Choose a mortgage CRM by matching it to your team’s actual workflow, exact LOS and POS editions, access model, and total cost. Do not choose by feature count or by the word “mortgage” alone. If a milestone must move from an LOS into a CRM and trigger a borrower follow-up, verify the field mapping, synchronization direction, duplicate behavior, consent gate, and failure path in a live demonstration.
This guide compares HubSpot, CANDID, Jungo, Shape, and BNTouch using official pages reviewed October 9, 2026. Pricing and capabilities are vendor-stated, not independently tested. The HubSpot comparison article reviewed here visibly displays “Updated: 07/23/25,” although extracted metadata identified a later date. Read the original comparison as a vendor-authored overview, not an independent product test.
The practical goal is a defensible buying decision: define the workflow, assign ownership of important fields, test the real data path, and compare complete commercial terms.
The short answer: choose by workflow fit, not by the word “mortgage”
Start with three decisions before reviewing feature lists:
- Name the workflow to improve. Examples include assigning new leads, tracking referrals through funded loans, or following up after a verified LOS milestone.
- List the exact systems and editions. Record the LOS, POS, lead sources, messaging tools, and product editions your team uses.
- Assign ownership for important fields. Decide which system controls loan status, borrower contact details, referral attribution, and communication preferences.
A connector name does not establish supported fields, direction, latency, permissions, conflict handling, or retry behavior. Those are purchase-test questions.
Compare five mortgage CRM options using current evidence
Use the table to create a shortlist, not to declare a universal winner. Normalize every proposal to the same users, contact volume, term, onboarding, usage, and connector assumptions.
| Product | Strongest documented fit | Published pricing | Verify before buying |
|---|---|---|---|
| HubSpot | Marketing-led lead generation and broader CRM use | Marketing Hub Starter displayed from $7 per seat; Professional $800 per month with 3 Core Seats; Enterprise $3,600 per month with 5 Core Seats. Free CRM is available. | Marketing contacts, seat types, onboarding, add-ons, and the actual POS integration path |
| CANDID | Mortgage-focused workflows and named LOS connections | No public price found in the reviewed official pages. Request a quote. | Exact LOS fields, synchronization behavior, workflow scope, and included onboarding |
| Jungo | Salesforce-based mortgage CRM with mapping and lead-distribution claims | Lite $96 per user per month when billed annually; Mortgage App $119; Bundle $149. Enterprise is custom. | One-year minimum contract, setup fees, required tier, and exact LOS connector |
| Shape | Mortgage workflows, routing, referrals, and communications | Core, Complete, and Enterprise are quote-based. Complete includes 5 users. Communications may add usage charges. | Team price at actual message volume and whether unlimited calling is needed |
| BNTouch | Mortgage CRM with borrower-facing and marketing features | Individual $165 per month. Team starts at $95 per user per month with a 2-user minimum. | Written Team price, activation fees, usage charges, and exact connector scope |
HubSpot: The official Marketing Hub pricing page lists the displayed prices above and one-time onboarding fees of $3,000 for Professional and $7,000 for Enterprise. Pricing varies by billing period, promotional offer, marketing contacts, seats, and add-ons. HubSpot’s seat documentation distinguishes access and seat types. Its free CRM does not establish that it is the only free plan in the market. For a POS connection, a solutions partner or API may be a project route, but neither proves a ready-made mortgage connector. Review the partner directory and API overview, then validate the selected POS.
CANDID: The official FAQ confirms integrations with Encompass, LendingPad, ARIVE, and Nano, along with preconfigured automation and white-glove onboarding. It does not publish detailed mappings, supported editions, synchronization direction, retry behavior, or public pricing. Request a quote and a field-level demonstration.
Jungo: The current pricing page lists Lite at $96 per user per month when billed annually, Mortgage App at $119, and Bundle at $149. It also states a minimum one-year contract and one-time setup fees of $249 for Mortgage App and $299 for Bundle. Jungo describes LOS synchronization, customizable field mapping, pipeline updates, lead distribution, business texting, and opt-outs. Confirm the plan and connector for the exact LOS edition.
Shape: The current pricing page uses Core, Complete, and Enterprise labels rather than older plan names. Pricing is quote-based, with monthly or annual billing. Communications use a usage-based package unless an unlimited calling option is selected. Shape’s mortgage page names several LOS connections, but the exact fields and implementation path still require a demonstration.
BNTouch: The official pricing page lists Individual at $165 per month and Team starting at $95 per user per month with a two-user minimum. Its pages present Team pricing and LOS lists inconsistently. Confirm the final price, activation fees, usage charges, included features, and target connector in writing.
For a fair comparison, request the same user types, marketing-contact volume, annual or monthly term, onboarding, setup, message volume, connector fees, add-ons, and required tier from every vendor. Prices and offers can change.
Define the data model and source of truth before you automate
A practical operating model assigns the LOS authority over loan milestones and loan-file status, the CRM authority over relationship history, tasks, referral attribution, and communication preferences, and the POS authority over borrower-facing intake when the architecture gives it that role. Confirm this model with your team rather than treating it as a vendor default.
Keep records at their natural grain:
- Contact: one person, such as a borrower, co-borrower, agent, or referral partner.
- Loan: one application or loan file. A person may have several loans.
- Referral event: one referral occurrence associated with a partner and loan.
- Communication event: one email, text, call, or other interaction.
- Milestone event: one observed loan-status change.
A contact total cannot stand in for funded-loan outcomes, referral conversion, or communication activity. Preserve source record IDs, event versions, and timestamps so reports can distinguish those measures.
A connector moves data; it does not decide which system owns the truth.
Before synchronization, document each important field’s source, destination, direction, data type, transformation, update frequency, conflict rule, and owner. For a milestone, specify which system is authoritative and which timestamp or version identifies a newer event.
Turn “integration” into a purchase test
Public vendor pages substantiate some high-level connection claims, but they do not consistently provide complete schemas, latency, permissions, concurrency behavior, conflict rules, or retry guarantees. Jungo describes LOS synchronization and field mapping. Shape documents a high-level bidirectional LendingDox workflow. CANDID confirms named LOS connections without detailed mappings. BNTouch lists LOS integrations without establishing identical field coverage for each one.
Ask every shortlisted vendor to demonstrate the same record path: one inbound milestone, one permitted outbound update if supported, a duplicate event, and a failed synchronization. For HubSpot and a POS, ask the POS vendor or proposed implementation partner to show the objects, authentication, permissions, write operations, and error handling. API access is a possible project path, not a ready-made connector.
For concurrent lead intake, “search, then create” is not enough. Two workers can search simultaneously, find no match, and create duplicate contacts. Use a database-enforced unique constraint on a stable source lead ID, a transactional upsert, or a vendor-supported idempotency mechanism where available. Normalize email and phone, preserve the source ID, and route ambiguous matches to review rather than silently merging them.
Compare the workflows that affect daily mortgage operations
| Trigger | AI job | Validation and action | Fallback owner |
|---|---|---|---|
| A website or third-party lead source submits a prospect | None for identity, consent, or state routing. Optional summary of free text after intake. | Normalize contact data, preserve source lead ID, validate consent and licensing coverage, then atomically create or update the CRM record. | Lead operations handles duplicates or missing data. Sales management handles no qualified assignee. |
| An LOS milestone changes | None for applying the authoritative status. AI may summarize status history for review. | Check the exact LOS edition, compare source timestamp or event version, map the field, log the attempt, and quarantine stale or failed writes. | The integration owner handles technical errors. Loan operations resolves conflicting business data. |
| A referral partner is associated with a loan | None for attribution. Optional activity summary for review. | Store a referral event linked to the partner and loan. Define whether one loan has one primary referrer or multiple credited partners. | The referral-program owner resolves attribution disputes. Loan operations resolves record matching. |
| A verified milestone makes follow-up eligible | Rank eligible leads or draft approved-fact language. Do not decide consent or eligibility. | Recheck consent, opt-out, suppression, loan status, and template version immediately before sending. Store a unique communication-event key and delivery result. | Marketing or compliance operations handles suppression issues. The loan officer reviews factual content. |
Shape documents lead capture, routing, referrals, messaging, trigger actions, and a high-level LendingDox workflow. Jungo documents lead distribution, LOS synchronization, field mapping, business texting, and opt-outs. The Shape trigger guide lists email, SMS, MMS, task, portal, LOS, and webhook actions. Its referral guide describes Referral Partner as a distinct relationship model. These sources establish vendor-stated capabilities, not a completed configuration or legal compliance for every campaign.
Example: consent-gated follow-up after a milestone
Use this as an illustrative operating sequence, not a vendor template. The LOS supplies a verified milestone. The CRM records the communication event. A deterministic check confirms current consent, opt-out status, loan status, and message eligibility before any send action. AI may draft wording from approved facts for review, but it does not decide whether outreach is permitted.
Use a communication-event key containing the source system, loan ID, milestone code, template version, recipient ID, and event version. That key identifies one attempted message at its declared event grain and prevents a replay from becoming a second send. If consent is missing, the milestone is stale, or the record is suppressed, save the exception without sending and assign it to the appropriate owner.
For referral reporting, use separate referral events when a loan can have several partners or referral attempts. Calculate funded-loan outcomes from loan records, not contact totals.
Where AI helps, and where a deterministic rule is safer
Use AI for bounded tasks such as ranking otherwise eligible leads, summarizing interaction history, suggesting a next step, or drafting language for review. Use deterministic rules for required identifiers, consent, opt-outs, loan status, state-based assignment, duplicate loan IDs, and closed or withdrawn records.
HubSpot’s current pricing comparison describes plan-dependent scoring, including rule-based scoring and Enterprise AI recommendations. Shape and BNTouch also advertise AI-related capabilities. These are vendor claims, not comparative tests of accuracy or outcomes. If an AI score is stored, retain source record IDs, source timestamps, model or rule version, rationale codes, generated time, and reviewer status.
Make consent a blocking rule, not a prediction. Let AI rank eligible work or draft a suggested message, then require human review when the content is sensitive or borrower-specific.
Run a structured CRM selection and validation process
Map one high-value workflow from trigger to outcome and name the operational owner at each step. Include roles for loan officers, processors, managers, and marketing staff. Test whether each role can see, edit, export, and audit only the borrower information appropriate to its work.
Request a field-level integration matrix and a written quote covering seats, marketing contacts, implementation, setup, onboarding, messaging, connector fees, add-ons, and contract term. Test exports, audit history, duplicate handling, synchronization exceptions, and permission boundaries with realistic sample records. Establish a baseline before rollout and select a measurable outcome such as fewer unassigned leads, shorter time to follow-up, fewer duplicate loan records, or more complete referral attribution.
- The exact LOS and POS products and editions are identified.
- Required fields have an owner, direction, data type, and conflict rule.
- Concurrent duplicates, stale updates, and failed synchronization have visible resolution paths.
- Roles, exports, audit history, and retention pass sample-record tests.
- The written quote includes material seat, contact, setup, usage, onboarding, and connector costs.
Advance a vendor only when the relevant workflow passes the demonstration checklist and the written quote identifies material costs and connector assumptions. Teams that need help mapping requirements across platforms can review CRM systems consulting. For HubSpot system planning and configuration, HubSpot systems consulting may also be relevant.
Frequently asked questions
Is a mortgage-specific CRM always better?
No. Choose based on verified workflow fit, exact integration scope, access model, and the implementation effort your team can support.
Does an LOS integration mean all borrower and loan fields synchronize?
No. Confirm the exact LOS edition, supported fields, direction, timing, permissions, and conflict behavior in a demonstration and written scope.
Can a CRM lead score indicate borrower eligibility?
No. A lead score can prioritize follow-up, but it is not an underwriting, credit, approval, or compliance decision.
How should a team compare CRM prices?
Normalize users, contacts, term, onboarding, setup, messaging, connectors, add-ons, and usage. Confirm the final amount in a written quote.
Is a vendor case study a performance benchmark?
No. HubSpot’s Accord Mortgages case study reports customer results and implementation details, but it is promotional and self-reported. Do not treat its figures as an independent benchmark or assume they will transfer to another lender.