Choosing a deposit platform without the feature grid.
A map of the market, ten questions to put to every vendor including us, and the situations where we would tell you to buy something else.
Four kinds of vendor call themselves a deposit solution
They do different jobs. The most common procurement mistake is to compare them on one feature grid.
| Category | What it does | Where it stops | Examples |
|---|---|---|---|
| Pricing and benchmark intelligence | Elasticity modelling, rate recommendations and peer benchmarking against pooled account data. The reference data is the asset. | It tells you the rate. It does not run the campaign, open the account or check whether the balance arrived. | Curinos, Nomis |
| Pricing execution | Takes a pricing decision and deploys it: product and fee configuration, relationship rates, deal management. | Someone upstream still has to decide what to price and why. | Zafin |
| Origination and digital sales | Opens the account fast across digital, branch and call centre, and markets to the existing base. | Strong once you know who to target. It does not explain why the book is short or size the gap. | MANTL, Narmi, Q2 |
| Core and digital banking suites | The system of record: ledger, accounts, servicing, channels. | Deposit growth is a report. Change cycles run in quarters. | Temenos, Finacle, FLEXCUBE, BaNCS, Mambu |
| EquiTrak us | Decides which initiative to run and on which clients, sizes it, drives it across channels with owners and dates, and measures whether the balance arrived and stayed. | We read benchmark data and hand off to origination. We do not replace either, and we do not replace the core. |
Ten questions that separate the shortlist
Feature checklists reward whoever wrote the longest list. These are harder to answer well, and the answers predict what year two looks like.
- Does it size the gap, or only show it?
- Most dashboards draw a variance. Few break it into named causes with a dollar value and a client list behind each. Ask: show me the shortfall broken into causes that sum to the total.
- Can it tell sticky balances from hot money before you price?
- Blanket pricing is the most expensive habit in deposit management. Ask: what does the model do differently for a twenty-year primary relationship and a rate shopper with the same balance?
- Does a recommendation become a task with an owner and a date?
- If the output is a PDF, someone has to rebuild it as work, and the link between the idea and the result is lost when they do. Ask: walk me from a recommendation to a funded account without leaving the product.
- Is the baseline locked at approval?
- If the comparison point can move afterwards, every campaign succeeds. This is the most common reason finance stops believing the numbers. Ask: what stops a campaign claiming growth that would have happened anyway?
- Does it measure the quality of funding, or only the volume?
- What matters is what kind of funding a campaign added, how long it will stay, and what it is worth after pricing, servicing and run-off. Ask: show me last quarter's campaign net of run-off.
- Who is accountable when an AI agent acts?
- A shared service account is not an answer. Each agent should have its own identity and a record of whose authority it acted under. Ask: show me one agent's identity, its risk tier, and the human sign-off on an action that moves money.
- What control model covers the AI?
- Model-risk frameworks were written for statistical models and are still catching up with generative and agent-based systems. The obligation stays with the bank either way. Ask: which controls do you provide for the parts our model-risk policy does not yet cover?
- Can it produce a full record on short notice?
- Supervisors increasingly expect the lifecycle record of a material AI system on demand: intended use, validation, monitoring, incidents. Ask: export the decision trail for one recommendation, now.
- What breaks if we swap a vendor in eighteen months?
- Vague answers about data export and integration depth are the tell. Exit terms are cheaper to negotiate before signature. Ask: if we replace our origination or benchmark provider, what stops working and for how long?
- What does the proof of value prove?
- A demo on vendor data proves the vendor can build a demo. A proof of value on your book proves whether the diagnosis holds. Ask: run it on our data and show us a finding we did not already know.
When EquiTrak is the wrong choice
- You need account opening and nothing else
- If opening a business account takes days, buy an origination platform first. We can tell it who to open accounts for. We are not a substitute for it.
- You want a rate benchmark and nothing else
- If the only question is what competitors are posting this week, a benchmarking subscription is cheaper and better.
- Your data cannot support a diagnosis
- If account-level balances, rates and maturities cannot be retrieved in a usable form, data comes first. We would say so during the proof of value, before you sign anything larger.
- Nobody owns the loop
- The platform assumes someone owns deposit growth and can decide. If that accountability is split across treasury, product and marketing with no tie-breaker, software will not fix it.
- You want fully autonomous execution
- Every consequential action here needs a person. If the goal is agents moving money without authorisation, we are the wrong vendor on purpose.
- You need it live next month
- A credible proof of value on your own book takes weeks. A production loop in thirty days is a dashboard.
Put the ten questions to us.
Send them ahead of a first meeting and we will answer each one in the product.