Skip to main content
Sales training strategy

Sales training software checklist: what to test in a demo

In a demo, look at your own scenario, not the presentation. This checklist gives specific questions in six areas — language, profiles, criteria, evidence, data and pilot — plus red flags and a post-demo decision table.

September 30, 20266 min read

Short answer

When choosing a sales training platform in a demo, look at your own scenario, not the presentation. Bring your team's hardest conversation to the demo — a price objection, for example — and run it live. Then ask questions in six areas: language, customer profile setup, control over evaluation criteria, evidence behind scores, data protection and pilot terms. For each question, ask to be shown the answer on screen rather than told it.

The checklist below breaks these six areas into practical questions. You can print it, take it to the demo and mark each line "shown", "only said" or "not available".

Preparing for the demo

  1. One real scenarioDescribe the conversation type you lose most in one paragraph: who the customer is, what they say, how a good conversation should end.
  2. Your standardThe criteria of a good conversation — even as a draft. You need to see how the platform takes it in.
  3. Two participantsA sales manager and an experienced rep. The manager looks at administration, the rep at the practice itself.
  4. A do-not-upload listDecide in advance which data you will never put into the platform — real customers' names and phone numbers, for example.

A 20-minute demo plan

  1. Minutes 1–5: the profileAsk the vendor to build a profile for your scenario on screen, or adapt a ready one to your information. Note how long it takes.
  2. Minutes 5–12: live practiceYour rep runs the conversation with that profile — in chat and on a call if possible. The manager does not step in, only takes notes.
  3. Minutes 12–17: the reportOpen the conversation's report together. Ask the reason for the score on each criterion, and ask your rep: what does this feedback tell them to do differently?
  4. Minutes 17–20: the manager's viewAsk them to show where team results, recurring mistakes and change over time are visible.

1. Language and channel

  • Can the AI customer hold a natural conversation in the languages your customers speak?
  • In which language is the evaluation report written, and can you choose it?
  • Where does practice happen — chat, calls, both? Which channel does your team really sell in?
  • How does an employee start practising — a separate app, a browser, a messenger?

2. Customer profile

  • Who builds the profile — you or the vendor? Which fields exist: customer type, behaviour, objections, language?
  • Can you attach a "good conversation" standard to the profile?
  • Does the profile rely on your product information, or does it play a generic "sales customer"?
  • What happens to earlier results when a profile changes?

3. Evaluation criteria

  • Can you set the criteria and their weights yourself, or is there one rubric for everyone?
  • Who calculates the weighting — your rule or the model? Can the model change a weight "on its own judgement"?
  • When criteria change, how are old scores shown — is the version they were scored with recorded?
  • What happens to a conversation with no written standard — is it still scored, or not scored at all?

4. Evidence behind the score

A report with a score but no explanation is useless for training. In the demo, open one conversation's report and ask: "Why 3 rather than 5 on this criterion?" A good answer points to the employee's specific message or transcript line and shows what could have been said there. A general sentence — "did not explore needs enough" — is not evidence.

Ask for the same conversation to be evaluated twice. How far apart are the scores? They may not be identical, but a large difference undermines employees' trust.

5. Data and privacy

  • Which employee data is stored: name, phone number, transcript, voice?
  • Who sees which report — employees their own, managers their team's?
  • How long is data kept and how is it deleted? Can this be written into the contract?
  • What rule or warning prevents real customer data from being put into profiles?

Obligations around personal data processing vary by country — in Azerbaijan the Law on Personal Data is the core text. Have your lawyer check the contract; a vendor's "it's all compliant" is not enough.

6. Pilot terms

  • How long does the pilot last and how many employees can take part?
  • Can you use your own profiles and standard in the pilot?
  • What data stays with you at the end of the pilot — reports, transcripts, profiles?
  • Who defines success, and is it written down before the pilot starts?

Red flags

  • A promise to "increase your sales by X per cent" — no training platform can guarantee that.
  • A claim that "the AI is completely objective" — evaluation models can be wrong and need human review.
  • Refusing to show your scenario in the demo — only a ready-made showcase scenario.
  • No explanation of the score, just a number.
  • Suggesting that scores feed straight into pay or dismissal decisions.

A decision table

After the demo, rate each of the six areas on three levels: "shown", "only said", "not available". Move the "only said" items to a list to test in the pilot. If there are two or more "not available" items in areas that matter to you (language or criteria control, say), a pilot is not worth it. For the wider frame of the choice, the comparison of AI simulation and classic training helps.

Limitations

The checklist organises demo questions but does not replace a pilot: some problems only show up after two or three weeks of real use — how willingly employees come back to practise, for example. The final word on legal compliance belongs to your lawyer.

Checking Vexvon AI Training against this list

For Vexvon AI Training the answers are: practice runs as a written chat in Telegram and as an AI test call to the employee's phone; interface and report languages are Azerbaijani, Russian and English. The company builds the profiles — customer type, behaviour, objections and a good-conversation standard; a test call with no standard is not scored. The company can set its own criteria and weights; weights come from your configuration, not from the model, and each evaluation keeps the weights and rubric version it used. Strengths and weaknesses are tied to the employee's own messages. We recommend checking all of this with your own scenario in the demo.

Next step

Print this list and write a one-paragraph description of your hardest conversation — with those two, any demo becomes more useful. More articles are in sales training strategy; for a demo with your own scenario, contact us.

Further reading on this topic: AI sales training pilot, sales training privacy.

Live demo

Ready? Let's start

See Vexvon live in a 10-minute demo.

  • A scenario built for your business
  • A live sample call
  • A tour of the platform
Get a demoorBook a meeting

Your details are used only for the demo and to get in touch.