Skip to main content
Skip to article

15 Questions to Ask Before Buying Academy Software

A vendor evaluation checklist for academy software: 15 questions across data, money, chess capability, AI safety, and operations, with the red flag for each.

By the ChessCore team · Published June 10, 2026 · Updated July 2, 2026 · 12 min read

Bottom line

Before buying academy software, ask fifteen questions across six areas: data ownership and export, money handling, chess capability, AI safety, operations, and contract terms. Each question has an answer you want and a red flag that should end the call. The checklist is vendor neutral on purpose: every question here applies to ChessCore exactly as it applies to anyone else.

TL;DR

  • Organise your vendor evaluation by what each question protects: your data, your money, your chess teaching, your reputation, and your exit.
  • For every question on the checklist there is an answer you want and a red flag; a vendor who dodges the question is giving you the red flag.
  • Run the demo on your own roster spreadsheet and your worst make-up class scenario, never on the vendor's polished sample data.
  • Read the contract for three things before signing: lock-in terms, price escalation, and what happens to your data when you leave.

Key facts

  • Human oversight of consequential automated outputs is a core control in the NIST AI Risk Management Framework, which makes 'can the AI message parents without approval?' a legitimate vendor question, not paranoia. (nist-ai-rmf)
  • Recurring fee collection runs on payment provider rails such as Razorpay subscriptions and Stripe Billing, so the name of the provider behind a vendor's 'integrated payments' is a fair and answerable question. (razorpay-subscriptions)
  • Game syncing for chess academies is technically verifiable: Lichess publishes a free public API, so a vendor claiming sync should be able to show a student's real games arriving from it. (lichess-api)
  • In ChessCore, every AI-drafted parent message waits in an approval queue until a named human approves it, and figures like attendance, fees, and ratings are injected from records, not written by the model. (product behavior)

What should you ask about your data?

Ask three data questions before anything else: can you export everything, who owns the records, and what happens when you leave. Your roster, attendance history, payment records, and student notes are the academy's institutional memory, and the software you are evaluating will hold all of it. Data questions come first because every other promise a vendor makes is recoverable; losing five years of records is not.

  1. 1Question 1: Can we export everything ourselves, any time, in a format another system can read? The answer you want: yes, self-serve, in CSV or another open format, covering students, attendance, payments, and notes. The red flag: 'contact support for an export', exports limited to PDF, or an export that omits payment history.
  2. 2Question 2: Who owns the data we put in? The answer you want: you do, stated plainly in the contract, with the vendor acting only as a processor. The red flag: licence language that grants the vendor broad rights to use your student data, or a salesperson who has to check.
  3. 3Question 3: What happens to our data when we cancel? The answer you want: a defined window to export after cancellation, then deletion on a stated schedule you can request in writing. The red flag: no deletion policy at all, or data retained indefinitely 'for compliance' with no citation of which compliance.

A useful follow-up on question 1 is to ask the salesperson to run the export live on the demo account and open the file. An export feature that exists on the pricing page but produces an unreadable dump is functionally a lock-in mechanism. Our own data handling commitments are documented on the security page, and the same live-export test applies to us.

What should you ask about money handling?

Ask who actually processes the payments, where the money lands, and how refunds and failures are handled. Fee collection is the workflow where software errors cost real money and parent trust at the same time, so vague answers here matter more than anywhere else.

  1. 1Question 4: Which payment provider sits underneath your billing? The answer you want: a named, regulated provider such as Razorpay for Indian academies [2] or Stripe [3], with the vendor able to say which one and why. The red flag: 'we have our own payment system' with no underlying provider named.
  2. 2Question 5: Does money settle directly to our account, or does it pass through yours? The answer you want: direct settlement from the provider to your academy's account. The red flag: the vendor collects on your behalf and pays you out later, which puts your cash flow inside their solvency.
  3. 3Question 6: How are refunds, failed mandates, and partial payments handled? The answer you want: a documented flow you can trigger yourself, with receipts updated automatically. The red flag: refunds require a support ticket, or failed UPI mandates simply vanish from the dashboard.
app.chesscore.io/fees

Fees · March · all batches

96% collected

Collected · this term

96%
NovJun

Aarav R. · March · ₹4,500

UPI AutoPay · receipt sent to Sharma family

Paid

Vihaan S. · March · ₹4,500

Payment link opened in WhatsApp

Due 5 Jul

Batch B2 · 14 of 14 reconciled

No bank-transfer screenshots needed

Auto

AutoPay mandates · 31 active

Renew monthly without reminders

Active
ChessCore's fee dashboard in the demo academy: collection status, UPI AutoPay mandates, and payment links, with every figure traceable to a provider record.

ChessCore's answers, for the record: Razorpay with UPI AutoPay for India and Stripe elsewhere, direct settlement, and receipts generated from provider records [2][3]. For plan-level details see the pricing page rather than taking a blog post's word for it. The point of stating this is that question 4 through 6 should be this easy for any vendor to answer.

What should you ask about the chess layer?

Ask whether student games sync automatically, which engine reviews them, and whether progress reports cite real evidence. This is the section where generic class management software quietly fails, because attendance and invoicing look identical for chess, music, or maths tuition, but the teaching layer does not. Our comparison of class management software and chess-native tools covers that gap in depth.

  1. 1Question 7: Do student games arrive automatically from where they actually play? The answer you want: first-class sync from Lichess via its public API [4], plus import from Chess.com. The red flag: coaches uploading PGN files by hand, which is the spreadsheet problem wearing a software badge.
  2. 2Question 8: What produces the engine evaluations in game reviews? The answer you want: a named engine, Stockfish in ChessCore's case, with the language model never evaluating positions. The red flag: 'our AI analyses the games', with no engine named.
  3. 3Question 9: When a report says a student improved, what evidence sits behind the sentence? The answer you want: linked games, rating history from synced platforms, and attendance records. The red flag: reports that are pure prose with no clickable trail back to data.

Question 9 is the quiet differentiator. Any tool can generate a paragraph that says a student is improving. A tool built for chess academies can show the parent the three games, the rating chart, and the attendance record that justify the paragraph. If you mostly need scheduling and invoicing, our spreadsheets versus academy software comparison is the honest place to start instead.

What should you ask about the AI?

Ask whether the AI can message parents without a human approving it, where the numbers in its drafts come from, and what it does when data is thin. 'AI-powered' appears on nearly every vendor site now; these three questions separate AI built for trust from AI built for the demo.

  1. 1Question 10: Can any AI-generated message reach a parent without a named human approving it first? The answer you want: no, never, with an approval queue you can see on screen. The red flag: auto-send features marketed as a convenience, or approval that can be switched off globally.
  2. 2Question 11: Where do the numbers in AI drafts come from? The answer you want: ratings, attendance counts, and fee amounts are injected from database records and visibly marked in the draft, so the model cannot alter them. The red flag: the model writes the numbers itself and the vendor calls occasional errors 'hallucinations' as if naming the failure excuses it.
  3. 3Question 12: What does the AI do when a student has almost no data? The answer you want: it says so, and the draft flags the gap. The red flag: a fluent, confident report about a student with two synced games and patchy attendance.

Why this is a buying question, not a tech question

Human review of consequential automated output is a named control in the NIST AI Risk Management Framework [1]. A message about a child's progress or a family's money is consequential output. You are not being difficult by asking; you are doing exactly what the governance framework says a deployer should do.

app.chesscore.io/reports/march-batch-b2

March progress report · Aarav R.

1 of 14 drafts for Batch B2

AI draft · awaiting you

Aarav attended 11 of 12 classes this month and played 18 rated games, moving from 1348 to 1395.

His endgame conversion improved clearly. Next month the focus is back-rank defense, based on the mistakes tagged in his last three games.

Highlighted numbers come from your attendance and rating records. The AI cannot change them.

Goes to the Sharma family after approval

EditApprove & send
An AI-drafted report in ChessCore's approval queue: record-backed numbers are highlighted, and nothing sends until a coach approves it.

We wrote a full piece on why every AI message to a parent needs an approval queue, and a companion guide to the wider set of AI guardrails worth demanding in coaching software. For this checklist, the short version is enough: if the vendor's answer to question 10 is anything other than an unconditional no, the remaining questions stop mattering.

What should you ask about operations, support, and pricing?

Ask how multiple coaches share the system, whether actions leave an audit trail, and whether pricing is public. These three round out the fifteen, and they predict what living with the software feels like in month eight rather than in the demo.

  1. 1Question 13: Can coaches have different roles and permissions? The answer you want: roles that let an assistant coach mark attendance without seeing payment data. The red flag: one shared login for the whole academy.
  2. 2Question 14: Is there an audit trail of who changed, sent, or approved what, and when? The answer you want: yes, visible to the owner, covering edits to records and every approved parent message. The red flag: no history at all, which turns every dispute into memory against memory.
  3. 3Question 15: Is pricing published, and what exactly is metered? The answer you want: a public pricing page and a plain answer on whether you pay per student, per coach, or per branch. The red flag: 'talk to sales' as the only path to a number for a small academy.

How do you run a demo that reveals the truth?

Run the demo on your data and your problems, never on the vendor's sample academy. A canned demo is a performance: the sample students all have rich game histories, the fees are all paid, and every workflow has been rehearsed on exactly that data. The software's real character only shows when it meets the mess you actually live with, so the single most useful thing you can do before any sales call is to prepare two artefacts and insist the demo runs on them.

The first artefact is your roster spreadsheet, exactly as it exists today, with its inconsistent name spellings, the column someone added in 2023 and never explained, the students with no Lichess username, and the family with three children under one phone number. Send it ahead and ask the vendor to import it live at the start of the call. You will learn more in those ten minutes than from any feature tour: how import errors are surfaced, whether duplicates are detected, what happens to the rows that do not fit, and whether the salesperson reaches for excuses or for tools. When ChessCore's demo academy shows 248 students imported from Sheets, that import flow is precisely the thing we are inviting you to stress with your own file.

The second artefact is your worst make-up class scenario, written out as one paragraph before the call. Every academy has one: the student who missed four classes across two batches, holds a credit from last term, has a sibling in another batch, and whose parent has already been promised something by a coach who has since left. Read it out and ask the vendor to model it on screen, end to end, including what the parent sees. Generic tools handle the happy path; this scenario is where scheduling, credits, communication, and records collide, and where you find out whether the software was built by people who have actually run classes.

  1. 1Send your real roster spreadsheet ahead and open the call with a live import of it.
  2. 2Read out your worst make-up class scenario and ask to see it modelled end to end.
  3. 3Ask for the data export to be run live on the freshly imported data, and open the file.
  4. 4Ask to see an AI draft for a student with thin data, and watch whether it admits the gap.
  5. 5Ask to see the audit trail entry for an action performed during the call itself.

The one-line demo policy

If the vendor will not run the demo on your data, treat that as the answer to most of the fifteen questions. Confidence in a product looks like 'send us your messiest spreadsheet'; the absence of that confidence looks like a beautifully rehearsed sample academy.

What belongs in the contract before you sign?

Three things belong in the contract in writing: the lock-in terms, the price escalation rules, and the data deletion process on exit. Verbal assurances from a sales call do not survive a vendor's pricing review eighteen months later, so anything you were promised out loud should appear on paper before you sign.

Contract termWhat you want in writingRed flag
Commitment and renewalMonthly or annual term with explicit notice period, no silent auto-renewal into a longer termMulti-year minimums for a small academy, or renewal terms that change without consent
Price escalationA cap or notice period on price increases for your existing planPricing 'subject to change at any time' with no notice commitment
Data export at exitSelf-serve export of all records available through the end of a defined post-cancellation windowExport access cut off the day the subscription lapses
Data deletion at exitDeletion on a stated schedule after the export window, confirmable in writing on requestIndefinite retention, or silence on deletion entirely
Payment unwindingClear ownership of provider accounts and mandates so subscriptions move with youPayment mandates bound to the vendor's account rather than yours

This checklist is vendor neutral on purpose, and that is not a rhetorical flourish: ChessCore expects to be asked all fifteen questions and to answer them on screen, on your data. If you want the wider context on what the category is and who needs it, start with our pillar guide to chess academy management software, then bring the checklist, your spreadsheet, and your worst scenario to a demo.

Frequently asked questions

What questions should I ask before buying academy software?

Ask fifteen questions across six areas: data (export, ownership, deletion on exit), money (the underlying payment provider, settlement, refunds), the chess layer (game sync, the engine behind reviews, evidence in reports), AI safety (approval before parent-facing sends, where numbers come from, behaviour on thin data), operations (roles, audit trail, public pricing), and the contract (lock-in, price escalation, data deletion). For each, know the answer you want and the red flag before the call starts.

How do I evaluate coaching software in a demo?

Refuse the canned demo and run the call on your own materials: send your real roster spreadsheet ahead and ask for a live import, then read out your worst make-up class scenario and ask to see it modelled end to end, including what the parent sees. Finish by running the data export live and opening the file. Vendors comfortable with that format have a working product; vendors who insist on their sample academy are telling you something.

What is the biggest red flag when buying academy software?

An AI feature that can message parents without a named human approving the send. It combines the worst failure modes on the checklist: a probabilistic writer, your academy's reputation, and a child's family on the receiving end, with no human gate between them. Close behind it are the absence of self-serve data export and a vendor who collects fees into their own account before paying you out.

Do these questions apply to ChessCore too?

Yes, deliberately. The checklist is vendor neutral: ChessCore's answers are first-class Lichess sync with Chess.com import, Stockfish for every engine evaluation, numbers injected from records into AI drafts, an approval queue in front of every parent-facing send, Razorpay and Stripe underneath payments, and self-serve export. We expect you to ask for all of it on screen during a demo rather than taking this paragraph's word for it.

Is an academy software checklist different from a school ERP checklist?

The data, money, operations, and contract questions overlap heavily, but the chess capability section does not: a school ERP evaluation never asks where engine evaluations come from or whether games sync from Lichess. If a tool answers the generic questions well but has no chess layer at all, it is class management software, a different category that suits some academies and quietly underserves others.

Sources

  1. [1]NIST AI Risk Management Framework · accessed 2026-06-10
  2. [2]Razorpay Subscriptions documentation · accessed 2026-06-10
  3. [3]Stripe Billing documentation · accessed 2026-06-10
  4. [4]Lichess.org API documentation · accessed 2026-06-10

Written by the ChessCore team

Drafted with AI, fact-checked and approved by a human before publishing, the same guardrail our product applies to every report it sends. Last updated July 2, 2026. Read our editorial standards.

Bring the checklist to a ChessCore demo

Send us your real roster spreadsheet and your worst make-up class scenario, then ask all fifteen questions on screen. That format is the demo we prefer to give.

Request a demo

Keep reading