Build a Fully Automated High-Intent UK SMB Prospecting & Salesforce Sales Engine

Build a Fully Automated High-Intent UK SMB Prospecting & Salesforce Sales Engine Project overview I am a UK SMB Account Executive selling Revolut Business. My role is currently extremely outbound and 360: I have to find companies, research them, identify decision-makers, find contact details, check whether I am allowed to prospect the account, create the account/contact in Salesforce, send emails, manage cadences, follow up, call, qualify and close. I want to eliminate as much of the non-revenue-generating work as possible. My strengths are calling prospects, communicating with decision-makers and closing. I want a system that handles almost everything else. The desired end state is: The system prospects. The system researches. The system enriches. The system qualifies. The system checks Salesforce. The system claims accounts. The system creates contacts. The system sends personalised emails. The system runs follow-up cadences. I call, reply and close. My objective is to make my current outbound role feel as close as possible to an inbound sales role. Core business objective The system should continuously identify net-new UK SMB businesses that are highly likely to need Revolut Business and are relatively quick/easy to close. I do NOT want a system designed to generate thousands of generic leads. Quality is significantly more important than quantity. My manager expects approximately five closed deals per week, so the system must optimise for: Speed-to-close Strong Revolut Business ICP fit Evidence of likely buying need Reachability of the relevant decision-maker Low wasted prospecting time High likelihood that the account is available for me to prospect Strong email response potential Strong calling opportunities Reasonable GP potential Fully automated execution wherever possible The primary success metric is not “number of leads generated”. The primary success metric is: Does this system help me close more deals, faster? Ideal daily experience Every working morning, I want to receive an email containing 20 high-quality, sales-ready UK businesses. The email should contain two sections. Section A — 10 newly incorporated companies These should be newly incorporated UK companies that: fit Revolut Business ICP criteria; appear commercially legitimate and likely to become active businesses; have an identifiable decision-maker; have usable contact information; are eligible to be prospected; are not already owned by another Account Executive in Salesforce; have a credible reason why Revolut Business may be useful; can potentially be closed quickly. These are primarily optimised for speed-to-close. Because newly incorporated businesses are often claimed quickly by other salespeople, I also want a real-time or near-real-time alert system. If an especially strong newly incorporated company becomes available during the day, I want to be alerted immediately rather than waiting for the next morning. Preferred alert channel: Email SMS or WhatsApp if practical and approved Section B — 10 established companies These should be established UK SMB businesses with annual turnover/revenue of up to approximately £30 million. The system should prioritise companies showing strong Revolut Business fit and likely buying need, including signals such as: heavy importing; heavy exporting; significant FX exposure; international suppliers; international customers; multiple currencies; international payments; expanding into new countries; entering new markets; international hiring; ecommerce; marketplace selling; overseas contractors; business travel; cross-border operations; multiple subsidiaries; international logistics; strong growth; other evidence that Revolut Business could solve a genuine business problem. Speed-to-close is more important than maximising theoretical GP. A company with moderate GP potential that can realistically close quickly should rank above a very high-GP account that appears likely to take months. Target industries The engine should search broadly across UK SMBs, but particularly prioritise businesses such as: Ecommerce Import/export companies Wholesalers Manufacturers Logistics businesses Travel businesses Recruitment agencies SaaS companies Marketplaces International service businesses Companies with significant overseas suppliers/customers Other strong Revolut Business ICPs discovered through data This should not be a rigid industry-only filter. The system should ultimately rank businesses based on likelihood of needing Revolut Business and likelihood of closing quickly. Hard exclusions The following should normally be filtered out before they reach me: Dormant companies Dissolved/non-trading businesses Companies with no credible Revolut Business use case Weak ICP-fit businesses Companies where no relevant decision-maker can be contacted Businesses already owned by another Account Executive Businesses outside my permitted UK SMB segment Companies likely to be ineligible for Revolut Business Restricted/high-risk sectors based on approved internal eligibility rules Mining companies Banks Financial institutions Duplicate accounts Suppressed/do-not-contact prospects Other businesses that cannot realistically be onboarded Eligibility rules should be configurable rather than hard-coded permanently. Data sources I want the freelancer to use a mixture of free/public sources and carefully chosen paid sources. Potential sources include: Public/free Companies House Companies House newly incorporated company data/API Companies House director/officer information UK Trade Info Company websites Google/search data Public news Public hiring/job information Public international expansion announcements Other legitimate UK company datasets Public ecommerce/technology signals where useful Existing tools I already have access to: Salesforce Lightning LinkedIn Sales Navigator Revolut Gmail integrated with Salesforce The system can use Sales Navigator data where this is technically and contractually permitted. Paid enrichment I am willing to use a paid provider if necessary for: decision-maker identification; direct business email; email verification; direct mobile numbers; company information; buying/intent signals. Possible tools may include Apollo, Lusha, Hunter or other suitable providers. The freelancer should recommend the best option. Total ongoing external software cost should ideally remain under £100/month. Do not recommend an unnecessarily expensive stack. Salesforce integration This is a core part of the project. I have already received approval to automate Salesforce functionality including: checking whether a business is already owned; determining whether I can prospect it; claiming available businesses; creating business/account records; creating contacts; adding contact information; enrolling leads/contacts into approved cadences. The freelancer should therefore design this using official Salesforce functionality/APIs/integration methods, not browser hacking. The system should use Salesforce Lightning / official Salesforce APIs, Flow or other approved functionality as appropriate. Critical Salesforce ownership check A lead is useless to me if another Account Executive already owns the company. Therefore the system must: Discover a prospect. Research/enrich the prospect. Search Salesforce for the business. Determine whether it is already owned. Discard the prospect if another AE owns it. If available, claim the account in my name. Continue searching until the required number of available leads is achieved. For example: If the system finds 20 strong established companies but 8 are already owned, it should not send me 12 leads. It should continue searching until it has found 10 high-quality established companies that are actually available to me. The same rule applies to newly incorporated companies. Automatic claiming system should:automatically claim an available account for me as soon as it passes all filters. This is particularly important for newly incorporated companies because speed matters and other Account Executives may be targeting the same businesses. If automatic claiming is not technically possible for a particular workflow, the system should alert me immediately so I can claim the business manually. Decision-maker identification The primary decision-maker pool should begin with Companies House directors/officers. Preferred personas include: Founder Owner CEO Managing Director CFO Finance Director Head of Finance Other relevant registered director Where a director cannot be reached, the system should attempt to find another appropriate senior decision-maker such as: Finance Manager COO Head of Operations Procurement Director Head of Procurement Ecommerce Director Other senior commercial/financial decision-maker relevant to the use case The system should not throw away an excellent company simply because a Companies House director cannot be directly contacted. Contact enrichment For each company, I ideally want: Company name Companies House number Industry Decision-maker full name Job title Direct business email Direct mobile number where available Office number as fallback Evidence/source for contact data Contact-data confidence Email-verification status Direct mobile and direct email are preferred. However, an otherwise excellent lead should not necessarily be discarded just because only an office number is available. Information I want for every prospect Every lead should arrive sales-ready. For each prospect I want: Basic details Company name Companies House number Decision-maker Name Job title Email Direct mobile preferably Office phone fallback Sales intelligence Why this company fits Revolut Business Why this company may have buying intent Likely pain point Likely Revolut Business use case Estimated FX exposure where reasonably inferable Import/export activity International activity International expansion signals Other relevant buying signals Closeability A one-line explanation: “Why this company should be relatively quick/easy to close.” Example: Expanding into Germany, recurring imports, strong FX exposure, owner-led company and direct MD mobile found. This explanation should be evidence-based. Lead scoring Every account should receive a configurable score. The score should optimise primarily for: 1. Speed-to-close Highest weighting. Signals may include: obvious immediate need; strong Revolut use case; owner-led business; reachable senior decision-maker; direct mobile/email available; simple company structure; credible urgency/trigger; recently incorporated; international expansion; straightforward onboarding profile. 2. Revolut Business ICP fit Including: international activity; FX; import/export; multiple currencies; ecommerce; business travel; cards/spend; international supplier/customer payments; other monetisable Revolut Business use cases. 3. Buying-intent evidence Including: international expansion; hiring; growth; new markets; strong cross-border requirements; relevant public intent data; other legitimate intent indicators. Signal recency should help the score but should not dominate it. An older signal is acceptable if the company remains highly relevant and appears likely to buy. 4. Contactability Including: relevant senior person found; direct email; direct mobile; verified email; confidence in identity. 5. Expected GP This should influence ranking but should NOT override speed-to-close. Important scoring principle The ranking should answer: “Who should I call first if my goal is to close as many appropriate Revolut Business deals as possible this week?” It should NOT simply answer: “Which company looks biggest or most impressive?” AI research AI can be used to research and summarise businesses. For each lead, AI should create a concise sales briefing containing: Company summary What the company does. Why Revolut Business is relevant Evidence-based reasons only. Likely pain point For example: FX costs supplier payments multiple currencies international collections business cards travel spend ecommerce settlement international expansion Why now Relevant trigger or buying-intent signal. Suggested call opener Short, natural and specific. Suggested email High-quality outbound email personalised using the company information. Unknowns Anything the AI could not verify. The system must not hallucinate. If information is unknown, it should say unknown rather than inventing an answer. Email automation Outbound emails must be sent through: Salesforce using my Revolut Gmail account. Do not build a completely separate cold-email platform. The system should: Claim the prospect. Create the relevant Salesforce contact. Select the correct cadence. Generate/personalise the first email. Send the first email automatically. Run follow-up emails automatically. Stop the cadence immediately when the prospect replies. Allow me to take over manually after the reply. The cadence should continue until the prospect replies. Email opens/clicks alone should not stop the cadence. Cadence strategy I want a short, aggressive but professional cadence designed to get a response within approximately one week. Likely structure: 3–4 emails Call steps during the week 5–7 business days Immediate stop on reply The freelancer should help design the actual cadence strategy. I want different cadences for different prospect categories rather than one generic sequence. Examples: Newly incorporated company cadence Messaging focused on: setting up business finances correctly; getting business banking/payment infrastructure in place; business cards; domestic/international payments; avoiding unnecessary banking friction early. Import/export cadence Messaging focused on: FX; international supplier/customer payments; multiple currencies; reducing international payment friction. International-expansion cadence Messaging focused on: new markets; foreign currencies; cross-border operations; employee/card requirements; international payments. Ecommerce/marketplace cadence Messaging focused on: merchant/payment flows; international collections; FX; supplier payments; cards/spend. Additional cadences should be created where data supports them. Email quality requirement I do not want generic templated emails. The goal of the messaging is: get replies → start conversations → book meetings if necessary → close deals. A hybrid approach is preferred: a small number of excellent proven frameworks; automatically personalised using the company’s situation; relevant intent/fit signals; role/persona; likely pain point. The system should avoid unnecessary long emails or obviously AI-written copy. Morning email Every morning I should receive one email containing two clearly separated tables. Table 1 — Newly incorporated 10 companies. Suggested columns: Rank Company Incorporation date Director / decision-maker Title Mobile Email Revolut-fit reason Closeability reason Suggested opener Salesforce status Cadence status Table 2 — Established high-intent businesses 10 companies. Suggested columns: Rank Company Revenue/size Decision-maker Title Mobile Email Key buying signal FX/import/export evidence Likely pain Closeability reason Suggested opener Salesforce status Cadence status The email should be concise enough that I can open it and start calling immediately. Real-time newly incorporated alerts Newly incorporated businesses are competitive internally. Therefore I want a faster process: New company detected → qualifying filters → director enrichment → Salesforce ownership check → automatic claim if available → contact creation → first email → alert me to call. The goal is to reach strong new incorporations before another Account Executive claims them. Feedback / learning loop The system should improve over time. At minimum, I reliably track: Closed Won Closed Lost The workflow should store enough information about every prospect to analyse which profiles convert. For example: lead source; new vs established; industry; company size; age; import/export signal; FX signal; international activity; decision-maker role; direct mobile available; direct email available; initial lead score; cadence used; primary use case; Closed Won / Closed Lost. Over time the scoring should prioritise more companies resembling the leads I actually close. This does not need to begin as an advanced machine-learning model. A transparent rules-based optimisation process is sufficient initially. Automation requirement Once the system is live, I want minimal manual involvement before the sales conversation. For qualifying prospects, the ideal workflow is: Discover company ↓ Research company ↓ Resolve company identity ↓ Check eligibility ↓ Identify decision-maker ↓ Enrich email/mobile ↓ Score fit + buying need + closeability ↓ Check Salesforce ownership ↓ Discard if already owned ↓ Automatically claim if available ↓ Create Salesforce contact ↓ Select appropriate cadence ↓ Generate personalised email ↓ Send first email automatically ↓ Continue cadence automatically ↓ Morning list / real-time alert tells me who to call ↓ Prospect replies ↓ Cadence stops immediately ↓ I take over, call/reply and close What I do NOT need automated I am happy to: dial phone numbers myself; conduct sales conversations; qualify prospects; respond to prospects; negotiate; close; handle unusual/complex prospects manually. Do not waste development budget building a power dialler unless it becomes clearly necessary later. Security / compliance requirements The project must use authorised/approved access. Salesforce integration has been approved. Use official APIs/integration mechanisms wherever possible. Technical requirements The system should include appropriate handling for: API authentication Pagination Rate limiting Retries Error handling Idempotency Duplicate prevention Salesforce matching Failed-record handling Logging Data freshness Source provenance Confidence scoring Email verification Workflow monitoring Configurable filters Configurable score weights Configurable excluded industries Configurable cadence assignment Re-running the system must not repeatedly create the same account/contact or repeatedly enrol the same person into a cadence. Preferred technology philosophy I am not attached to any particular technology. Possible tools include: Salesforce APIs Salesforce Flow Companies House APIs UK Trade Info Python n8n Make Clay Apollo Hunter Lusha Airtable Google Sheets PostgreSQL/Supabase serverless functions LLM APIs However: do not build a complicated stack just because you can. The system should be: reliable; fast; cheap to operate; understandable; maintainable; easy to hand over. Ongoing external software cost should ideally remain under £100/month. Budget I want the freelancer to optimise scope around the highest-impact functionality. If everything cannot be completed within this budget, prioritise: external lead discovery; decision-maker enrichment; Salesforce availability check; automatic claiming; Salesforce contact creation; lead scoring; morning email; automatic cadence enrolment; personalised email generation; learning/optimisation layer. Timeline I want a usable system live as soon as reasonably possible. My sales targets are urgent. I would strongly prefer the freelancer to build this incrementally rather than disappearing for several weeks. Suggested approach: Phase 1 — Working MVP Get the first functioning prospect engine operating quickly. It should already: discover companies; enrich them; check Salesforce; provide usable leads. Phase 2 — Salesforce automation Add: claiming; contact creation; cadence enrolment; automatic emails. Phase 3 — Optimisation Improve: scoring; copy; data sources; conversion feedback; error handling; speed. I would rather have useful automation operating early and improve it than wait for a perfect final system. Suggested milestones Milestone 1 — Architecture + proof of concept Deliver: architecture diagram; data-source plan; Salesforce integration approach; sample output; proof that Companies House/company discovery works; proof that Salesforce availability can be checked. Milestone 2 — Lead engine Deliver: new-company discovery; established-company discovery; enrichment; decision-maker identification; contact data; scoring; hard exclusions. Milestone 3 — Salesforce automation Deliver: ownership check; automatic account claiming; contact creation; deduplication; cadence enrolment. Milestone 4 — Messaging Deliver: personalised email generation; multiple cadence strategies; automatic first-email send; follow-up sequence; stop-on-reply logic. Milestone 5 — Delivery + optimisation Deliver: morning email; real-time new-company alerts; Closed Won/Closed Lost feedback loop; monitoring; documentation; handover. Ownership and handover This is non-negotiable. At the end of the project I must receive: all source code; all automation workflows; all prompts; all scoring logic; all configuration; all schemas; all API documentation; setup instructions; operating instructions; troubleshooting instructions; architecture diagram; credentials/configuration handover process; list of ongoing subscriptions; list of external dependencies. Where practical, the system should be built inside accounts/workspaces that I or the company control. I do not want to remain permanently dependent on the freelancer. A recorded handover session/walkthrough should be included. Please include a short post-launch bug-fix period. Definition of success I will consider this project successful if my working day changes from: search for company → research → find director → find email → find number → check Salesforce → claim → create contact → write email → send email → follow up → call to: open morning email → see ranked sales-ready companies → call decision-maker → reply → close. The system should make the role significantly less 360/outbound. The ideal outcome is that I wake up every day with: 10 strong newly incorporated businesses; 10 strong established businesses; all available for me to prospect; relevant decision-maker identified; email/mobile enriched; Salesforce account already claimed where possible; contact already created; personalised first email already sent; follow-up cadence already running; clear explanation of why each company is worth calling; clear suggested opener. Ideal freelancer profile I am looking for a strong: RevOps / GTM Automation Engineer or Salesforce + GTM Systems Engineer Experience should ideally include: Salesforce APIs Salesforce Lightning Salesforce Flow CRM automation B2B enrichment company identity matching Companies House REST APIs lead scoring sales automation outbound cadences email personalisation n8n / Make / similar Python/API development LLM structured outputs error handling idempotency sales operations A generic AI chatbot developer is not what I am looking for. Application questions Please answer all of the following. 1. Describe the closest system you have previously built to this. What did it automate? 2. How would you automatically identify newly incorporated UK businesses as quickly as possible? 3. How would you find established UK companies showing strong cross-border, FX, import/export or international activity? 4. What data sources would you use while keeping ongoing software cost under £100/month? 5. How would you identify and enrich the relevant decision-maker? 6. How would you prevent the system from creating duplicate Salesforce accounts/contacts? 7. How would you check Salesforce ownership before claiming an account? 8. How would you make the claiming process idempotent and safe if two processes run simultaneously? 9. How would you automatically enrol a prospect into the appropriate Salesforce cadence? 10. How would you personalise emails without allowing an LLM to hallucinate company information? 11. How would you stop a cadence immediately when a prospect replies? 12. How would you design the scoring model to prioritise speed-to-close, rather than simply company size? 13. How would you handle API errors, rate limits and partial workflow failures? 14. What would you build first if I told you I needed useful results as quickly as possible? 15. What parts of this project do you believe are unrealistic or should NOT be automated? I particularly value candidates who challenge bad assumptions rather than simply saying yes to everything. Red flags I will not hire someone who: recommends browser bots as the main Salesforce solution; asks for my password; cannot explain API authentication; cannot explain idempotency; cannot explain duplicate prevention; cannot explain Salesforce Account/Contact relationships; wants to scrape LinkedIn aggressively; relies entirely on generic ChatGPT prompts; cannot explain how data sources will be verified; wants to spend hundreds per month on unnecessary tools; promises that every generated lead will definitely close; cannot explain failure handling; cannot provide documentation; wants to retain control of the finished system. Final project goal I am not hiring someone to “generate leads”. I am hiring someone to build a sales-ready opportunity engine. I want the system to perform the repetitive outbound prospecting work so that the Account Executive can focus almost entirely on: CALL → CONVERSE → QUALIFY → REPLY → CLOSE.