Hiring
Start with what the salesperson will sell and who must approve it. An AI application for document review requires a different sales conversation from an ML platform used by engineers or a data science engagement sold around a defined business problem. Write a hiring brief that names the product, buyer, deployment requirements and limits of the seller’s authority.
For a United States territory, specify account segments, named-account ownership, travel expectations and time-zone coverage rather than assuming nationwide means unrestricted prospecting. Map likely stakeholders: a business sponsor, technical evaluator, data owner, security reviewer, procurement contact and budget approver. Validate this map against your own deals.
Decide who owns renewals and expansion before advertising the position.
Build the hiring process around your actual sales cycle, not a generic software-sales template. Review recent opportunities from initial interest through technical evaluation, data access, security review, procurement and deployment approval. Record where deals stalled and which dependencies the seller could influence.
For AI products, ask how prospects establish data permissions, define acceptable outputs, evaluate errors and assign human review. Establish which claims sellers may make and which require technical approval. A polished demonstration is not a validated production result: a controlled example does not establish performance on a buyer’s workflows.
The NIST AI Risk Management Framework offers voluntary guidance for considering trustworthiness when developing, using and evaluating AI systems. Use it as a discussion aid, not evidence that your product is certified or compliant.
Have product, security and legal owners confirm current requirements and escalation routes before incorporating them into interview answers or sales materials.
Use the same scorecard for every candidate for a given role. Set weights before interviews and define what acceptable performance looks like. Do not let familiarity with AI terminology substitute for commercial judgment.
| Capability | Observable evidence |
|---|---|
| Business discovery | Connects a workflow problem to a buyer-owned outcome and asks how it will be measured. |
| Technical judgment | Separates known capabilities, assumptions and questions requiring a specialist. |
| Evaluation planning | Defines data access, acceptance criteria, error handling, owners and a decision date. |
| Account navigation | Identifies budget authority, technical influence and approval dependencies. |
| Commercial execution | Explains qualification, forecasting and negotiation decisions using specific examples. |
| Claim discipline | Declines unsupported performance claims and proposes a way to validate them. |
Score each dimension from one to four: unsupported assertion, partial evidence, convincing evidence, or repeatable evidence with thoughtful trade-offs. Record examples before discussing an overall recommendation.
Define base pay, target variable compensation, quota, crediting rules and ramp as one operating plan. Specify whether credit follows signed contracts, recognized revenue, paid pilots or another clearly defined event. Resolve shared credit between account executives, technical sellers and expansion owners.
Model quota against your own opportunity creation, conversion, contract values and sales-cycle evidence. Test what happens when a pilot succeeds technically but purchasing is delayed. Avoid rewarding pilot volume without a credible route to a commercial decision.
For compensation research, use the BLS occupational wage tables. Find the occupation that best matches the actual duties, then verify the geography, reference year and wage definitions. Do not present a broad national occupation as a local AI-sales salary quote or assume its wages equal your proposed base-plus-variable package.
Document ramp expectations, payout timing, accelerators, cancellations and territory changes before making offers. Have the appropriate internal advisers review the plan.
Source for evidence of selling comparable complexity: multiple decision-makers, technical evaluation and a measurable business case. Adjacent enterprise-software or workflow experience may be more useful than an AI job title. Browse salespeople on Salesfolks, then assess candidates against the same role requirements.
Use a recruiter screen, a structured hiring-manager interview, a bounded work sample and references with candidate permission. Assign each interviewer specific scorecard dimensions to reduce repetitive questioning.
Give candidates the same fictional brief: a buyer wants to classify incoming business documents. A potential pilot would use customer-approved sample data, agreed evaluation criteria and an explicit escalation path for errors. Provide synthetic materials only; do not request confidential customer data or reusable unpaid prospecting work.
Allow 30 minutes of preparation and 20 minutes of discussion. Ask for five discovery questions, a stakeholder map and a one-page pilot plan. The plan should cover sample suitability, success and failure criteria, human review, error escalation, responsibilities and a go/no-go meeting. Introduce a request to call a successful demonstration “production proven.” Assess whether the candidate corrects that claim without derailing the conversation.
During references, explore a relevant example already discussed with the candidate: “How did this person handle uncertainty in a customer commitment?” Seek context, not blanket endorsements.
Teach product capabilities, unsupported claims, customer workflows and escalation routes. Have the hire shadow discovery and technical reviews, build an account hypothesis and explain the difference between a demo, a pilot and production validation.
Have the seller lead discovery, map stakeholders and draft evaluation plans with a technical partner. Review recordings where permitted and available, CRM notes and follow-up messages. Coach one observable behavior at a time, such as establishing decision criteria before offering a pilot.
Assess independent qualification, forecast reasoning and handoff quality. Match revenue expectations to your documented sales cycle rather than requiring arbitrary early closes. Use weekly coaching to inspect missing evidence, next-step ownership and stalled dependencies.
Not necessarily. Require evidence that they can learn the product, manage technical evaluations and sell to the relevant buyers. Test those abilities rather than relying on category labels.
Only if the scope and support make that realistic. Define what the seller can demonstrate independently and which questions require engineering, security or product involvement.
Do not press for confidential information. Use anonymized explanations, permitted references and the standardized hypothetical exercise to assess their reasoning and personal contribution.
Use the linked NIST guidance for risk discussions and BLS tables for occupation- and geography-specific wage research. Consult O*NET when refining task and skill requirements; confirm that the occupation selected fits your role rather than copying a generic description.
When your brief and scorecard are ready, create a Salesfolks account and check current employer plans. Verify current plan details before choosing an option, and keep selection decisions anchored to observable evidence.
Candidate perspective: AI & Machine Learning Sales Jobs: A Candidate’s Guide.