Object model: Account (HubSpot: Company) = the organisation · Contact = a person at it · Lead = Salesforce's pre-qualification object, converted into Account + Contact (+ Opportunity); HubSpot instead uses lifecycle stages · Opportunity (HubSpot: Deal) = a specific potential purchase defined by amount + stage + close date · Activity = logged calls, emails, meetings, tasks. Also: Products/Price Book, Quote, Contract, Case, Campaign, contact roles (the buying group). Every record has an owner; picklists exist so reports can count things; validation rules enforce exit criteria at data entry.

Two jobs at once: the CRM is the company's memory (narrative — what they said, what changed) and its forecast (structure — stage, date, amount). Do only one and it fails at the other.

Salesforce ↔ HubSpot: org ↔ portal · Account ↔ Company · Opportunity ↔ Deal · field ↔ property · Flow ↔ Workflow · Cadence ↔ Sequence · List View ↔ Saved view. Salesforce = enterprise, every org configured differently. HubSpot = scale-up, opinionated, runnable without a specialist.

Five phrases to own: Salesforce admin (the person configuring objects, fields, validation rules, reports, permissions) · reports and dashboards (a report is a filtered, grouped, summarised query; a dashboard is a screen of charts built from reports) · opportunity stages (a picklist that is a claim about evidence) · pipeline hygiene (real stages, buyer-validated dates, real amounts, prompt closed-losts, dated next steps) · single source of truth (one authoritative system; the enemy is the shadow spreadsheet).

A day of CRM work: morning hygiene sweep (no next step, no activity 14 days) → run the call → log within five minutes (who / their words / what changed / what was agreed) → advance the stage only if the exit criterion is met → set the close date from the buyer's process → add new stakeholders as contact roles → next step with a name and a date. An empty next step is the most damning field in the system — it proves no next action was agreed. It also fails if undated, or if only you do it.

The stack in four rings: data in (Sales Navigator, ZoomInfo, Cognism, Apollo) · engagement out (Outreach, Salesloft, HubSpot Sequences) · intelligence back (Gong, Chorus, Clari) · paperwork through (Calendly/Chili Piper, PandaDoc, CPQ, DocuSign). Everything writes back to the CRM or it is failing. Glue: activity capture, bi-directional sync, enrichment.

CPQ = Configure, Price, Quote: valid combinations, priced by rule, quoted with discount approval workflows — margin protection and error prevention, part of quote-to-cash. Plan for the approval delay; better, design the deal to avoid the exception.

AI — good at: call summarisation and CRM field population · research briefs · first-draft outreach · deal-risk and forecast-hygiene flags · coaching at scale. AI — embarrassing at: fabricated personalisation (never send an unverified claim about a person or company) · generic sequences at scale (reply rates fell; Google/Yahoo/Microsoft enforce authentication, one-click unsubscribe and low spam thresholds) · over-trusted scores (a score is not evidence; exit criteria are) · unedited product claims in front of customers (a compliance risk in regulated markets). Etiquette: assistance yes, deception no; never simulate an observation you did not make; apply the defensibility test ("how did you know that?"); keep personal data in approved tools; announce recording bots and accept refusal.

UK rules, short version: UK GDPR = personal data; PECR = electronic marketing; both apply. PECR corporate subscribers (Ltd, LLP, Scottish partnerships) sit outside the consent rule for unsolicited marketing email; individual subscribers (people, sole traders, ordinary partnerships) do not. Never conceal sender identity; always give a working opt-out. Calls: screen TPS + CTPS + your own do-not-call list, identify yourself, display a number. Lawful basis is usually legitimate interests — documented assessment, not assertion. Tell people where their data came from. Objection to direct marketing is absolute — stop, everywhere, immediately. Never put patient or special-category data in a sales CRM. The Data (Use and Access) Act 2025 raised maximum PECR penalties from £500,000 to UK GDPR levels (up to £17.5m or 4% of global turnover).

No-access credibility plan (under 20 hours): free HubSpot CRM — 10 companies, 20 contacts, a named stage model with written exit criteria, 6 deals with dated next steps, one report, one dashboard (~~2 hrs) · Trailhead + a free Developer Edition org at developer.salesforce.com/signup — custom field, report, dashboard, validation rule (~~8–12 hrs) · a Sales Navigator or Apollo trial to build a real target list (~1 week, part-time). Certifications: HubSpot Academy (free, best value per hour) · Salesforce Platform Foundations, formerly Associate (entry-level, free learning, paid exam) · Salesforce Administrator (serious, but over-engineered for an AE — right for a RevOps path).

The line: "Not in a commercial environment — I've built my own: a HubSpot instance with a defined stage model and exit criteria, and a hygiene dashboard I run on myself. What I don't know is how your org is configured, and that's what I'd expect to learn in the first fortnight." Never claim enterprise experience you do not have; one follow-up question turns a skills gap into an honesty problem.