HubSpot audits
Know where the portal
needs attention.
Unpack workflow health, trace dependencies and ask an AI advisor about the evidence. Add wider Portal Audit, a branded client report and CLI/MCP access to portal findings. Know what needs work before you build.
Client portal / Northline
A clearer picture
of the portal.
Portal score
A starting point for the conversation.
- 01 Review first
Workflow health
Review inactive automations
Inactive workflows remain in the portal.
Confirm which still belong in the portal.
- 02 Discuss
Sales setup
Review stage probabilities
Pipeline stage probabilities need review.
Confirm that each stage reflects the sales process.
Before the implementation brief
A shared view of
what needs work.
An inherited portal rarely comes with the full story. Portal Audit brings the setup, findings and recommendations into one reviewable assessment, so you can separate a useful next step from a guess.
- Portal context
- Review the areas available to the audit, from CRM setup and sales processes to marketing, automation, service and reporting.
- Scored findings
- Move between the portal-level picture and the underlying sections. Use the evidence and recommendations to frame a discussion, not to replace one.
- Your presentation
- Bring your agency’s branding to the report. Review the narrative, share the report and export a PDF for the client conversation.
- Visible limits
- Read coverage and unavailable data alongside the findings. A partial view is not proof that a portal has no issues.
AI HubSpot workflow audit
What does an AI HubSpot workflow audit check?
Workflow Audit reads every workflow in the portal, inactive ones included by default, and checks each one’s configuration, its conflicts with other workflows and the dependencies between them. Each workflow gets a 0–100 health score from rules; AI explains the evidence but doesn’t set the score. The audit only reads your portal, so nothing changes until you decide what to fix.
- 01 / What it checks
Configuration, conflicts and dependencies.
For each workflow: naming, enrollment triggers, branching, suppression, re-enrollment, goals and actions. Between workflows: competing property writes, overlapping enrollment, circular trigger chains, re-enrollment without a goal or suppression, contradictory branches, missing assets and seven-day email exposure, each rated critical, warning, info or suggestion. It also traces workflows that enroll into each other or share lists, properties and pipeline stages.
- 02 / How scoring works
A rule-based score you can unpack.
Configuration, conflict severity, complexity and available activity add up to 100 × earned points ÷ applicable maximum. Activity that isn’t available is left out rather than scored as zero. Open any workflow to see what moved its score and which conflicts involve it.
- 03 / What you do next
Work through it by severity.
Resolve critical conflicts first, then high ones, then the lowest-scoring workflows. Check the dependency paths before you touch a connected workflow. To build the fix, Projects (currently in limited release) creates a corrected replacement, switched off, for you to review. Then re-run the audit and compare.
Inside Workflow Audit
Don’t just count workflows.
Investigate the impact.
Inherited automation needs more than a map. Jetstack combines explainable health, typed dependency paths, evidence-linked findings and saved review decisions, with a read-only AI advisor to investigate the snapshot.
- A score you can unpack
- Configuration, conflict severity, complexity and available activity contribute to workflow health. The score is 100 × earned points ÷ applicable maximum, rounded to a whole number. Unavailable activity is excluded, not treated as zero. Portal Audit uses its own template-weighted scoring model.
- Typed relationships, not just lines
- Trace explicit enrollment, potential property handoffs, shared writers and shared usage. Inspect upstream/downstream path witnesses and impact coverage before deciding what a change might affect.
- An advisor grounded in the audit
- Ask about workflows, findings, connections, scores, history, reviews and coverage. The AI advisor retrieves from the completed snapshot; it does not make live HubSpot changes or observe live record execution.
- Evidence that stays in context
- Review findings, save decisions and compare compatible audit versions. Missing data stays visible. Potential conflicts and configuration paths are signals to investigate, not proof that the same records actually collided at runtime.
Your evidence. Your AI workspace.
Take portal findings straight into Claude.
Run an eligible portal audit from your terminal or MCP-compatible assistant, then retrieve its report data. No manual PDF export needed for that data flow.
- 01 / Portal Audit
Run. Check. Retrieve.
Native MCP tools
run_audit,get_audit_statusandget_audit_dataexpose the supported portal-audit workflow. The CLI provides the corresponding run, status and data commands. - 02 / Existing report
Bring the evidence to the conversation.
With authorized access and a known run ID, retrieve portal report data, scores, recommendations and any stored narrative. Ask a compatible assistant to analyze that evidence; verify its conclusions before acting.
- 03 / Workflow Audit
Investigate through the right surface.
Standalone workflow audits have REST results, evidence, dependency and path endpoints. Their read-only advisor is in the app; dedicated workflow-audit CLI/MCP wrappers are not currently supplied.
Connections, permissions, plan/template access and available data constrain coverage. CLI/MCP creation uses the public-API path; extension-only capture still needs the app and Chrome extension. An audit or an assistant answer does not authorize implementation.
From findings to a brief
The next step is a decision, not an automatic fix.
Use the assessment to agree the right work. Keep diagnosis and implementation deliberately separate.
- 01 / Diagnose
Read the evidence.
Inspect the score in context. Review the findings, recommendations and coverage before drawing conclusions about the portal.
- 02 / Choose
Agree what matters.
Discuss the business process behind the finding. Choose a scope with your client, including what should stay as it is.
- 03 / Hand off
Write the implementation brief.
Take the agreed scope into Projects or your team’s delivery process. A report does not authorize changes to HubSpot.
Observe first
A diagnostic.
Not a cleanup script.
The audit reads the portal; it does not delete records, rewrite workflows or deploy fixes. Its job is to make the current setup easier to understand.
Start from 40+ ready-made templates, from the Full Portal Assessment to a five-minute pulse check, or build your own. HubSpot subscriptions, permissions and browser-only areas affect the evidence available, and the report shows its coverage so you know exactly what it measured.
The Jetstack advantage
Workflow depth. A bigger delivery platform.
Investigate workflow health, paths and evidence with a read-only AI advisor. Add broader Portal Audit, CLI/MCP report access and Projects for the next brief.
Jetstack vs Howly
Howly maps your workflows and stops there. Jetstack audits the whole portal, explains every score and builds the fix.
Jetstack vs Entflow
Entflow grades workflows and writes a report. Jetstack explains the grade, audits the whole portal and builds the fix.
Jetstack vs Portal-iQ
Portal-iQ sends a PDF. Jetstack gives you live evidence, an AI advisor and the platform to build what the audit finds.
Questions, answered
Before you run an audit.
Does Portal Audit change anything in HubSpot?
No. Portal Audit is a read-only diagnostic. It collects evidence and produces findings, scores and a report. Deciding what to change, and carrying out that work, is a separate step.
How are Portal Audit and Workflow Audit scored?
Portal Audit uses configured data-point rules and weights. Standalone Workflow Audit uses applicable configuration, conflict, complexity and activity factors, normalized to a 0–100 health score. Missing activity metrics are not scored as zero. AI helps explain evidence; it does not invent either score. Coverage and access still determine what can be assessed.
How much does a HubSpot audit cost with Jetstack AI?
Portal Audit is part of the Jetstack AI platform, published at $1,250 per month. The same rate covers Projects and Portal reuse. Or $1,000 a month billed annually, with no per-seat or per-portal fees; the pricing page has the details.
How long does a HubSpot audit take?
A Portal Audit or Workflow Audit run usually takes minutes. The time depends on the size of the portal, how many workflows it has and whether AI insights are generated. Reviewing the findings with the people who own each process takes longer than the run, so plan time for that conversation.
Who can audit my HubSpot portal and fix workflows?
You, your agency or a HubSpot partner can run the audit with Jetstack AI: it reads the portal and shows which workflows conflict, what they depend on and how each one scores. Fixing is a separate, approved step. Projects, currently in limited release, builds a corrected workflow as a new asset, switched off, and you retire the old one in HubSpot.
Can I share the report under my agency’s brand?
You can configure report branding and share the report or export a branded PDF. Review the findings and narrative before sharing them with a client.
Will every HubSpot portal have the same coverage?
No. Coverage depends on the audit template, the portal’s HubSpot subscriptions, permissions and available data. Some areas need the Chrome extension. Review the run’s coverage and any unavailable data alongside the score.
Can I ask an AI advisor about the workflow findings?
Yes. The Workflow Audit advisor retrieves evidence from a completed audit snapshot: workflows, findings, relationships, path witnesses, scores, coverage and review decisions. It is read-only, not a live HubSpot execution monitor or an automatic repair agent. Inspect the evidence and business context before acting.
Can I use Claude, CLI or MCP with audits?
Portal audits have native CLI/MCP tools to run an eligible public-API audit, check status and retrieve report data. Existing data needs authorized access and a known audit run ID; extension-only capture still needs the app/extension. Standalone workflow audits have REST results, evidence and dependency/path investigation, but no dedicated CLI/MCP wrappers. Their advisor is in-app.
Does a finding automatically become a fix in Projects?
No. Review the evidence with your client and choose the work you want to take forward. Projects can be the implementation workspace for that agreed brief, with its own review and approval steps.
Elsewhere in the platform