HubSpot Smart CRM: Data Model, Editions, AI Inputs and Custom Objects
HubSpot Smart CRM is the database beneath every hub and a product sold in editions: its data model, edition limits, the inputs its AI reads, and custom objects.
Paul Maxwell, PhD
AUTHOR
GET WEEKLY REVOPS INSIGHTS
No spam. Unsubscribe anytime.
Two misreadings of HubSpot Smart CRM cost a business money. The first treats the name as a marketing label, and misses that HubSpot sells Smart CRM in three editions whose limits decide whether a portal can label an association or create a custom object. The second treats it as a product that becomes intelligent once bought, and misses that the AI features built on it take their facts about an account from what the records hold. A deal whose buyers were never associated with it is summarised fluently, left out of revenue attribution and missing the meetings logged on its contacts, and nothing on the deal record says so.
This article explains what HubSpot Smart CRM is, what is sold under the name, and what depends on it. It starts with the data layer and the editions that license it, then separates the questions the name conflates. It follows one association through four features that read it, tests the finding on a sample portal, and sets out when a custom object is warranted. It closes with an audit procedure, costs and returns, symptoms, common questions and the limits of the evidence.
An object is a type of record, such as contacts, companies, deals or tickets. A property is a field on an object. An association is a relationship between two records, which HubSpot's object documentation defines as always two-way. An activity is a logged call, email, meeting or note. Smart CRM is HubSpot's name for the database holding all of these, and also the name of a product sold in editions.
Smart CRM Components: Objects, Properties, Associations and Activities
HubSpot's Product and Services Catalog describes Smart CRM as the "underlying database that powers each HubSpot product". The object documentation states that all objects run on one data platform, which lets a ticket be associated with a contact a marketing form created, and one report span both.
Every account carries the standard objects, and a Super Admin can activate further HubSpot-defined objects (appointments, courses, listings, services and projects) in the data model tool at More > Data Management > Data Model. Contacts are deduplicated by email address and companies by domain name, while deals and tickets are keyed only on Record ID, which cannot recognise two deals describing one opportunity; duplicate management covers what the tools can merge.
The property documentation states that a property's object type and internal name can never be edited, that some properties in use on records or in segments and workflows cannot be edited at all, and that changing a field type can invalidate the values already stored. A field type is therefore cheapest to settle before the first record carries a value.
Association labels state what a relationship means, such as the buyer on a deal or the parent of a company. Under the association label documentation, a Super Admin can create up to 50 labels per object pair, and segments, workflows and custom reports can all filter on a label. Activities appear on the timeline of every record they are associated with, and some are associated automatically at the moment they are logged, a rule the section on associations follows.
Smart CRM Editions and Their Documented Limits
Smart CRM is also a product. The catalog states that its features are included with the Starter, Professional or Enterprise edition of any premium product, and that Smart CRM can be bought on its own in the same three editions. A free edition carries select functionality for up to two users and 1,000 contacts. The knowledge base follows the pattern with one gap: the association label article lists Professional or Enterprise of Marketing, Sales, Service, Content or Data Hub, or of Smart CRM, and omits Revenue Hub, while the custom objects article lists Enterprise of any hub or of Smart CRM.
| Capability | First edition | Documented limit |
|---|---|---|
| CapabilityCustom properties | First editionStarter | Documented limit1,000 per object at every paid edition; 10 in total in the free tools |
| CapabilityPipelines | First editionStarter | Documented limit15, 100 and 350 per account across all object types, by edition |
| CapabilityData quality overview (beta) | First editionStarter | Documented limitAccount scanned every two weeks at Starter, daily at Professional and Enterprise; the knowledge base lists standalone Smart CRM only from Professional |
| CapabilityAssociation labels | First editionProfessional | Documented limit50 per object pair; a paired label counts as one |
| CapabilityCalculated properties | First editionProfessional | Documented limit40 at Professional, 200 at Enterprise |
| CapabilityDuplicate management | First editionProfessional | Documented limitContacts and companies |
| CapabilityCustom objects | First editionEnterprise | Documented limit10 definitions and 1,000,000 records on Marketing, Sales, Service or Content Hub Enterprise; 20 and 1,500,000 on Data Hub; 10 and 1,500,000 on Revenue Hub |
| CapabilityField-level permissions | First editionEnterprise | Documented limitEnforced in the interface, not the API, as set out in field-level permissions |
The Name and the Four Questions It Conflates
Discussion of Smart CRM runs four questions together: what the layer is and what is sold under the name, both answered above; what changed with the name; and whether data quality now governs AI output. The last two rest on evidence of different strength.
The earliest HubSpot page located that uses the name is the Fall 2024 Spotlight, a page dated 1 October 2024 that describes enriching records "in HubSpot's Smart CRM". That establishes the name was in use by that date, not when it was introduced. The knowledge base still writes "your HubSpot CRM", including in the custom objects article whose availability header lists Smart CRM as a subscription, so HubSpot's own documentation uses both names for one database.
Neither Operations Hub nor Commerce Hub appears in the catalog or in the visible text of HubSpot's products page, which places programmable automation, data sync and data quality under Data Hub, and quotes, invoices, subscriptions and payments under Revenue Hub. Two traces of the older names remained in September 2026: the address hubspot.com/products/operations redirected to the Data Hub page and hubspot.com/products/commerce to the Revenue Hub page, and the products page's structured data still listed "Commerce Hub" at the second address. HubSpot has stated both changes: a footnote to its Fall 2025 Spotlight reads "Data Hub replaces Operations Hub", and the Revenue Hub product page describes "Revenue Hub, formerly Commerce Hub".
The link between data and AI output has a documented mechanism and a vendor-reported magnitude without a stated method. The record summary documentation states that the summary card is "dynamically influenced by the activities logged to the record", and HubSpot's Fall 2026 Spotlight says every Breeze Assistant answer comes straight from CRM data. The same page gives what the AI reads a new name, Growth Context, which it defines as everything HubSpot knows about a business, its team and its customers. It reports that customers using AI with high-quality context, against those with low-quality context, generated 5x more MQLs, closed 4.7x more deals and resolved 6.8x more tickets. The page does not state how context quality was measured or how the two groups were formed, and a comparison between customers cannot separate the effect of context from the kind of business that keeps complete records. The finding below therefore concerns which records a feature can read, not how much its output improves.
The Contact-to-Deal Association and the Features That Read It
The finding: in Smart CRM, the association between a contact and a deal is an input to four features, each documented in a separate HubSpot article, and the data quality tools measure properties one at a time, so no figure they report shows how many deals carry every input those features need. One of the four readings is not restored when the association is added later. Because each reading is documented on its own, the combined effect of one missing association is stated in none of the articles cited below.
Revenue attribution reads it first. The attribution documentation requires at least one associated contact, since "without an associated contact, the revenue will not be attributed to any interactions". It recommends identifying the right contact and associating it with the deal, after which the report can read that contact's interactions. It adds that the W-shaped and full path models return no value when a deal was created before any of its associated contacts.
The deal's traffic source reads it second. The default deal properties set Original Traffic Source on a deal from the associated contact with the earliest activity, falling back to the company's value when no contact is associated, so a deal without its contacts reports where the company came from rather than the buyer.
Automatic activity association reads it third, and only forward. An activity first logged on a contact is associated with the contact's primary company and its five most recent associated open deals, and the activity association documentation requires the records to be associated already when the activity is logged. The one automatic route to earlier activity is at deal creation: the create-deal panel offers an "Add timeline activity" checkbox that copies the selected contact's calls, emails, notes, meetings and tasks from a chosen period, up to 2,000 of them. A contact associated with an existing deal after three meetings brings none of them, and the deal's summary, built from activities logged to the deal, does not mention them.
The contact's record summary reads it fourth. Under the record summary documentation, the summary of a prospect contact shows Number of open deals, Potential revenue and Upcoming close dates, figures a deal reaches only through its association with the contact. The summary a rep opens before a call therefore omits the deal.
Through the company association, Breeze Assistant's company summary reports open deals, potential revenue and upcoming close dates for a prospect, and ARR from closed won deals for a customer, figures a deal reaches only through its association with the company. Activities logged on a deal are also associated automatically with its primary company.
Against all of this, HubSpot's data quality tools report on properties: duplicates, properties with no data, unused properties, and a fill rate per property in the report view. A fill rate describes one column, while the features above read a row: a deal with its properties and its associations, and in two of the four readings the order in which they were created.
Deal Readiness on Sample Data
The figures in this section are sample data constructed for this article, not taken from any client portal. The portal holds 400 open deals, measured on four inputs the readers above depend on: the Amount, Close date and Deal type properties, which the deal summary lists, and at least one associated contact.
| Deals | Inputs missing |
|---|---|
| Deals244 | Inputs missingNone |
| Deals48 | Inputs missingAssociated contact |
| Deals40 | Inputs missingAmount |
| Deals28 | Inputs missingDeal type |
| Deals20 | Inputs missingAssociated contact and deal type |
| Deals12 | Inputs missingClose date |
| Deals8 | Inputs missingAssociated contact and amount |
Counted one at a time, the inputs look healthy. Amount is missing on 40 plus 8 deals, leaving 352 populated (88.0 per cent). Close date is present on 388 (97.0 per cent), deal type on 352 (88.0 per cent), and a contact association on 324 (81.0 per cent), since 48, 20 and 8 deals lack one. The four fill rates average 88.5 per cent.
| Reader and the inputs it needs | Deals usable | Share |
|---|---|---|
| Reader and the inputs it needsWeighted amount in the forecast: Amount | Deals usable352 | Share88.0% |
| Reader and the inputs it needsRevenue attribution, if the deals close as they stand: contact, Amount, Close date | Deals usable272 | Share68.0% |
| Reader and the inputs it needsDeal summary with every measured input present: all four | Deals usable244 | Share61.0% |
The forecast row reads Weighted amount, which the default deal properties define as Amount multiplied by Deal probability. Attribution needs the contact, amount and close date together, which only the 244 complete deals and the 28 missing a deal type satisfy. Only 244 deals carry all four inputs, a per-record share 27.5 points below the mean fill rate.
Of the 128 deals attribution cannot use, 76 lack a contact association. Associating contacts with the 48 and the 20 would raise attribution coverage from 272 to 340 deals (85.0 per cent), while filling every missing amount would raise it only to 312 (78.0 per cent). The contact association trails amount by seven points of fill rate and moves 68 deals against 40 in the reader that needs both.
Custom Objects at the Edges of the Standard Model
The standard objects describe a company buying through a deal. A contractor's bid, a managed service agreement and a serialised unit do not fit that shape: a recurring agreement recorded as a deal closes once and carries no later state, the problem set out in service agreement object design for MSPs.
The decision rule is whether the thing has a lifecycle of its own. An entity that is created, changes state and is reported on independently of any one deal is an object; a fact that only ever describes one deal is a property of it. A unit that is sold, installed, serviced and replaced passes the test, as the illustrative serialised equipment case study shows.
Under the custom objects documentation, Enterprise is required, and each object needs a primary display property that can be set to require unique values. Neither the object's internal name nor its properties' internal names can be edited once the object exists. An object in use by other tools cannot be deleted until those references are removed, and a deleted object cannot be restored, so a definition deserves the review a schema migration gets.
Below Enterprise, the object documentation states no Enterprise requirement for activating the appointments, courses, listings, services and projects objects, so the first check is whether one of them appears as activatable in the portal. A separate deal or ticket pipeline is the other route.
Smart CRM Data Layer Audit
- Open More > Data Management > Data Model and read the Limits tab, which reports usage against limits for records, associations, pipelines, custom and calculated properties, association labels and custom objects. Record each figure with the date.
- List the features in use that read the layer and the inputs each one's documentation names, as in the reader table above.
- In the deals index, save a view of open deals and add one filter at a time: Number of associated contacts is equal to 0, then Amount, Close date and Deal type each "is unknown". The filter documentation treats 0 as a known number and matches "is equal to 0" to both 0 and an empty value, so the equality filter finds deals with no contacts where "is unknown" can miss them.
- Save a second view combining the opposite conditions (contacts greater than 0, and all three properties known) and divide its count by the open-deal total. Set that per-record share beside the per-property fill rates, which the data quality documentation places in the report viewer: open a deal report using the three properties and choose More > Data Quality in the right sidebar.
- Open Settings > Data Management > Objects, select Activities, and read the automatic association settings for activities logged on contacts, confirming that meetings reach associated open deals.
- Repair associations first. Associate contacts with the deals that lack them, then associate the meetings already on those contacts with each deal by hand, since automatic association applies only between records associated before logging. For new deals, select the contact in the create-deal panel and tick "Add timeline activity", the only step that copies earlier activity automatically. Fill missing properties at their source afterwards.
- Verify. Rerun both views and confirm the per-record share rose by the number of deals repaired. Open three repaired deals, choose Actions > Summarize, and confirm the summary reflects the meetings associated by hand. Summarize one of their contacts and confirm the deal is counted in its open deals, then confirm any repaired closed won deal now appears in the revenue attribution report.
Costs and Returns of a Shared Data Layer
The layer buys the absence of a sync between functions that work in HubSpot: a form submission, a sales meeting and a support ticket land on one contact, and the AI features read that record without an integration.
The costs come in three kinds. Licensing: labels need Professional, custom objects need Enterprise, and calculated properties and pipelines are capped per edition. Labour at the moment of work: an association made after the meeting does not carry the meeting, so contacts must be attached when a deal is created rather than in a later cleanup. Irreversibility: internal names are permanent and deleted custom objects are unrecoverable, so a careless model is expensive to unwind.
The case is strongest where several hubs serve teams that share customers, deals originate inside HubSpot, and AI features are in use or planned. It is weakest where one team uses HubSpot alone, deals arrive by import without their contacts, or the central entity needs a custom object the edition lacks. Two alternatives win on specifics: an ERP remains the system of record for inventory and the general ledger, neither of which appears among HubSpot's objects, and a data warehouse suits analysis joining CRM records to product usage or finance at a grain HubSpot's objects do not hold. The firm is a HubSpot Solutions Partner and is paid to implement this platform, which is the interest to weigh against this section.
Smart CRM Symptoms and Their Mechanisms
Meetings appear on a contact's timeline and not on the deal. The contact was associated after the meetings were logged, the deal had already closed, the contact had five open deals associated more recently, or the activity settings were changed. The repair is to associate the existing meetings by hand, and to select contacts at deal creation with "Add timeline activity" ticked.
A company summary shows fewer open deals than the pipeline holds for that company. Some deals are associated with a duplicate company record or with none, and the summary reads only associated deals, so associate them and merge the duplicates. The same mechanism explains a contact's summary showing no open deals while the contact is negotiating one: the deal was created without the contact.
A deal's Original Traffic Source describes the company rather than the buyer. The deal has no associated contact and the property has fallen back to the company's value, which associating the buying contact corrects.
A custom report filtered to "Deal type is none of Existing Business" undercounts new business. The report builder documentation states that negative conditions such as "is none of" exclude records with no value, so populate the property or add an "is unknown" branch.
The W-shaped or full path model returns nothing for a group of deals. Those deals were created before their associated contacts, as when contacts are imported after deals; import contacts first in future loads.
Common Questions About HubSpot Smart CRM
What is HubSpot Smart CRM? HubSpot Smart CRM is the database beneath every HubSpot hub: the objects, properties, associations and activities that Marketing, Sales, Service, Content, Data and Revenue Hub all read and write.
HubSpot Smart CRM: what is it as a product, and is it sold separately? It is sold in Starter, Professional and Enterprise editions, and the same features come with the matching edition of any premium hub.
What are the HubSpot Smart CRM features? The data model features are custom properties, pipelines, association labels, calculated properties, duplicate management, custom objects and field-level permissions, gated by edition as the table above sets out.
HubSpot Smart CRM vs Sales Hub: which does a sales team need? Sales Hub is an application over Smart CRM and includes the Smart CRM features of its edition, while Smart CRM alone provides the database without Sales Hub's selling tools. A team that records contacts, deals and activities can work in Smart CRM alone; one that needs sequences needs Sales Hub or Service Hub at Professional or above, per the object documentation.
What is HubSpot Smart CRM pricing? The catalog lists standalone Smart CRM from $20 per seat a month at Starter, $50 at Professional and $75 at Enterprise, as of September 2026. An account holding any premium hub already has the Smart CRM features of that hub's edition.
Is smart CRM HubSpot's free CRM? The free tools include select Smart CRM functionality, limited to two users, 1,000 contacts and 10 custom properties, and the paid editions are the same database with higher limits.
Boundaries of the Evidence
This article does not review the AI features, which change between releases, or advise on edition choice, which HubSpot Starter vs Professional covers; attribution models and coverage arithmetic are treated in attribution reporting.
Every product fact is as documented in September 2026 in the sources linked at each claim, which are authoritative over this text as the platform changes. Several features cited are labelled beta, and the Operations Hub and Commerce Hub mapping rests on a Spotlight footnote, a product page and the redirects, since no dated announcement of either change was found. The sample portal shows how per-property and per-record measures diverge, not the prevalence of portals like it. The link between complete records and better AI output is documented as a mechanism; the only magnitude found is HubSpot's own comparison between customers, published without a method.
In Summary
HubSpot Smart CRM is the database beneath every hub and a product sold in three editions whose limits govern the data model, from association labels at Professional to custom objects at Enterprise. The name was in use by October 2024, and the current lineup lists Data Hub and Revenue Hub, to which the old Operations Hub and Commerce Hub addresses redirect.
The contact-to-deal association feeds four features documented in four separate articles: revenue attribution, the deal's Original Traffic Source, automatic activity association and the contact's record summary. HubSpot's data quality tools report fill rates one property at a time, so no figure they produce counts the deals that carry the association and every property the same features read. On the sample portal, four inputs averaging 88.5 per cent populated left 61.0 per cent of deals complete.
Contacts belong on a deal when it is created, with "Add timeline activity" ticked, before its first meeting is logged. The four features do not recover alike from a late association: attribution recovers when a missing contact is associated later, but automatic activity association runs only forward, and the only automatic copy of earlier activity happens in the create-deal panel.