies flagged vs. investigated — measures whether the alerting threshold is calibrated correctly

Time from charge to identification of unapproved vendor spend (was weeks, target is days or same-week)

Recurring subscription count — tracks whether the team is consolidating tools or accumulating them

AWS spend as a percentage of total IT budget — flagged to finance if it exceeds 40%

Comparison

What this replaces

The other ways teams handle this today, and how the Starch version compares.

Torii or Blissfully

Great for SaaS discovery and license management, but they don't connect to your bank feeds — so actual dollar amounts require manual reconciliation, and you're still building spend reports by hand.

QuickBooks or your bookkeeper's reports

Accurate but slow — you get the closed-month view 2-3 weeks after the fact, which means anomalies are historical by the time anyone sees them.

AWS Cost Explorer (for cloud spend only)

Excellent for AWS line-item drill-down, but it covers one vendor — you still have no unified view of what Okta, Datadog, Jamf, and twenty other SaaS tools are actually costing you month to month.

Google Sheets with bank export CSVs

Free and flexible, but it's manual every time — someone has to pull the export, clean it, re-categorize vendors, and rebuild the pivot table. That someone is probably you.

On Starch RECOMMENDED

One platform — transaction insights, runway analysis all running on connected data. Setup in plain English; numbers stay current via scheduled syncs and live agent queries.

Try it on Starch →

FAQ

Frequently asked questions

Does Starch actually read our bank transactions, or does it just pull from accounting software?

Starch syncs directly from your Plaid-connected bank accounts — so it sees the actual charge hit your account, not the version that's been categorized and massaged in QuickBooks two weeks later. That's what makes the anomaly detection useful: it's working from live transaction data, not closed-month reports.

We use AWS for most of our infrastructure spend. Does Starch connect to that too?

Yes. Starch connects directly to AWS — Cost Explorer, CloudWatch, and Organizations are all available. AWS data is queried on demand rather than synced on a schedule, so when your dashboard or automation asks for it, Starch pulls the current numbers. It won't store a historical snapshot, but for 'what is AWS costing us right now' it works well.

What if a charge comes from a vendor that's not in any SaaS catalog — a contractor invoice, a one-off tool, something obscure?

Plaid sees all transactions from your connected bank accounts regardless of vendor — there's no catalog to match against. Any charge that hits the account shows up. You can tell Starch to group unknown vendors into a 'review' category and flag them weekly.

Is this SOC 2 certified? We'd need to clear it with our security team before connecting bank accounts.

Starch is not SOC 2 Type II certified today — worth being upfront about that. If your security team requires a certified vendor for financial data connections, that's a real constraint to weigh. Plaid itself is SOC 2 compliant and handles the banking credential layer.

Can Starch also pull from our Okta or Jamf to cross-reference which employees are actually using the tools we're paying for?

Okta and Jamf are both reachable from Starch's integration catalog — connect them and the agent can query them live when your app runs. You can build a surface that shows Okta active user counts alongside the vendor's monthly charge, so you're looking at cost-per-active-seat rather than cost-per-licensed-seat. That's a custom app you'd describe to Starch — it's not a pre-built template, but it's a straightforward one to build.

We're a two-person team. How long does this actually take to set up?

Connecting Plaid and getting the Transaction Insights dashboard running takes under 30 minutes — it's a pre-built app you start from, not a blank canvas. The custom anomaly rules, Slack alerts, and category groupings are natural-language descriptions you give Starch; budget another hour to get the alerting tuned to your thresholds. The quarterly finance export view is the longest piece if you want it formatted a specific way, but that's still a description, not a build.