salesprompter-cli 0.1.20 → 0.1.23

package/dist/deel-outreach.js

@@ -0,0 +1,454 @@
+import { classifyDirectPathSegment } from "./direct-path.js";
+import { parseHistoricalQuery } from "./historical-queries.js";
+function marketCountries(market) {
+    if (market === "dach") {
+        return ["DE", "AT", "CH"];
+    }
+    if (market === "europe") {
+        return ["DE", "AT", "CH", "NL", "GB", "FR", "SE", "DK"];
+    }
+    return [];
+}
+function sqlStringList(values) {
+    return values.map((value) => `'${value.replaceAll("'", "\\'")}'`).join(", ");
+}
+function broadTitlePredicateSql(field) {
+    return [
+        `LOWER(${field}) LIKE '%head of hr%'`,
+        `LOWER(${field}) LIKE '%head of human resources%'`,
+        `LOWER(${field}) LIKE '%head of people%'`,
+        `LOWER(${field}) LIKE '%head of payroll%'`,
+        `LOWER(${field}) LIKE '%people services%'`,
+        `LOWER(${field}) LIKE '%shared services%'`,
+        `LOWER(${field}) LIKE '%total rewards%'`,
+        `LOWER(${field}) LIKE '%compensation%'`,
+        `LOWER(${field}) LIKE '%benefits%'`,
+        `LOWER(${field}) LIKE '%hr business partner%'`,
+        `LOWER(${field}) LIKE '%hrbp%'`,
+        `LOWER(${field}) LIKE '%people operations%'`,
+        `LOWER(${field}) LIKE '%people ops%'`,
+        `LOWER(${field}) LIKE '%hr operations%'`,
+        `LOWER(${field}) LIKE '%hr administration%'`,
+        `LOWER(${field}) LIKE '%hr director%'`,
+        `LOWER(${field}) LIKE '%director hr%'`,
+        `LOWER(${field}) LIKE '%chief people officer%'`,
+        `LOWER(${field}) LIKE '%chro%'`,
+        `LOWER(${field}) LIKE '%leiter personal%'`,
+        `LOWER(${field}) LIKE '%leiter der personalabteilung%'`,
+        `LOWER(${field}) LIKE '%personalleitung%'`,
+        `LOWER(${field}) LIKE '%bereichsleiter personal%'`,
+        `LOWER(${field}) LIKE '%geschäftsbereichsleiter personal%'`,
+        `LOWER(${field}) LIKE '%geschaeftsbereichsleiter personal%'`
+    ].join("\n      OR ");
+}
+function titleExclusionPredicateSql(field) {
+    return [
+        `LOWER(${field}) NOT LIKE '%assistant%'`,
+        `LOWER(${field}) NOT LIKE '%assistent%'`,
+        `LOWER(${field}) NOT LIKE '%assistentin%'`,
+        `LOWER(${field}) NOT LIKE '%recruitment%'`,
+        `LOWER(${field}) NOT LIKE '%recruiter%'`,
+        `LOWER(${field}) NOT LIKE '%talent acquisition specialist%'`
+    ].join("\n    AND ");
+}
+export function buildDeelOutreachExportSql(options) {
+    const countries = marketCountries(options.market);
+    const countryClause = countries.length > 0 ? `UPPER(CAST(p.company_countryCode AS STRING)) IN (${sqlStringList(countries)})` : "TRUE";
+    return `WITH hr_leadlists AS (
+  SELECT
+    CAST(leadListId AS STRING) AS leadListId,
+    CAST(query AS STRING) AS query,
+    leadList_container_ts
+  FROM \`icpidentifier.SalesPrompter.leadLists_raw\`
+  WHERE query IS NOT NULL
+    AND LOWER(CAST(query AS STRING)) LIKE '%/sales/search/people%'
+    AND REGEXP_CONTAINS(LOWER(CAST(query AS STRING)), r'type%3afunction[^)]*id%3a12[^)]*selectiontype%3aincluded')
+    AND NOT REGEXP_CONTAINS(LOWER(CAST(query AS STRING)), r'id%3a12[^)]*selectiontype%3aexcluded')
+  QUALIFY ROW_NUMBER() OVER (
+    PARTITION BY CAST(leadListId AS STRING)
+    ORDER BY leadList_container_ts DESC
+  ) = 1
+)
+SELECT
+  CAST(p.leadListId AS STRING) AS leadListId,
+  CAST(p.contactId AS STRING) AS contactId,
+  CAST(p.linkedin_contacts_companyId AS STRING) AS companyId,
+  CAST(p.firstName_cleaned AS STRING) AS firstName,
+  CAST(p.lastName_cleaned AS STRING) AS lastName,
+  CONCAT(
+    COALESCE(CAST(p.firstName_cleaned AS STRING), ''),
+    IF(p.firstName_cleaned IS NOT NULL AND p.lastName_cleaned IS NOT NULL, ' ', ''),
+    COALESCE(CAST(p.lastName_cleaned AS STRING), '')
+  ) AS fullName,
+  CAST(p.jobTitle AS STRING) AS jobTitle,
+  CAST(p.companyName AS STRING) AS companyName,
+  CAST(p.linkedin_companies_handle AS STRING) AS linkedin_companies_handle,
+  CONCAT('https://www.linkedin.com/company/', CAST(p.linkedin_contacts_companyId AS STRING)) AS companyUrl,
+  CAST(p.companySize AS STRING) AS companySize,
+  CAST(p.company_countryCode AS STRING) AS countryCode,
+  CAST(p.headquarters AS STRING) AS headquarters,
+  CAST(p.domain AS STRING) AS domain,
+  CAST(p.domain_linkedin AS STRING) AS domain_linkedin,
+  CAST(p.industry AS STRING) AS industry,
+  CAST(p.contact_location AS STRING) AS contact_location,
+  CAST(p.contact_companyLocation AS STRING) AS contact_companyLocation,
+  SAFE_CAST(p.tenureAtCompany AS FLOAT64) AS tenureAtCompany,
+  SAFE_CAST(p.tenureAtPosition AS FLOAT64) AS tenureAtPosition,
+  CAST(p.contact_summary AS STRING) AS contact_summary,
+  CAST(p.contact_titleDescription AS STRING) AS contact_titleDescription,
+  CAST(p.leadList_container_ts AS STRING) AS leadList_container_ts,
+  CAST(p.timestamp AS STRING) AS company_ts,
+  l.query AS leadListQuery,
+  CAST(p.email AS STRING) AS email,
+  SAFE_CAST(p.emailScore AS INT64) AS emailScore,
+  SAFE_CAST(p.gender AS INT64) AS gender,
+  COALESCE(p.emails_accept_all, FALSE) AS emails_accept_all
+FROM \`icpidentifier.SalesGPT.leadPool_new\` p
+JOIN hr_leadlists l ON CAST(p.leadListId AS STRING) = l.leadListId
+WHERE CAST(p.functionId AS STRING) = '12'
+  AND ${countryClause}
+  AND CAST(p.email AS STRING) IS NOT NULL
+  AND COALESCE(p.email_invalid, FALSE) = FALSE
+  AND COALESCE(p.email_bounced, FALSE) = FALSE
+  AND COALESCE(p.company_blacklisted, FALSE) = FALSE
+  AND COALESCE(p.company_inSequence, FALSE) = FALSE
+  AND COALESCE(p.contact_replied, FALSE) = FALSE
+  AND COALESCE(p.contact_bounced, FALSE) = FALSE
+  AND COALESCE(p.jobTitle_blacklisted, FALSE) = FALSE
+  AND COALESCE(p.domainBlacklisted, FALSE) = FALSE
+  AND CAST(p.domain AS STRING) IS NOT NULL
+  AND CAST(p.firstName_cleaned AS STRING) IS NOT NULL
+  AND CAST(p.lastName_cleaned AS STRING) IS NOT NULL
+  AND SAFE_CAST(p.emailScore AS INT64) >= ${Math.trunc(options.minEmailScore)}
+  AND CAST(p.companySize AS STRING) IN ('200-499', '501-1000', '1001-5000', '5001-10.000', '10.000+')
+  AND (
+      ${broadTitlePredicateSql("CAST(p.jobTitle AS STRING)")}
+  )
+  AND ${titleExclusionPredicateSql("CAST(p.jobTitle AS STRING)")}
+QUALIFY ROW_NUMBER() OVER (
+  PARTITION BY LOWER(CAST(p.email AS STRING))
+  ORDER BY SAFE_CAST(p.emailScore AS INT64) DESC NULLS LAST, p.timestamp DESC, p.contact_ts DESC
+) = 1
+ORDER BY SAFE_CAST(p.emailScore AS INT64) DESC NULLS LAST, p.timestamp DESC
+LIMIT ${Math.trunc(options.limit)}`;
+}
+function toNullableString(value) {
+    if (value === null || value === undefined) {
+        return null;
+    }
+    const normalized = String(value).trim();
+    return normalized.length > 0 ? normalized : null;
+}
+function toNumber(value) {
+    if (value === null || value === undefined) {
+        return null;
+    }
+    const numeric = Number(value);
+    return Number.isFinite(numeric) ? numeric : null;
+}
+function toBoolean(value) {
+    if (value === null || value === undefined) {
+        return null;
+    }
+    if (typeof value === "boolean") {
+        return value;
+    }
+    const normalized = String(value).trim().toLowerCase();
+    if (normalized === "true") {
+        return true;
+    }
+    if (normalized === "false") {
+        return false;
+    }
+    return null;
+}
+function normalizeCountryCode(value) {
+    return value === null ? null : value.trim().toUpperCase();
+}
+function isDachCountry(countryCode) {
+    return ["DE", "AT", "CH"].includes(countryCode ?? "");
+}
+function employeeCountFromBucket(bucket) {
+    switch ((bucket ?? "").trim()) {
+        case "200-499":
+            return 350;
+        case "501-1000":
+            return 750;
+        case "1001-5000":
+            return 2500;
+        case "5001-10.000":
+            return 7500;
+        case "10.000+":
+            return 10000;
+        default:
+            return 500;
+    }
+}
+export function buildDeelLeadPoolContactSql(options) {
+    const normalizedIds = options.contactIds
+        .map((contactId) => contactId.trim())
+        .filter((contactId) => contactId.length > 0);
+    if (normalizedIds.length === 0) {
+        throw new Error("At least one contactId is required to build the leadPool contact SQL.");
+    }
+    const contactList = normalizedIds.map((id) => `'${id.replaceAll("'", "\\'")}'`).join(", ");
+    const minScore = options.minEmailScore ?? 0;
+    return `SELECT
+  CAST(p.leadListId AS STRING) AS leadListId,
+  CAST(p.contactId AS STRING) AS contactId,
+  CAST(p.linkedin_contacts_companyId AS STRING) AS companyId,
+  CAST(p.firstName_cleaned AS STRING) AS firstName,
+  CAST(p.lastName_cleaned AS STRING) AS lastName,
+  CONCAT(
+    COALESCE(CAST(p.firstName_cleaned AS STRING), ''),
+    IF(p.firstName_cleaned IS NOT NULL AND p.lastName_cleaned IS NOT NULL, ' ', ''),
+    COALESCE(CAST(p.lastName_cleaned AS STRING), '')
+  ) AS fullName,
+  CAST(p.jobTitle AS STRING) AS jobTitle,
+  CAST(p.companyName AS STRING) AS companyName,
+  CAST(p.linkedin_companies_handle AS STRING) AS linkedin_companies_handle,
+  CONCAT('https://www.linkedin.com/company/', CAST(p.linkedin_contacts_companyId AS STRING)) AS companyUrl,
+  CAST(p.companySize AS STRING) AS companySize,
+  CAST(p.company_countryCode AS STRING) AS countryCode,
+  CAST(p.headquarters AS STRING) AS headquarters,
+  CAST(p.domain AS STRING) AS domain,
+  CAST(p.domain_linkedin AS STRING) AS domain_linkedin,
+  CAST(p.industry AS STRING) AS industry,
+  CAST(p.contact_location AS STRING) AS contact_location,
+  CAST(p.contact_companyLocation AS STRING) AS contact_companyLocation,
+  SAFE_CAST(p.tenureAtCompany AS FLOAT64) AS tenureAtCompany,
+  SAFE_CAST(p.tenureAtPosition AS FLOAT64) AS tenureAtPosition,
+  CAST(p.contact_summary AS STRING) AS contact_summary,
+  CAST(p.contact_titleDescription AS STRING) AS contact_titleDescription,
+  CAST(p.leadList_container_ts AS STRING) AS leadList_container_ts,
+  CAST(p.timestamp AS STRING) AS company_ts,
+  CAST(p.email AS STRING) AS email,
+  SAFE_CAST(p.emailScore AS INT64) AS emailScore,
+  SAFE_CAST(p.gender AS INT64) AS gender,
+  COALESCE(p.emails_accept_all, FALSE) AS emails_accept_all,
+  CAST(p.jobTitle AS STRING) AS jobTitleRaw,
+  CAST(p.domain AS STRING) AS domainRaw
+FROM \`icpidentifier.SalesGPT.leadPool_new\` p
+WHERE CAST(p.contactId AS STRING) IN (${contactList})
+  AND CAST(p.email AS STRING) IS NOT NULL
+  AND COALESCE(p.email_invalid, FALSE) = FALSE
+  AND COALESCE(p.email_bounced, FALSE) = FALSE
+  AND COALESCE(p.company_blacklisted, FALSE) = FALSE
+  AND COALESCE(p.company_inSequence, FALSE) = FALSE
+  AND COALESCE(p.contact_replied, FALSE) = FALSE
+  AND COALESCE(p.contact_bounced, FALSE) = FALSE
+  AND COALESCE(p.jobTitle_blacklisted, FALSE) = FALSE
+  AND COALESCE(p.domainBlacklisted, FALSE) = FALSE
+  AND CAST(p.domain AS STRING) IS NOT NULL
+  AND CAST(p.firstName_cleaned AS STRING) IS NOT NULL
+  AND CAST(p.lastName_cleaned AS STRING) IS NOT NULL
+  AND SAFE_CAST(p.emailScore AS INT64) >= ${Math.trunc(minScore)}
+  AND CAST(p.companySize AS STRING) IN ('200-499', '501-1000', '1001-5000', '5001-10.000', '10.000+')
+QUALIFY ROW_NUMBER() OVER (
+  PARTITION BY LOWER(CAST(p.email AS STRING))
+  ORDER BY SAFE_CAST(p.emailScore AS INT64) DESC NULLS LAST, p.timestamp DESC, p.contact_ts DESC
+) = 1
+ORDER BY RANDOM()`;
+}
+function buildGermanSalutation(row) {
+    const lastName = row.lastName?.trim() ?? "";
+    const firstName = row.firstName?.trim() ?? "";
+    if (row.gender === 1 && lastName.length > 0) {
+        return `Sehr geehrter Herr ${lastName},`;
+    }
+    if (row.gender === 2 && lastName.length > 0) {
+        return `Sehr geehrte Frau ${lastName},`;
+    }
+    if (firstName.length > 0 || lastName.length > 0) {
+        return `Guten Tag ${[firstName, lastName].filter((value) => value.length > 0).join(" ")},`;
+    }
+    return null;
+}
+function deriveLanguage(countryCode) {
+    return isDachCountry(countryCode) ? "de" : "en";
+}
+function deriveRegion(countryCode) {
+    if (isDachCountry(countryCode)) {
+        return "DACH";
+    }
+    return countryCode ?? "GLOBAL";
+}
+function computeLeadScore(row) {
+    const emailScore = row.emailScore ?? 70;
+    const segment = row.segment ?? "broader-hr";
+    const segmentBonus = segment === "leadership" ? 10 : segment === "payroll-people-services" ? 8 : 4;
+    const acceptAllPenalty = row.emails_accept_all ? 4 : 0;
+    const tenureBonus = (row.tenureAtCompany ?? 0) >= 1 ? 2 : 0;
+    return Math.max(60, Math.min(99, emailScore + segmentBonus + tenureBonus - acceptAllPenalty));
+}
+function gradeFromScore(score) {
+    if (score >= 90) {
+        return "A";
+    }
+    if (score >= 80) {
+        return "B";
+    }
+    if (score >= 70) {
+        return "C";
+    }
+    return "D";
+}
+export function normalizeDeelOutreachRows(rows) {
+    return rows.map((row) => {
+        const leadListQuery = toNullableString(row.leadListQuery);
+        const parsedQuery = leadListQuery === null ? null : parseHistoricalQuery(leadListQuery);
+        const countryCode = normalizeCountryCode(toNullableString(row.countryCode));
+        return {
+            leadListId: toNullableString(row.leadListId),
+            contactId: toNullableString(row.contactId),
+            companyId: toNullableString(row.companyId),
+            firstName: toNullableString(row.firstName),
+            lastName: toNullableString(row.lastName),
+            fullName: toNullableString(row.fullName),
+            jobTitle: toNullableString(row.jobTitle),
+            companyName: toNullableString(row.companyName),
+            linkedin_companies_handle: toNullableString(row.linkedin_companies_handle),
+            companyUrl: toNullableString(row.companyUrl),
+            companySize: toNullableString(row.companySize),
+            countryCode,
+            headquarters: toNullableString(row.headquarters),
+            domain: toNullableString(row.domain),
+            domain_linkedin: toNullableString(row.domain_linkedin),
+            industry: toNullableString(row.industry),
+            contact_location: toNullableString(row.contact_location),
+            contact_companyLocation: toNullableString(row.contact_companyLocation),
+            tenureAtCompany: toNumber(row.tenureAtCompany),
+            tenureAtPosition: toNumber(row.tenureAtPosition),
+            contact_summary: toNullableString(row.contact_summary),
+            contact_titleDescription: toNullableString(row.contact_titleDescription),
+            leadList_container_ts: toNullableString(row.leadList_container_ts),
+            company_ts: toNullableString(row.company_ts),
+            leadListQuery,
+            leadListQueryDecoded: parsedQuery?.decodedQuery ?? null,
+            leadListFunctionFilters: parsedQuery?.functionFilters.filter((filter) => filter.selection === "INCLUDED").map((filter) => filter.text) ?? [],
+            leadListTitleFilters: parsedQuery?.titleFilters.filter((filter) => filter.selection === "INCLUDED").map((filter) => filter.text) ?? [],
+            leadListRegionFilters: parsedQuery?.regionFilters.filter((filter) => filter.selection === "INCLUDED").map((filter) => filter.text) ?? [],
+            leadListHeadquartersFilters: parsedQuery?.headquartersFilters
+                .filter((filter) => filter.selection === "INCLUDED")
+                .map((filter) => filter.text) ?? [],
+            leadListHeadcountFilters: parsedQuery?.headcountFilters.filter((filter) => filter.selection === "INCLUDED").map((filter) => filter.text) ?? [],
+            email: toNullableString(row.email),
+            emailScore: toNumber(row.emailScore),
+            gender: toNumber(row.gender),
+            emails_accept_all: toBoolean(row.emails_accept_all),
+            language: deriveLanguage(countryCode),
+            marketSegment: isDachCountry(countryCode) ? "dach" : "non-dach"
+        };
+    });
+}
+function rowToScoredLead(row) {
+    const contactName = row.fullName?.trim() ||
+        [row.firstName, row.lastName].filter((value) => value !== null && value.length > 0).join(" ").trim();
+    const title = row.jobTitle?.trim() ?? "";
+    const { segment, reason } = classifyDirectPathSegment(title);
+    const score = computeLeadScore({ ...row, segment, segmentReason: reason });
+    return {
+        companyName: row.companyName ?? row.domain ?? "Unknown Company",
+        domain: row.domain ?? row.domain_linkedin ?? "unknown.example",
+        industry: row.industry ?? "Unknown",
+        region: deriveRegion(row.countryCode),
+        employeeCount: employeeCountFromBucket(row.companySize),
+        contactName: contactName.length > 0 ? contactName : row.email ?? "Unknown Contact",
+        title: title.length > 0 ? title : "Human Resources",
+        email: row.email ?? "unknown@example.com",
+        source: "salesprompter-deel-leadlists",
+        signals: [
+            `language:${row.language}`,
+            `market:${row.marketSegment}`,
+            `segment:${segment}`,
+            row.countryCode ? `country:${row.countryCode}` : null
+        ].filter((value) => value !== null),
+        techStack: [],
+        crmFit: segment === "broader-hr" ? "medium" : "high",
+        outreachFit: score >= 85 ? "high" : "medium",
+        buyingStage: segment === "payroll-people-services" ? "solution-aware" : "problem-aware",
+        notes: [
+            row.leadListId ? `leadListId:${row.leadListId}` : null,
+            row.leadListRegionFilters.length > 0 ? `regions:${row.leadListRegionFilters.join(", ")}` : null
+        ].filter((value) => value !== null),
+        score,
+        grade: gradeFromScore(score),
+        rationale: [
+            `segment ${segment}: ${reason}`,
+            row.emailScore !== null ? `email score ${row.emailScore}` : "email score unavailable",
+            row.language === "de" ? "DACH lead routed to German outreach" : "non-DACH lead routed to English outreach"
+        ],
+        customVariables: {
+            language: row.language,
+            salesprompter_language: row.language,
+            salesprompter_market_scope: row.marketSegment,
+            salesprompter_country_code: row.countryCode,
+            salesprompter_segment: segment,
+            salesprompter_segment_reason: reason,
+            salesprompter_email_score: row.emailScore,
+            salesprompter_emails_accept_all: row.emails_accept_all,
+            salesprompter_leadlist_id: row.leadListId,
+            salesprompter_leadlist_regions: row.leadListRegionFilters.join(", "),
+            salesprompter_leadlist_titles: row.leadListTitleFilters.join(", "),
+            salesprompter_leadlist_functions: row.leadListFunctionFilters.join(", "),
+            salesprompter_linkedin_company_handle: row.linkedin_companies_handle,
+            gender: row.gender,
+            anrede_zeile: row.language === "de" ? buildGermanSalutation(row) : null
+        }
+    };
+}
+function average(values) {
+    if (values.length === 0) {
+        return null;
+    }
+    const total = values.reduce((sum, value) => sum + value, 0);
+    return Number((total / values.length).toFixed(2));
+}
+export function buildDeelOutreachPack(market, rows) {
+    const locales = {
+        de: [],
+        en: []
+    };
+    const countries = new Map();
+    const titles = new Map();
+    const segmentCounts = {
+        leadership: 0,
+        "payroll-people-services": 0,
+        "broader-hr": 0
+    };
+    const emailScores = {
+        de: [],
+        en: []
+    };
+    for (const row of rows) {
+        const { segment } = classifyDirectPathSegment(row.jobTitle ?? "");
+        segmentCounts[segment] += 1;
+        const country = row.countryCode ?? "unknown";
+        countries.set(country, (countries.get(country) ?? 0) + 1);
+        const title = row.jobTitle ?? "unknown";
+        titles.set(title, (titles.get(title) ?? 0) + 1);
+        if (row.emailScore !== null) {
+            emailScores[row.language].push(row.emailScore);
+        }
+        locales[row.language].push(rowToScoredLead({ ...row, segment }));
+    }
+    return {
+        vendor: "deel",
+        market,
+        total: rows.length,
+        locales,
+        summary: {
+            localeCounts: {
+                de: locales.de.length,
+                en: locales.en.length
+            },
+            segmentCounts,
+            countries: [...countries.entries()].sort((a, b) => b[1] - a[1] || a[0].localeCompare(b[0])),
+            averageEmailScoreByLocale: {
+                de: average(emailScores.de),
+                en: average(emailScores.en)
+            },
+            topTitles: [...titles.entries()].sort((a, b) => b[1] - a[1]).slice(0, 20)
+        }
+    };
+}