PART 02 · FLAGSHIP

AI × Sales

How to test lead scoring, rep assistance, forecasts, outbound, pricing, and conversational qualification while keeping people, consent, and consequential decisions under accountable control.

Representative evaluation · consent and suppression · drafts before writes

Lead Scoring & Qualification

A score is a fallible prioritisation signal, not a fact about a person or permission to contact them.

BASELINE RISK

Opaque prioritisation

Definitions differ by rep → missing CRM fields are mistaken for low intent → protected or proxy traits can influence treatment → people may be contacted without current eligibility.

CONTROLLED PILOT

Decision support with review

Define the decision and lawful inputs → exclude sensitive/proxy features → test by relevant groups and missingness → show reasons and uncertainty → let a trained person override → check consent and suppression before any contact.

Baseline worksheet: record current lead volume and observed win rate. It does not predict who will buy or an improvement.

Evidence to collect

VALIDITY
Out-of-time test
Evaluate calibration, precision/recall, missing data, drift, and value against the current process on a representative held-out period.
Evidence: versioned evaluation
FAIRNESS
Treatment review
Test error and opportunity differences across relevant groups and regions. Investigate proxies and provide a review/appeal path.
Evidence: fairness and override log
VALUE
Randomised or phased pilot
Compare incremental qualified opportunities, customer complaints, rep workload, and false exclusions—not score accuracy alone.
Evidence: signed pilot review
Vendor links below are discovery material, not evidence that a score is accurate, fair, lawful, or useful in your market. Verify dated documentation, data provenance, and case methodology before a pilot.
MYTH
"More third-party signals automatically make a lead score better."
Tap to see reality →
REALITY
Unknown provenance can add privacy, consent, bias, staleness, and identity-matching risk. Use the minimum necessary lawful data and test each feature's incremental value.

Sales Copilots

Assist with notes and drafts, but preserve recording consent, source context, human judgment, and CRM accountability.

BASELINE RISK

Fragmented records

Notes vary by rep → recording notice and retention are inconsistent → summaries omit uncertainty → CRM writes are hard to attribute.

CONTROLLED PILOT

Consented draft assistance

Provide required recording notice and consent → restrict retention/access → produce source-linked draft notes → show uncertainty → rep corrects and approves → CRM write uses least privilege, exact-field approval, idempotency, and an audit log.

Baseline worksheet: total the team's current reported admin time. It does not assume that time disappears or becomes selling time.

Pilot evidence to collect

ACCURACY
Source-linked notes
Sample names, amounts, dates, objections, commitments, and next steps; track omissions and unsupported inferences.
Evidence: corrected transcript sample
PRIVACY
Consent and lifecycle
Verify recording notice/consent by jurisdiction, purpose, access, retention, deletion, export, and redaction of sensitive data.
Evidence: approved data map
WORKFLOW
Accepted time and error cost
Measure total review time, corrected fields, missed follow-ups, rep overrides, and customer impact before expanding.
Evidence: pilot issue log
Recording and employee-monitoring obligations vary. Verify current local law, works-policy requirements, platform terms, and vendor data handling before capture.
MYTH
"A call summary is objective because the model heard the whole call."
Tap to see reality →
REALITY
Transcription and summaries can lose dialect, speaker identity, negation, emotion, and context. Keep the source available under policy, show uncertainty, and require the accountable rep to review material fields.

Sales Forecasting

A forecast is a dated estimate with uncertainty, not a single “real number.”

BASELINE RISK

Uncalibrated point estimates

Stage definitions drift → missing activity is treated as negative evidence → overrides are undocumented → finance receives a point estimate without assumptions or uncertainty.

CONTROLLED PILOT

Versioned scenarios

Define outcomes and cut-off date → validate data completeness → backtest out of time → report calibrated ranges and scenario assumptions → capture human overrides and reasons → monitor drift and decision impact.

Baseline worksheet: expose current pipeline and management's stated accuracy assumption. It does not translate an accuracy percentage into revenue.

Pilot evidence to collect

BACKTEST
Out-of-time performance
Report error by horizon, segment, region, deal size, and data completeness. Compare against simple baselines and the current process.
Evidence: reproducible backtest
DECISION
Scenario usefulness
Test whether intervals and risk flags improve a defined planning decision without creating harmful pressure or hiding uncertainty.
Evidence: forecast and override log
Accuracy percentages are meaningless without the metric, horizon, denominator, evaluation window, exclusions, and baseline. Verify all vendor claims against a dated, precise methodology.

AI SDRs & Outbound

Use assistance to research and draft only for eligible, relevant outreach; do not automate unsolicited bulk contact.

BASELINE RISK

Uncontrolled prospecting

Purchased or scraped data has unclear provenance → relevance is assumed → suppression is stale → volume harms people, brand, and sender reputation.

CONTROLLED PILOT

Eligible, reviewed outreach

Document lawful purpose and source → minimise data → check consent, suppression, region, and channel rules → draft a truthful accessible message → human approves exact audience/content/volume/window → cap frequency → handle opt-out and escalation immediately.

Baseline worksheet: calculate today's reported SDR cost, meeting count, and simple cost per meeting. It does not estimate an AI replacement.

Pilot evidence to collect

ELIGIBLE
Audience provenance
For every recipient, retain data source, lawful purpose, region/channel eligibility, suppression check, and reason the contact is relevant.
Evidence: eligibility ledger
HARM
Complaint and opt-out guardrails
Measure bounces, spam reports, opt-outs, negative replies, misrepresentation, accessibility, and domain health—not meetings alone.
Evidence: guardrail dashboard
CONTROL
Draft-only execution
Least-privilege sender, immutable suppression, idempotency, daily/recipient caps, signed exact approval, complete logs, stop switch, and rollback are mandatory.
Evidence: control test
No database size or model changes contact law, consent, suppression, or platform terms. Verify current requirements in every recipient jurisdiction before a pilot.
MYTH
"A personalised message is automatically acceptable outreach."
Tap to see reality →
REALITY
Personalisation can be inaccurate or invasive and does not establish consent or lawful purpose. Send only to eligible recipients, use truthful context, minimise data, honour suppression, and provide an accessible opt-out.

Pricing & Negotiations

Use analysis to draft a recommendation; an authorised person owns price, fairness, margin, and customer impact.

BASELINE RISK

Inconsistent discretion

Discount reasons are missing → margin and contractual constraints are checked late → similar customers can receive unexplained treatment.

CONTROLLED PILOT

Bounded recommendation

Use approved non-sensitive features → show comparables and uncertainty → test fairness and contractual constraints deterministically → create a PAUSED quote draft → authorised owner approves exact price, terms, customer, and expiry.

Baseline worksheet: expose the gross value of current discounts. It is not margin leakage or recoverable revenue.

Pilot evidence to collect

VALID
Backtest and baseline
Compare recommendations to simple rules out of time; include product cost, terms, taxes, returns, churn, and missing-data sensitivity.
Evidence: reproducible backtest
FAIR
Comparable treatment
Test unexplained differences for comparable customers and relevant groups. Exclude sensitive traits and review proxy effects.
Evidence: pricing fairness review
SAFE
Approval-bound quote
Enforce floors, caps, contract rules, expiry, idempotency, signed exact approval, write logs, stop control, and quote withdrawal/rollback.
Evidence: policy test + audit log
Vendor margin claims are not transferable benchmarks. Verify precise dated methods, included costs, product version, customer mix, and comparison design.

Conversational AI (WhatsApp & Chat)

Offer a clearly disclosed, accessible assistant on an opted-in channel and make human help easy to reach.

BASELINE RISK

Inconsistent intake

Notice, consent, sensitive questions, accessibility, retention, and handoff quality vary by conversation.

CONTROLLED PILOT

Disclosed, limited assistance

Disclose automation → confirm channel eligibility → collect only necessary fields → avoid sensitive inference → show accessible choices → escalate uncertainty, distress, complaints, regulated topics, or human requests → require approval before booking or CRM writes.

Baseline worksheet: total current inquiry volume and reported handling time. It does not estimate automation, time saving, or meetings.

Pilot evidence to collect

TASK
Representative completion
Test real intents, Arabic dialects, code-switching, noise, ambiguity, accessibility needs, and unsupported requests.
Evidence: reviewed conversation set
ESCALATE
Human handoff quality
Measure detection and transfer of distress, complaints, legal/financial/medical topics, vulnerable users, uncertainty, and direct human requests.
Evidence: escalation recall + audit
PRIVACY
Data minimisation
Review notice, consent, retention, deletion, access, sensitive-field redaction, channel rules, and CRM write approval.
Evidence: approved data-flow map
Channel capabilities and rules change. Verify current official WhatsApp/platform documentation, template/consent requirements, data handling, and regional obligations on the pilot date.

Reference pattern: a controlled sales workflow.

Start read-only and test on representative and adversarial cases. Separate tenants and use least-privilege service identities. Models may produce explanations and drafts; they cannot grant consent or approve actions. Every contact, CRM update, meeting, quote, or price write stays PAUSED until deterministic policy validates eligibility, suppression, caps, and an unexpired signed approval bound to the exact recipient/resource, content, values, and time window. Use idempotency, complete audit logs, human escalation, stop controls, and tested rollback.

HubSpotSalesforceDynamics 365WhatsApp APILinkedIn
Educational architecture only. Production readiness requires local legal, privacy, security, fairness, accessibility, reliability, and channel-policy review.

Vendor discovery links — verify before use

These links identify products mentioned in the earlier lesson. They do not prove an outcome. Locate precise, dated official documentation, terms, security/privacy material, and case methodology before relying on a feature or claim.

  1. 6sense Revenue AI — 6sense.com
  2. Demandbase One — demandbase.com
  3. Madkudu — madkudu.com
  4. Gong — gong.io
  5. Dooly.ai — dooly.ai
  6. Clari — clari.com
  7. BoostUp.ai — boostup.ai
  8. Aviso — aviso.com
  9. 11x.ai — 11x.ai
  10. Artisan — artisan.co
  11. Clay — clay.com
  12. PROS Holdings — pros.com
  13. Zilliant — zilliant.com
  14. DealHub — dealhub.io
  15. Salesloft — salesloft.com
  16. Haptik — haptik.ai
  17. Air.ai — air.ai