Your hiring has outgrown a shared mailbox, but a separate applicant tracking system means a second login, a second database and a sync that never quite matches your CRM. ATS puts hiring inside the HubSpot portal you already run. Candidates apply on your own website. Every application lands as a CRM record. Recruiters work from cards on the records they already open every day, and every candidate gets a score against the role that shows exactly how it was reached.

The decision underneath it: a candidate is a Contact. Not a recruiting record that syncs to one — the Contact itself, with recruiting fields on it. So the candidate who later becomes a customer, a partner or a referral is one record with one history, and every list, workflow, report and email you already use works on hiring too.

KindPublic HubSpot app, installed on your portal
Built forTeams whose hiring has outgrown a mailbox but does not justify an enterprise ATS
For candidatesA careers site on your own domain, a page per role, an application form with CV upload, a status link that needs no password
For recruiters5 cards on HubSpot records, a full-screen recruiting app page, 6 workflow actions
Résumé readingReads a PDF or Word CV into the candidate's fields, with your own AI key
Matching10 weighted dimensions, every rule a setting, a breakdown behind every score
Stores data inYour own portal — custom objects and Contact properties. No outside database holds candidate data
How to get itNot distributed. Ask for a walkthrough

Important

Where it stands, 2026-11-16. Every surface on this page runs on our demo portal, and every screenshot here was taken there today from an invented set of roles and candidates — including one real application through the careers page, whose CV the app read with Claude. It runs on one portal; the next step is the same on your own roles and candidates. Reading résumés, placing addresses on a map and measuring commutes each use a provider key of your own, and each is off until you add one.

Why not a HubSpot form, or a separate ATS#

A HubSpot form on a careers pageA standalone ATSATS
Where candidates liveForm submissions on a ContactA second database, synced to the CRM (or not)The Contact itself
A pipeline per roleBuild it yourself from properties and listsYesYes — 12 stages, moved from a card, a page or a workflow
Reading the CVNoUsuallyYes, into the fields that get scored
Ranking candidates against a roleNoOften a score you cannot inspect10 dimensions, each with its own sub-score, every rule a setting you can change
AutomationHubSpot workflowsThe vendor's own rules engineHubSpot workflows, with 5 ATS actions
ReportingHubSpot reports on form dataThe vendor's reportsHubSpot reports on real records, plus the app's own dashboard
LoginsOneTwoOne

When it fits#

  • You already run your customer relationships in HubSpot and want candidates in the same place, not in a silo next to it.
  • You hire steadily but not at enterprise volume — a handful of open roles at a time, a recruiter or a hiring manager doing the screening.
  • You want to see why a candidate ranks where they do, and change the rule when the ranking is wrong for your bar.
  • You want hiring automation without another tool — "when an application scores above 80, notify the hiring manager" is a HubSpot workflow.

It is not the right product for regulated, high-volume hiring with formal compliance reporting, or if candidates must hold their own accounts. An enterprise ATS with a HubSpot connector is the better fit there.

Before you start: three decisions#

  1. Which tier. The app installs 4 custom objects by default (postings, applications, interviews, documents). That fits a standard portal. 6 and 19 objects are available when you need candidate profiles, offers or compliance fields, and upgrading never deletes anything. See ats-setup.md § The tier decision.
  2. Which provider keys. Résumé reading needs an AI key (Claude or another provider), turning addresses into map coordinates needs an OpenCage key, and travel times need a Google key. Each is optional, each is billed by its provider to you, and each states what it will spend before it spends it. A daily cap you choose limits résumé reading.
  3. Where the careers pages go. The careers site is three pages on your HubSpot website — listing, role, application — that use your theme. Plan the URL (/careers is usual).

How it works#

1. Candidates find the role on your own site#

The careers listing on a company website, headed "Open Positions". Below a search box ("Search roles, locations…") and two filters, "All departments" and "Any workplace", three roles are listed as cards: Field Sales Representative (Sales, Cork, Ireland, On-site, Full-time, EUR 45,000 – EUR 60,000), Customer Success Manager (Customer Success, Remote (Ireland & UK), Remote, Full-time, EUR 50,000 – EUR 65,000) and Senior Product Manager (Product, Dublin, Ireland, Hybrid, Full-time, EUR 85,000 – EUR 105,000).

The careers listing, in the website's own theme. Look at the filters and the salary range on each card — both come straight from the job posting record.

The listing searches and filters live postings by department, location, work model and employment type, and a filtered view is a link you can share ("our open Engineering roles"). Each role has its own page on a permanent address, indexable by search engines with the structured data Google Jobs reads, and a job-board feed lets aggregators pick up every role from one URL.

The page for one role, "Senior Product Manager". Under the title: tags Product, Dublin, Ireland, Hybrid and Full-time; the salary range EUR 85,000 – EUR 105,000; share buttons for LinkedIn, X and Email; an "Apply for this role" button. Below, "About the role" with a short description, then "Responsibilities" and "Requirements" as bullet lists.

One role's page. A salary you choose to hide is simply absent — here, and in the job-board feed.

2. They apply in one step, with their CV#

The application form "Apply: Senior Product Manager", filled in for Niamh O'Callaghan: first and last name, email niamh.ocallaghan@example.com, phone, a LinkedIn URL and a two-line cover letter. Under "Resume (optional, PDF or Word, up to 10 MB)" the attached file reads "cv-niamh-ocallaghan.pdf (90.3 KB) — click to change", and the consent box "I consent to my data being processed for recruitment purposes." is ticked.

The application form, just before it is sent. Every field is checked as the candidate types; the CV is optional.

Submitting creates or updates the candidate's Contact, creates the application linked to the role, and stores the CV as a document on the record. A second application from the same person to the same role is recognised rather than doubled. The candidate lands on a thank-you page with a reference, and can later ask for a status link by email — no password, no account — that shows a simple status (Received, In review, Interviewing, Offer, Closed), never your internal stage names or a rejection reason.

3. The app reads the CV into the fields that get scored#

The Document Inspector card for cv-niamh-ocallaghan.pdf, tagged Resume and Parsed, application/pdf, 90.3 KB, uploaded 16/11/2026, with Open file and Re-parse. Under "What was read from this document": "17 of 22 fields found · Claude claude-sonnet-5". First name, last name, email and phone are marked "Not written — already set on the contact." City Dublin, Country Ireland, a skills list (Product strategy, Roadmapping, Product discovery, A/B testing, SQL, Amplitude, Jira, Figma, Agile, Scrum, OKRs, Pricing, Stakeholder management, Certified Scrum Product Owner), Languages English, Irish, French, Seniority, Education level and Years of experience 9 are each marked "Written".

The CV from the application above, read seconds after it arrived. Look at the left column: what the candidate typed in the form is kept; what the CV adds is written.

Reading happens after the candidate has their confirmation, so a slow model never costs anyone their application. It never blanks a field the CV does not mention, and never overwrites what a recruiter typed. A value that does not fit your option lists is reported on the card rather than silently dropped. Once the city is known, the candidate is placed on the map, so distance to every role is ready too.

A CV for someone who did not apply. You found a candidate yourself, for example on LinkedIn. Save their profile with LinkedIn's own Save to PDF, open their contact in HubSpot, and upload the file under Add a CV for this candidate on the Candidate 360 card. It accepts one PDF or Word (.docx) file, up to 10 MB. The app reads it the same way it reads a CV from the careers page and fills the same fields. The card then shows what it found and links to the document. The file stays private in your portal. HubSpot stores the upload as private and the app checks that before it reads anything, so the file never sits behind a public link. A file HubSpot marks as sensitive data is refused rather than read. No application is created, so none of your "thanks for applying" emails go out.

4. Recruiters see who fits, and why#

The Job Posting Command Centre card on the Field Sales Representative posting, Matches tab, after "Rank candidates": "2 candidates scored". A "Show travel time" button. #1 Seán Murphy, 95/100, with ticked tags B2B Sales, Prospecting and Negotiation, and "22 km away — inside the 50 km radius". #2 Eoin Walsh, 30/100 with a "gate triggered" tag, crossed tags B2B Sales, Prospecting and Negotiation, and "0 km away — inside the 50 km radius". Below: "Reports for this posting" and "Open the recruiting dashboard".

On the posting record: candidates within the role's radius, ranked. Seán has every must-have skill; Eoin is local but has none of them, so the must-have gate holds him down whatever else he scores.

Ranking scores everyone who is open to work against the role across 10 dimensions — skills, seniority, location, salary, experience, education, languages, industry, keywords — and changes nothing until you save a score to an application. Distance is a straight line from stored coordinates and costs nothing; Show travel time fetches the real driving or public-transport commute on demand, is stored nowhere, and never enters a score. How the arithmetic works, and how to change it: ats-matching.md.

Each ranked candidate is marked Applied or Not applied for this role. To act on the ones who have not applied, tick them and press Shortlist for outreach: the app marks each contact as shortlisted for this posting (the contact properties Shortlisted For Posting, Shortlisted For Posting ID and Shortlisted At), up to 50 at a time. That is all it does. It creates no application, so none of your "thanks for applying" workflows fire, and it sends no email. To approach them, build an active contact list on Shortlisted For Posting and use it in your own sequence or workflow, where your subscription settings govern the send. A contact carries one shortlist mark at a time: shortlisting them for another role replaces it.

Whether you may email people who have not applied is your legal question, not the app's: consent rules differ per country, and a candidate's open-to-work mark comes from an earlier application or from a recruiter, never from anything the person said about this role.

5. Each application has its own workspace#

The Application Workspace card for Aoife Brennan, aoife.brennan@ats-demo.example.com, posting Senior Product Manager, source Organic, applied 31/10/2026. Under Stage: "Current stage: Final interview" and the buttons "Advance to Offer preparation", "Reject" and "Withdraw". Under Match: 97/100, "Scoring rules v3".

On the application record: where the candidate is, and the next step in one click. The same card lists the interviews and documents below the score.

Stage changes happen here, on the pipeline page, or from a workflow — always through the same rule, so a rejection needs a reason wherever it is made. The score stored on the application keeps the per-dimension breakdown and the settings that produced it, so a score you look at six months later can still be explained.

The candidate's Contact carries a Candidate 360 card: every application this person has made, their documents, and the other open roles they fit best. Interviews get their own card, where an interview cannot be marked completed without notes.

The Candidate 360 card on Aoife Brennan's contact. Under Applications: Senior Product Manager, tagged Final interview and 97/100, applied 31/10/2026, with an Open application link. Under Documents: "No documents on file." Under "Best matching open jobs", after pressing "Find best matching open jobs": Senior Product Manager 97/100 "Already applied", 0 km away, scored on remote or hybrid working, not distance; Backend Engineer 30/100, 0 km away; Field Sales Representative 30/100, 220 km away, outside the 50 km radius; Customer Success Manager 30/100.

On the candidate's Contact: what they applied for, and every other open role scored for them. Here the one she applied for is also her best fit, and the sales role in Cork is out of reach on distance as well as skills.

6. The whole pipeline, in one place#

The recruiting app page dashboard. Four figures: Live postings 4, Active applications 10, Interviews next 7 days 3, Hires this quarter 1. Left, "Pipeline right now": a bar per stage — Submitted 3 (30%), Recruiter review 1, Screen 1, Hiring manager 1, Interview 2 (20%), Final interview 1, Offer out 1 (10% each) — with the caption "Active applications per stage. The percentage is that stage's share of all active applications." Right, "Applications, last 12 weeks", a line chart by week, and "11 applications in the last 30 days."

The recruiting app page inside HubSpot. Every figure is read from your CRM records when the page opens.

The Pipeline tab of the recruiting app page: a Posting filter, Refresh, and "10 active applications". Under "Submitted (3)": Niamh O'Callaghan — Senior Product Manager, applied today; Sophie Laurent — Customer Success Manager, 2 days in stage; Mark Henderson — Field Sales Representative, 5 days in stage. Under "Recruiter review (1)": Ciara Doyle, 5 days in stage. Under "Screen (1)": Tomasz Nowak, 8 days in stage. Each row has a "Move to" stage picker on the right.

The pipeline, grouped by stage, filterable by posting. "Move to" changes the stage without opening the record.

The app page also lists postings, upcoming interviews and reports, including a hiring funnel per posting, and has a form to create a new posting.

7. Automate with HubSpot's own workflows#

Six workflow actions — set application stage, recalculate match score, close job posting, create interview, read résumé and find best matching job postings — sit in HubSpot's workflow editor next to everything else. "Score every new application", "move high scorers to recruiter review", "close the posting and thank the remaining applicants after the deadline" are ordinary workflows. An action placed in a workflow of the wrong object type stops with a plain message and changes nothing.

Find best matching job postings is for contact workflows, and it is how a recruiter hears about a strong new candidate without opening a record. It scores the contact against every live posting, with the same rules and the same result as the Candidate 360 card, and returns the best posting at or above a minimum score you set (70 unless you change it): its title, its record ID, its owner and its score, plus how many postings reached the minimum. It sends nothing itself. Follow it with a branch on that count and HubSpot's own internal notification or task. Enrol the workflow on the contact property Candidate profile ready, not on contact creation: that property is filled in once the CV has been read and the address turned into coordinates, and a candidate scored before then has no skills or location to score. It is only filled in when the CV was read successfully, so a candidate whose CV could not be read does not trigger the alert.

Where the data lives and what leaves the portal#

Everything ATS stores is a record or a property in your portal. The app's own server answers the careers pages and the cards and keeps no copy of candidate data. Three things leave the portal, each only when you have added that provider's key: the text of a CV goes to your AI provider to be read; an address goes to OpenCage to become coordinates (stored on the record — OpenCage allows that); two sets of coordinates go to Google when someone presses Show travel time (stored nowhere — Google does not allow it).

Data model#

Most HubSpot portals allow around 10 custom objects, so the app installs in tiers and starts small.

TierObjectsWhat it is for
minimal2Only what the careers site needs to run
lean (default)4ats_job_posting, ats_application, ats_interview, ats_candidate_document — a working ATS
standard6Adds a candidate profile and an offer object
full19Requisitions, compliance, scorecards, skill junctions. Needs a raised custom-object allowance; installed deliberately
ObjectWhat it holds
ats_job_postingA role: its public content, address, status, and the criteria candidates are scored against
ats_applicationOne person applying to one role: stage, dates, and the stored score with its breakdown
ats_interviewA scheduled interview, its notes and its outcome
ats_candidate_documentAn uploaded CV or attachment, and what was read from it
contact (extended)Recruiting fields on the standard Contact — skills, seniority, experience, education, languages, salary expectation, work preference, open to work, location, the shortlist mark, and when the candidate's profile was ready to match
app_settingThe app's settings, including every matching rule

Associations use labels, and object type ids are looked up at runtime, because they differ on every portal.

What it leaves to other tools#

  • Posting into job boards' own systems. Roles reach aggregators through the feed they poll and through Google Jobs; an integration that posts into Indeed's or LinkedIn's systems is not part of the app.
  • Matching on meaning. Skills match literally, widened by a synonym list you edit (Node matches Node.js). The semantic option currently behaves as synonym matching.
  • Candidate accounts. The status link is the candidate's whole self-service, by design.
  • Scorecard and offer-letter authoring. The objects exist at the higher tiers; filling them is record-level editing.
  • Formal compliance reporting. Compliance objects exist at the full tier as data; a regulated process still needs its own review.
  • CVs in the old .doc format. PDF and .docx are read; .doc is refused at upload.
  • Importing from LinkedIn. LinkedIn opens its recruiting API only to certified partners. Candidates reach the app through your careers page. For someone you find on LinkedIn, save the profile as a PDF and add it on their contact (step 3): the app reads it like any other CV.
  • Email and calendar. HubSpot's own Outlook and Gmail connection already logs mail and meetings on the candidate's contact; the app adds nothing there.
  • Sending outreach. The shortlist marks who to approach; your own sequences and workflows do the approaching.

Where to go next#

You areRead
Deciding how candidates should be ranked, or tuning itats-matching.md — every dimension's arithmetic, all the settings, and how to change them safely
Building against it, or reviewing its securityats-integration.md — the careers site, the job-board feed, the match API and how each route is protected
Installing or running it on a portalats-setup.md — tiers, keys, what install creates, adding the cards to your record pages, sample data
Deciding whether it fitsThis page, then ask for a walkthrough

Click the image to switch between fit and full size. Esc closes.