Skip to main content

KB-015: GHL Prospecting Tool

Purpose: Document how the GHL Prospecting tool works, what the audit reports contain, and how SC uses prospecting to identify and qualify potential clients.

Who This Is For: All Symphony Core team members involved in sales or business development


Quick Answer​

GHL Prospecting is a built-in tool that generates audit reports on any business's online presence -- covering their website performance, SEO, Google Business Profile, online reputation, listings accuracy, tech stack, and AI search visibility (whether AI assistants can find, read and act on the business). SC has a Premium subscription ($29/month) that includes AI-powered lead discovery and unlimited manual reports. Use it to find businesses with gaps in their digital presence, then reach out with specific findings to start a sales conversation.


What is GHL Prospecting?​

GHL Prospecting is a lead generation and business audit tool built into GoHighLevel. It analyzes a business's public online footprint and produces a structured report highlighting strengths and weaknesses.

Free vs. Premium​

FeatureFreePremium ($29/month)
Manual prospect reportsLimitedUnlimited
AI Lead DiscoveryNoYes -- auto-discovers businesses matching your criteria
Email outreach via Lead ConnectorNoYes
Prospect CRM pipelineBasicFull pipeline with automation triggers
Report depthSummary onlyFull audit (seven visible sections on SC's reports)

SC has Premium. The subscription is active on the primary agency account.

Two workspaces, two separate prospect sets​

This is the single most misleading thing about the tool, and it cost a day of investigation on 2026-09-08. Prospecting appears in two places, each with its own prospects:

WorkspacePathWhat it holds
Sub-account — this is where the motion runsMarketing > Prospecting (/v2/location/<locationId>/marketing/prospecting)The Prospect AI agents and the prospects that feed the Sales Pipeline
AgencyProspecting, in the agency left navA separate prospect set. Reading it tells you nothing about the motion

Looking at the agency workspace and reporting its prospects as the motion's supply is the specific mistake to avoid. Always confirm the URL contains /location/<id>/marketing/prospecting before drawing conclusions.

Subscription topology. The agency Prospecting Premium subscription includes one linked sub-account. Under Prospecting > Settings > Manage subscription it reads "You can link only one sub-account, and this can only be done once." That link is already spent on Symphony Core LLC — it is one-time and irreversible, so it is not a decision to re-open. Cancelling the agency subscription also cancels the linked sub-account's.


What the Audit Report Contains​

Eight scored sections plus an overall score. Listings is hidden on SC's reports, so seven are visible and averaged, rendered across about 12 pages. Not all sections apply to every business, and a section that was switched off on the agent simply does not appear.

Corrected 2026-09-23 against a real report. The names below are what the report actually prints, verified page by page against Symphony Core's own audit, filed in Drive at sc-docs-storage/03-sales/enablement/2026-09-ref-symphony-core-own-marketing-audit-report.pdf (the tool's own filename was report_2026-09-23_en-US.pdf). Earlier versions of this table carried three names the tool has never emitted, and a blog post was drafted from them before the error was caught.

Section as printedListed on the summary page asWhat It MeasuresCommon Issues Found
Business Critical InformationBusiness DetailsFour fixed checks, identical for every business: text-enabled phone number, website hosting, chat widget, review repliesNumber not SMS-capable; no chat widget
Techno Stack AnalysisTechno StackAnalytics, tag managers, ad pixels, conversion trackingDetection fails on JS-rendered sites and behind tag managers. A missing mark means not detected, not not present
Google Business ProfileGoogle Business ProfileAddress, photos, website link, operational hours, phone numberMissing GBP entirely; incomplete profile elements
Listings (+ Detailed Listings)Listings31 named directories, each graded Match / Partial Match / No Match on name, address and phoneMissing from most directories; stale address after a move
Analyzing Your Reputation Game (+ Positive Reviews)ReputationReview count, rating, reply percentage, average response time, per-platform totalsA platform with no presence is scored as a zero, not as not-applicable. See the score warning below
Website Performance (+ Website Health)Website PerformanceMobile and desktop load times, Core Web Vitals, a pass/fix/consider count per deviceScores move between runs; treat as directional
AI search visibilityAI search visibilityA readiness score out of 100 ("Ready" from 70) across four areas: local presence (NAP consistency across GBP, directories and reviews), crawler access (robots.txt for GPTBot, ClaudeBot, PerplexityBot, Google-Extended), AI readability (JS-dependent content, LocalBusiness structured data field by field, preview image, llms.txt) and agent readiness (can an assistant find the phone, call, send an inquiry or book). It does not query any AI assistantStructured data missing most fields; content behind JavaScript; no labeled inquiry form or booking button on the page tested. Agent readiness tests one page only. See KB-017
Local SEO Heatmaps (+ Local Rank Tracker)Local SEO HeatmapsA grid of map rankings for one search term, then a ranked competitor table with ratings and review countsVoid if the audit matched the wrong business or address

⚠️ AI search visibility is a Labs feature. It appeared on SC's own report by 2026-11-11; the Labs card read "Live in 20 days" that day, so general release is about 2026-12-01. Not yet verified: whether reports from the /audit widget and the sub-account Prospecting workspace include it (launch backlog item AV2-01).

Four rendered sections never appear on the summary page: Detailed Listings, Positive Reviews, Website Health and Local Rank Tracker. They are continuations of the section above them and carry that section's score.

⚠️ Local Rank Tracker is dropped from the SC-authored report. The Marketing Audit Production PRD §7 removes it as low-value and actively misleading on a wrong business match. It is in the widget output, and the correction worksheet carries a row for it, but it does not carry through.

Reading the Scores​

Every score is a percentage. There are no letter grades; earlier versions of this page said there might be, and that was wrong. The report prints its own five bands on the summary page:

BandLabel
0-20%Worse
21-40%Bad
41-60%Average
61-80%Good
81-100%Excellent

⚠️ The overall score is the plain average of the section scores, despite the report describing it as "weighted". Verified on two runs of SC's own audit: six sections averaged 75.8 and printed 76%; adding Listings at 5% made seven sections averaging 65.7, and it printed 66%.

The consequence is that enabling an audit changes the overall score without anything about the business changing. The Listings audit was off on the agent until 2026-09-09, so any report generated before that date has no Listings score and a correspondingly inflated overall. Since 2026-09-29, Listings has been hidden again through report customization; the agent still runs the audit, but the section no longer counts toward the average. Since 2026-11-11 the AI search visibility section is also in the average, so SC's reports average seven sections again. Never compare an overall score across reports without checking both show the same sections.

How each section score is calculated, and which formulas actually reproduce, is in KB-017: Marketing Audit Scoring Logic.

⚠️ A low section score is not always a finding. SC's Reputation section scores 37% ("Bad") on 8 reviews, all positive, 5.0 average, 100% replied. The drag is zero Facebook reviews, counted as a zero rather than as not-applicable. Read the underlying metrics before leading with a section percentage in outreach.


How Prospect AI Works​

With Premium, you configure one or more Prospect AI agents that automatically find businesses matching set criteria. Each agent is configured in three steps: agent details (niche, location, allocation, schedule), audit preferences, and lead enrichment.

  1. Set target parameters: Niche, city and postal code, search radius, prospect filters
  2. The agent scans public data: Sourced from Google Maps data
  3. Prospects appear in the queue: Each discovered business gets an audit report
  4. Review and approve: You decide which prospects to pursue

Agents run on a schedule, weekly or monthly, with the monthly allocation split across the selected weeks.

What the agent returns​

Verified on the prospect cards in the Symphony Core sub-account, 2026-08-18:

FieldReliably present
Business nameYes
Business addressYes
Business phoneYes
Website URLYes, where one exists
GBP ratingYes
NicheYes
AI conversion score, plus a prospect scoreYes
Contact person's emailNo. Frequently absent

Measured 2026-09-08: 24 of 46 prospects (52%) had no email address at all. v1.1 recorded this qualitatively as "frequently absent"; it is now a number, taken across the full Prospecting Tool contact set in the SC sub-account.

This was the original reason the motion was call-first — the one field a cold email cadence depends on is the one the agent is least likely to supply, while the phone number is supplied reliably. The motion switched to email-first in cadence SOP v2.3 (2026-08-27) as an operator-fit decision, which means roughly half of every batch needs manual email discovery per the Prospect Email Discovery SOP before it can enter the cadence. Plan batch sizes against the sendable half, not the raw count. See the GHL Prospecting Outreach Cadence SOP.

Limits worth knowing​

LimitValue
Basic prospectingUp to 50 leads per month
Premium prospectingUp to 100 leads per month
Scope of the limitPer account, not per agent. Adding agents does not raise the ceiling
EnrichmentCapped at 3 contact points per prospect
API, webhooks, exportNone documented. The agent cannot be driven or read programmatically. Report generation is separately available as a workflow action - see the automation boundary below

Because the source is Google Maps data, narrow targeting criteria will exhaust a geography after repeated runs. If an agent starts returning nothing, widen the radius or loosen the filters before assuming a fault.

How allocation actually works​

Each agent has a monthly prospect allocation set on a slider in step 1. That allocation is drawn against the per-account cap, not a per-agent one. The agent edit screen shows the arithmetic directly:

Monthly Prospect AI leads limit (max): 100 | Used by other agents: 0 | Current agent: 25 | Available: 75

Consequences:

  • Adding agents never raises the ceiling. Extra agents split the same 100. They buy diversification of niche or geography, not volume.
  • Unallocated capacity is simply unused. SC pays for Premium either way, so an agent allocated 25 of 100 leaves 75 paid-for leads on the table each month.
  • The monthly allocation is split across the scheduled run weeks, so a weekly agent at 25/month yields roughly 6 per run.

Hard per-agent limits (verified 2026-09-09)​

Two caps shape how you can respond to exhaustion. Neither is documented by the vendor, and both were found by hitting them:

LimitBehaviour
Maximum 3 built-in niches per agentClicking a fourth is rejected with a toast: "You can select up to 3 niches." The selection simply does not change
One City / Area / Postal per agentOne anchor locality per agent. A search radius (5 / 10 / 15 / 25 mi / Entire city) widens coverage around it — this control sits below the fold and is easy to miss

The "Include additional niches" picker is a large catalogue of other business types (accountant, retail, and so on), but it does not contain the nine built-in trades — searching it for plumber returns "No Data", because Plumber is a built-in chip.

Consequence: broader niche coverage requires more agents, and more agents split the same 100-lead account cap. Broader geographic coverage does not — raise the search radius first.

Do not reach for geography by reflex. The full option set includes a Prospect filters block (website presence, GBP rating, conversion score) that combines as AND and shrinks the pool far harder than any locality choice. A zero-yield run with several filters active says nothing about geography. The complete catalogue, the filter-stack trap, and the correct order of suspicion are in the Prospect AI Agent Configuration Reference.

No new prospects: triage in order​

Run these in sequence. Do not skip to a theory — on 2026-09-08 two plausible explanations (a billing banner, then a missed human click) were both wrong, and the authoritative reading took two clicks.

  1. Open the agent. Marketing > Prospecting > Prospect AI > the agent. Read two fields: Last run, and the header counter Total leads generated this month: n/100.
  2. Did it run and return zero? The pool is empty for the current criteria. Check the prospect filters first — Custom finding strength unlocks website / GBP-rating / conversion-score filters, which combine as AND and are the most common cause — then the search radius, then the niches. This is the expected end-state of over-narrow criteria, not a bug.
  3. Did it not run at all? Check the schedule and that the agent is Active.
  4. Is the counter at the cap? Then allocation is spent for the month; raise the allocation or wait.
  5. Only now look downstream — pipeline counts and anything else. Compare Prospects found with Added to pipeline on the agent card: if those two are close, nothing downstream is stuck. Note that pipeline entry may be automatic (step 3, "Also add to a pipeline stage"), in which case there is no human step to blame at all — check the agent before assuming one.

The payment-method banner is unrelated. It is the SaaS reselling prompt and has no bearing on prospect generation — see agency-owned-subaccounts-carry-no-reselling.

The automation boundary — read before promising anything scripted​

The Prospecting module cannot be driven programmatically or by a browser agent. Verified repeatedly on 2026-09-08 across two machines:

  • No API, webhook or export. The agent cannot be configured, triggered or read through the GHL API.
  • The module's own report link must be copied from the UI by hand. There is no call that returns it.
  • It renders in an iframe with no accessibility tree. Element-finding returns nothing, tab clicks do not register, and screenshots intermittently time out with a frozen renderer.

Every step inside the module — creating an agent, editing it, refreshing an audit, adding a prospect to the pipeline, harvesting that module's report link — is a human click. Do not plan automation through this module, and do not promise a scripted batch pull.

This boundary is the module's, not the report's. A workflow action, Generate marketing audit report, produces an audit report from contact fields and returns its URL to later nodes in the same run — so a report link can be obtained without a human, just not from this module. The two are different artifacts made by different mechanisms, and the distinction decides whether a given piece of work is automatable. See the C-021 Prospecting Workflow Spec.

Best practice: Review AI-discovered leads weekly. Not every discovered business is a good fit -- filter for businesses that match SC's ideal client profile (local service businesses, professional services, small businesses with 1-50 employees).


How to Create a Manual Prospect Report​

Manual reports are useful when you already know a specific business you want to target.

Steps​

  1. Navigate to the sub-account where you want to create the prospect (use Symphony Core LLC sub-account for SC prospecting)
  2. Go to: Marketing > Prospecting (left sidebar). Not Contacts > Prospecting — that path was wrong in v1.2 and earlier
  3. Click: "Add Prospect" or "New Report"
  4. Enter the business details:
    • Business name (required)
    • Website URL (required for website/SEO sections)
    • City and state (required for GBP/listings sections)
    • Phone number (optional but improves accuracy)
  5. Click: "Generate Report"
  6. Wait: Report generation takes 30-90 seconds
  7. Review the report: Check each section for accuracy before using it for outreach

Important Notes​

  • Reports are generated at the sub-account level, not the agency level
  • Each report is tied to a prospect contact record in that sub-account
  • You can generate reports for businesses even if they are not yet in your CRM

How to Refresh a Report​

Audit data goes stale over time. Website changes, new reviews come in, and listings get updated.

When to Refresh​

  • Before first outreach: Always refresh if the report is more than 7 days old
  • Before a follow-up: Refresh if more than 14 days have passed since the last report
  • After a significant event: If the prospect mentions they made changes, refresh to see updated data

How to Refresh​

  1. Open the existing prospect record
  2. Click "Refresh Report" or "Re-run Audit"
  3. Wait for the new report to generate
  4. Compare with the previous report to note improvements or declines

Adding a Prospect to the CRM Pipeline​

There are two paths, and confusing them has already caused a misdiagnosis.

Path A — automatic (what SC uses). If the agent has step 3's "Also add to a pipeline stage" enabled with a pipeline and stage selected, every prospect the agent finds is added with no human involvement. SC's agent Noah is configured this way, pointed at Sales Pipeline / New Lead. On 2026-09-08 a feed stall was attributed to "nobody clicking Add to Pipeline"; with auto-add configured, nobody ever needed to, and the agent's own found 40 / added 38 counters said so. Check the agent's step 3 before blaming a human step.

Path B — manual. From the Prospecting tab, click "Add to Pipeline" on an individual prospect record. Use this for prospects you want to work selectively, or when auto-add is off.

Either path produces the same result:

  • Source field: automatically set to "Prospecting" — do not change this (used for ROI tracking)
  • Initial stage: New Lead in the Sales Pipeline. Outreach state is then carried by tags, per the Outreach Cadence SOP

What Gets Created​

  • A Contact record (if one doesn't already exist) with the business details from the audit
  • An Opportunity linked to that contact in the Sales Pipeline at New Lead
  • The audit report is accessible from the contact record

Sending Email from the Prospecting Tool​

GHL Prospecting supports sending outreach emails directly from the tool. These emails are sent via Lead Connector (GHL's built-in email infrastructure), not through Google Workspace.

Why Lead Connector, Not Google Workspace​

FactorLead ConnectorGoogle Workspace
Daily send limitsHigher (hundreds/day)Low (Gmail: ~500/day, shared across all uses)
Deliverability trackingBuilt-in open/click trackingRequires external tools
Reply handlingReplies land in GHL conversation tabReplies go to Gmail inbox
CRM integrationAutomatic -- logged to contact recordManual or requires Zapier
Risk to primary domainSeparate sending infrastructureShares reputation with business email

Rule: All prospecting outreach must go through Lead Connector. Never use Google Workspace (rohit@symphonycore.com or team@) for cold prospecting email.

Sending Process​

  1. Open the prospect record
  2. Click "Send Email"
  3. Select or compose your email (see Prospecting Outreach Email Templates)
  4. Email sends via Lead Connector
  5. Activity is logged on the contact record
  6. Replies appear in the GHL Conversations tab for that sub-account

Workflow Automation Possibilities​

GHL workflows can be triggered by prospecting events. These are aspirational -- implement as volume justifies:

TriggerPossible Automation
Prospect added to pipelineAuto-assign to team member; send Slack notification
Email sentStart follow-up timer (3-day delay for Follow-Up 1)
Email openedLog engagement; adjust follow-up timing
Reply receivedNotify team via Slack; move to "Responded" stage
No reply after cadenceMove to "Not Ready Now" stage; schedule 60-day re-engagement
Opportunity moved to "Qualified"Create new Opportunity in Sales Pipeline Stage 1

For the outreach cadence, see the GHL Prospecting Outreach Cadence SOP. Any automation of these triggers is gated behind the Prospecting Outreach Automation PRD.


Glossary​

TermDefinition
Audit ReportThe multi-section analysis GHL generates for a prospect's online presence
Lead ConnectorGHL's built-in email/SMS sending infrastructure (separate from Google Workspace)
NAPName, Address, Phone -- the core business identity fields checked for consistency
ProspectA business that has been audited but not yet contacted or qualified
Prospect AIPremium feature. Scheduled agents that automatically find and audit businesses matching set criteria, sourced from Google Maps data
AI conversion scoreThe agent's own estimate of how likely a prospect is to convert, shown on the prospect card
Core Web VitalsGoogle's page experience metrics (LCP, FID, CLS) measured in the Website Performance section
Review VelocityThe rate at which a business receives new reviews over time
Schema MarkupStructured data on a website that helps search engines understand business information


Document History​

VersionDateChanges
1.72026-11-12Adds the AI search visibility section (Labs, seen on SC's report 2026-11-11): its four areas, that it does not query any AI assistant, and that it now counts in the overall. Section count updated in the Quick Answer, Free vs Premium and the report contents. Records that /audit widget and sub-account coverage is unverified. Document Information footer brought up to date
1.62026-09-30§Reading the Scores notes that Listings has been hidden through report customization since 2026-09-29, and links to the new KB-017 for per-section scoring formulas
1.52026-09-23§What the Audit Report Contains corrected against a real report, page by page. Three section names in the old table were never emitted by the tool: Business Details, Technology Stack and SEO Analysis are actually Business Critical Information, Techno Stack Analysis and Local SEO Heatmaps. Four rendered sections were missing entirely. Struck the claim that scores may be letter grades, since every observed score is a percentage, and replaced the uncorroborated 80/60 thresholds with the report's own five bands. Records that the overall score is a plain average rather than the weighted composite it claims, so switching an audit on moves it without the business changing: turning Listings on cost SC 10 points
1.42026-09-17Scoped the automation boundary to the Prospecting module itself. The previous wording generalised it into a claim that report URLs have no API surface at all, which the Generate marketing audit report workflow action falsifies: it builds a report from contact fields and returns its link to later nodes in the same run. The module's own limits are unchanged - it is still an iframe with no accessibility tree and no API.
1.32026-09-09Added the hard per-agent limits (max 3 niches; one anchor locality plus a search radius) and their consequence: widening coverage requires more agents, which split the same 100-lead cap. Two Prospecting workspaces documented (agency vs sub-account, separate prospect sets) after a day was lost reading the wrong one; nav path corrected to Marketing > Prospecting (v1.2's "Contacts > Prospecting" was wrong). Added the subscription topology (one-time irreversible sub-account link, already spent), how allocation draws against the per-account cap, the "No new prospects: triage in order" runbook, and the automation boundary (iframe, no accessibility tree, no API - every step is a human click). Upgraded the missing-email warning from qualitative to measured: 52% of prospects have no email, and the motion is now email-first, so half of each batch needs manual discovery. Added a cross-link to the new Prospect AI Agent Configuration Reference, which catalogues every option and flags the prospect-filter stack as the highest-risk setting.
1.22026-08-22Removed the stale "Prospecting Pipeline" glossary entry and the two remaining references to it (the four-stage pipeline was never built); opportunities land in the Sales Pipeline at New Lead
1.12026-08-18Documented the Prospect AI agent model, the fields the agent actually returns (phone reliably present, contact email frequently absent), account-level lead limits, and the absence of any API or webhook. Repointed the pipeline entry and cadence references
1.02026-03-02Initial version documenting GHL Prospecting tool, audit report sections, Lead Connector email, and pipeline integration

Document Information

  • Version: 1.7
  • Created: 2026-03-02
  • Updated: 2026-11-12
  • Location: 06-team-training/platform-training/kb-015-ghl-prospecting-tool.md