A governed measurement system for US residential mortgage brokerages, from search impression and intake through consultation, application, and funded loan.
A mortgage dashboard can look healthy while the funded-loan column stays blank. The usual cause is not one bad channel. It is a broken definition chain: a call click becomes a “lead,” a scheduled consultation becomes an “opportunity,” and every downstream team reports a different cohort.
Mortgage broker marketing KPIs become useful only when every stage has one entry rule, timestamp, source system, owner, and exit rule. This guide builds that chain for a US residential brokerage. It covers acquisition and handoffs, not underwriting, rates, product selection, borrower eligibility, or compensation.
Operating rule: report impressions, clicks, intake events, enquiries, consultations, applications, submitted files, and funded loans separately. Use the brokerage's own mature cohorts. If a field is missing, mark it unavailable or pending; never replace it with an industry average or zero.
Compliance notice: This marketing measurement guide is not financial, legal, lending, licensing, tax, or investment advice. Confirm terminology, advertising content, identifiers, state requirements, disclosures, data handling, and implementation with your compliance officer, CCO, and qualified counsel. Equal housing opportunity where applicable. Advertising is not a commitment to lend. Past performance is not indicative of future results.
Start with the mortgage jobs and constraints the dashboard must explain
Build the dashboard from the brokerage's verified operating scope, not from a downloaded mortgage KPI template. Record supported enquiry types, licensed coverage, branch model, team capacity, timing categories, privacy limits, and controlled economic fields first. An unavailable or inapplicable value is more useful than an assumed national average.
A purchase enquiry tied to an offer deadline consumes a different intake path from a non-urgent refinance review. An investor or non-standard-documentation request may be unsupported. A licensed state may still have branch, sponsorship, advertising, identifier, or public-location conditions that need review. NMLS is the system of record for participating jurisdictions, but the relevant state or territorial authority grants or denies license authority. Use Regulation Z §1026.24 as a qualified-review gate when an advertisement states dwelling-secured credit terms; the rule addresses actual availability, clear-and-conspicuous disclosures, and prohibited advertising practices.
Mortgage job-and-constraint card
| Field | Controlled entry | Pause condition |
|---|---|---|
| Supported and excluded jobs | Purchase; rate-and-term refinance; cash-out refinance; investment property; non-standard documentation: include only if actually offered and approved | Offering or terminology unverified |
| Licensed footprint | States, responsible firm, branch, public-facing location, and required identifiers verified by compliance | Authority or advertising treatment unclear |
| Capacity | Available loan-officer consultation slots, processor capacity state, internal review hours | Intake cannot honor advertised availability |
| Timing | Planned research; offer/pre-approval deadline; closing deadline; non-urgent review | Category could imply borrower advice |
| Data boundary | Approved marketing fields, consent state, opaque join key, suppression threshold | Borrower PII or financial documents would enter marketing tools |
| Economics | Internal loan-amount band and gross-fee field; finance owns definitions | Field would be presented as advice, pricing, or a portable benchmark |
| Local density | Declared observed-competitor method and review date | Research is mistaken for market share |
Seasonality belongs on this card as an evidence question. Compare the brokerage's own declared windows while controlling for purchase/refinance mix, local market, rate-cycle context, campaign changes, and licensed-state coverage. Do not label a month “busy season” because a national article says so. The SBA's planning guidance supports examining demand, location, saturation, and alternatives; it does not predict channel performance.
Lock the stage dictionary before calculating a KPI
Define the full sequence before anyone writes a formula: impression, click, call click, connected call or form, unique enquiry, qualified enquiry, booked consultation, completed consultation, application, submitted file, and funded loan. Each row needs its own clock, system, owner, next-stage key, and exclusions.
| Stage | Entry rule and timestamp | System / owner | Next key and exclusions |
|---|---|---|---|
| Impression | Search Console counts a result shown; GSC date | Search Console / SEO | Filtered page-query-country-device cohort; note suppressed queries and partial days |
| Click | Search Console counts a Search click; GSC date | Search Console / SEO | Landing session/campaign key; exclude non-comparable filters |
| Call click | Unique eligible user triggers approved phone action; event time | Site analytics / digital analytics | Approved event ID; exclude repeat, test, bot, and actionless pages |
| Connected call | Approved call system records a human connection under written rule; connection time | Call log / intake | Opaque enquiry key; exclude abandoned, staff, vendor, and test calls |
| Form | Backend accepts a submission; server time | Form backend / web-intake | Opaque enquiry key; exclude spam, tests, vendors, employment, duplicates |
| Unique enquiry | One connected call or valid form after deduplication; first received time | CRM / intake | CRM enquiry ID; exclude invalid consent and unsupported records per policy |
| Qualified enquiry | Meets written state, job, timing, capacity, consent, and contactability rule; disposition time | CRM / intake + compliance | CRM enquiry ID; exclude spam, duplicates, unsupported scope, uncontactable |
| Booked job / consultation | Confirmed broker consultation; first valid booking time | CRM + calendar / scheduling | Appointment ID; count reschedules once, exclude tests and duplicates |
| Completed job / consultation | Consultation marked held; attendance time | Calendar + CRM / brokerage operations | Appointment-to-CRM ID; cancellations and no-shows remain separately visible |
| Application | Valid application start under the written LOS rule; LOS start time | LOS + CRM / LOS owner | Approved opaque key; exclude tests, duplicates, existing applications, unsupported jobs |
| Submitted file | LOS reaches the brokerage-defined submitted state; LOS state time | LOS / brokerage operations | LOS record ID; exclude tests and duplicates; no approval inference |
| Approval / clear-to-close | Optional only if the brokerage uses a reviewed definition; LOS state time | LOS / brokerage operations | LOS record ID; keep separate from closing and funding |
| Funded loan | LOS records funded under the approved rule; funding time | LOS / operations + compliance | Opaque loan key; exclude duplicate and test records; retain explained fallout |
| Fallout | Written neutral disposition occurs; disposition time | CRM or LOS / stage owner | Prior-stage ID; never relabel unknown as denied or unfunded |
| Duplicate | Written matching rule identifies the same enquiry/application; detection time | CRM or LOS / data owner | Surviving record ID; preserve audit trail |
Google Analytics recommends distinct events such as generate_lead, qualify_lead, disqualify_lead, working_lead, and close_convert_lead. Those labels can support collection, but the brokerage's written mortgage rules govern meaning. Where teams go wrong is reusing one “lead date” for form receipt, qualification, and booking; cohort lag then becomes impossible to diagnose.
Turn a governed stage map into a publishable content plan. theStacc Content SEO can research keywords, draft and score content, queue it, and publish it to a connected CMS. Compliance Profiles add configured disclosures during planning and require the licensed professional's review.
Measure impressions and clicks without calling either a lead
Use Search Console for Google Search impressions and clicks, then preserve identical page, query, country, and device filters when calculating search CTR. Split licensed geographies and brand versus non-brand queries under a documented rule. Neither average position nor traffic shows that an enquiry, consultation, application, or funded loan occurred.
Google documents impressions and clicks as separate Search Console metrics. Keep them separate in the dashboard and use campaign tags for other attributable channels. For deeper organic definitions, use the SEO KPI guide; for editorial-program measures, use the content marketing KPI guide. This page owns only their handoff into mortgage intake.
| KPI | Numerator / denominator | Window, source, owner | Exclusions |
|---|---|---|---|
| Search CTR | Google Search clicks / impressions under identical page, query, country, and device filters | Declared 28-day or calendar month; Search Console; SEO owner | Non-comparable filters, partial days, unsupported geographies; note suppressed query data |
| Call-click rate | Unique users triggering approved call-click event / unique eligible landing-page users from same attributable source cohort | Declared 28-day acquisition window; site analytics + event log; digital analytics owner | Repeat clicks, staff/tests, bots, pages without call action |
| Valid-form rate | Unique valid forms with source and timestamp / unique eligible form starts; if unavailable, eligible landing users and label the denominator | Declared 28-day acquisition window; form backend + site analytics; web/intake owner | Spam, tests, duplicates, vendors, employment, unsupported scope |
A search click may land on a licensed-state page and still leave. A call click may open the dialer without connecting. Make that visible. If acquisition filters change mid-month, close the old window and start another instead of blending unlike evidence.
Keep call clicks and forms separate at intake
A call click is an interface event, a connected call is a human contact, and a form is a submitted record. Track all three independently. Privacy-approved call measurement, server-side form source capture, spam and duplicate rules, unsupported-scope flags, and a named human intake owner are required before qualification begins.
Use one approved source token carried from landing session into the call or form record. The intake owner then assigns a unique enquiry ID after deduplication. Do not copy a borrower's name, email, financial details, documents, application fields, or conversation transcript into analytics. When a vendor cannot support the approved retention and access rules, leave the join unavailable rather than exporting more data.
Source-system and privacy map
| System | Permitted join | Retention / access owner | Never place in marketing tools |
|---|---|---|---|
| Search Console / site analytics | Page, campaign, aggregate cohort token | Approved policy / analytics owner | Borrower identity, financial or application data |
| Call system / form backend | Opaque enquiry token and source | Approved policy / intake owner | Documents, sensitive transcript, unapproved consent fields |
| CRM / calendar | Enquiry and appointment IDs | Approved policy / operations owner | PII exported to ad or web analytics |
| LOS / accounting | Compliance-approved opaque ID; aggregated cohort output | Approved policy / LOS and finance owners | Application documents, borrower financials, account or payment data |
What actually breaks is mundane: a mobile call lacks its campaign token, one person submits twice, or staff test the form after launch. Put “source missing,” “duplicate,” and “test” in the failure register. Silent cleanup makes the valid-form rate look better while hiding collection defects.
Measure qualification, booked jobs, and completed jobs as three handoffs
Qualification, booking, and attendance answer different operating questions. Qualification applies the approved state, request, timing, capacity, consent, and contactability rule. A booked job means a scheduled broker consultation; a completed job means it occurred. Cancellations, no-shows, reschedules, and capacity declines remain separate dispositions.
| KPI | Numerator / denominator | Window, source, owner | Exclusions |
|---|---|---|---|
| Qualified-enquiry rate | Unique enquiries qualified under written state/job/timing/capacity rule / all unique attributable enquiries in same intake cohort | 28-day intake cohort + stated qualification lag; CRM/intake joined to sources; intake owner with compliance approval | Spam, duplicates, vendors, employment, unsupported state/job, records lacking required consent |
| Booked-job rate | Unique qualified enquiries with confirmed broker consultation / all unique qualified enquiries in same cohort | 28-day intake cohort + declared booking lag; CRM + calendar; intake/scheduling owner | Reschedules counted once, duplicate bookings, tests, bookings outside cohort rule |
| Completed-job rate | Unique scheduled consultations marked held / all unique booked consultations in same cohort | Booked-consultation cohort + declared attendance lag; calendar + CRM disposition; brokerage operations owner | Reschedules once; cancellations/no-shows stay in denominator; internal/tests excluded |
Capacity declines deserve their own count. If consultation slots disappear in one licensed market, marketing may be doing its job while scheduling cannot accept the demand. Conversely, a high booking count with many no-shows points to confirmation or expectation-setting, not an acquisition diagnosis. Neither pattern authorizes a claim about borrower quality.
Connect marketing to applications and funded loans without rewriting lending operations
Join completed consultations to application starts, submitted files, optional approval states, and funded loans with compliance-approved CRM and LOS identifiers. Marketing may receive suppressed cohort counts and status timestamps, not borrower files. The measurement layer documents handoffs; it does not instruct underwriting, approval, rates, eligibility, or product selection.
| KPI | Numerator / denominator | Window, source, owner | Exclusions |
|---|---|---|---|
| Application-start rate | Unique completed consultations followed by valid application start / all completed consultations eligible under written rule | Completed-consultation cohort + declared application lag; CRM + LOS approved key; operations/LOS owner | Duplicate/tests, existing applications, unsupported jobs, privacy-suppressed records |
| Funded-loan rate | Unique cohort applications recorded funded / all unique valid applications started in same cohort | Application cohort + declared lag long enough for observed cycle; LOS + CRM source; operations owner with compliance sign-off | Duplicate, withdrawn, test; fallout stays in denominator unless written rule explains otherwise |
| Cost per funded loan | Direct attributable channel spend + explicitly costed internal marketing/review labor / unique funded loans under approved attribution rule | Declared acquisition cohort + full stated funding lag; invoices/payroll assumptions + CRM/LOS; finance owner with marketing/compliance sign-off | Unattributable spend, uncosted labor, duplicates, non-funded files; disclose omitted shared overhead |
Submitted file and approval or clear-to-close can appear as counts if the brokerage uses reviewed definitions, but neither is the funded-loan numerator. Recent cohorts will often show pending downstream outcomes. The correct display is “cohort not mature” with its observed lag, not a zero funded-loan rate.
Segment KPI evidence by brokerage economics
Segment only with controlled brokerage fields: supported job type, licensed market or branch, source, purchase/refinance mix, internal loan-amount band, deadline category, team-capacity state, and observed competitive-density band. Suppress small cohorts and personally identifiable data. Never present internal amounts or fees as portable industry economics.
Job type and urgency segmentation
| Dimension | Allowed values | Reporting guardrail |
|---|---|---|
| Purpose | Purchase; rate-and-term refinance; cash-out refinance; investment; documentation profile: each only if offered | Unknown stays separate; commercial excluded |
| Timing | Planned research; offer/pre-approval deadline; closing deadline; non-urgent review | Operational routing only, never borrower advice |
| Economics | Brokerage-defined loan-amount band and gross-fee field | Controlled records only; finance owns; suppress small cells |
| Capacity | Consultation slots available/constrained; processor capacity available/constrained | Record state at enquiry time |
Local-density worksheet
| Market record | Required evidence | Limitation |
|---|---|---|
| State / branch / public location | Licensed coverage verified by brokerage; public-location eligibility reviewed | No universal branch or identifier assumption |
| Observed competitors | Count named local and organic results under a recorded query method | Snapshot, not market share |
| Audit trail | Review date, researcher, query/location method, compliance reviewer | Refresh after material market or coverage change |
Do not compare refinance enquiries in one state during one rate context with purchase enquiries in another state after a campaign change. Hold source, market, mix, cohort maturity, and capacity state constant where possible. If the cell falls below the approved privacy threshold, suppress it and roll up only to an approved level.
Run a governed review cadence
Review one declared cohort window at a time, allow each downstream lag to mature, and assign every stage to marketing, intake, operations, compliance, or finance. Diagnose missing or invalid tracking before changing a channel. A keep, change, or stop decision should state the evidence, limitation, owner, and next review date.
Dashboard wireframe
| Tile field | What to show | Operator question |
|---|---|---|
| Stage count | One tile each for impression, click, call click, connected call, form, unique enquiry, qualified enquiry, booked consultation, completed consultation, application, submitted file, funded loan | Where did volume change? |
| Rate | Approved numerator and denominator shown beside the value | Is the handoff definition intact? |
| Cost | Declared attributable spend and included labor | What is omitted? |
| Evidence | Window, cohort start, maturity lag, source, owner, exclusions | Are comparisons valid? |
| Quality | Complete, partial, pending, suppressed, or broken | Is this a tracking failure? |
Failure-state register
Use neutral, stage-specific values: spam, duplicate, wrong state, unsupported loan purpose, uncontactable, capacity decline, canceled consultation, no-show, rescheduled, incomplete application, withdrawn file, denied/not-approved outcome, unfunded/unknown, source missing, and compliance suppression. A qualified reviewer should approve the vocabulary and access. Unknown is a data state, not permission to infer why a file did not fund.
Start the review with data-quality flags. If forms fell because source capture broke, changing channel allocation compounds the error. If qualified enquiries are stable but booking falls only when slots are constrained, the owner is scheduling or capacity. Channel choice itself belongs in the separate Google Ads versus SEO guide; this dashboard measures the choice after it is made.
theStacc Compliance Profiles can inject configured license details, responsible-firm wording, and not-advice language during planning, steer drafts away from prohibited claims, and gate each draft through a human verdict of None, Hold, or Block. Automated and agent-key callers cannot override that verdict. The licensed professional remains responsible. The Content SEO module handles keyword research, drafting, scoring, queuing, and connected-CMS publishing; it does not replace compliance, LOS, CRM, intake, or finance controls.
Build content around a measurement system your reviewers can inspect. Bring the job card, stage dictionary, approved disclosures, and review owners to the strategy call.
Frequently asked questions about mortgage broker marketing KPIs
These answers cover implementation questions that appear after the stage dictionary is agreed. They preserve the boundary between marketing evidence and lending outcomes, add practical rules for cohort maturity and privacy, and avoid compensation, profitability, rate, underwriting, approval, or borrower-qualification advice. A qualified reviewer should approve the final operating definitions.
Which marketing KPIs should a mortgage broker track?
A mortgage broker should track each handoff from impression through funded loan: impressions, clicks, call clicks, connected calls, forms, unique enquiries, qualified enquiries, booked consultations, completed consultations, applications, submitted files, and funded loans. Add cost, cohort window, source, owner, exclusions, and a data-quality flag to every rate; never compress the chain into one conversion metric.
Is a call click or form submission a mortgage lead?
No. A call click records an attempted action, while a form records submitted data; neither proves a connected conversation, a unique person, a supported request, or consent. Keep each as its own intake event. Apply bot, test, spam, duplicate, supported-state, supported-job, and privacy rules before creating a unique enquiry, then apply the written qualification rule separately.
What makes a mortgage enquiry qualified?
A qualified mortgage enquiry meets the brokerage's written, compliance-approved rule for licensed-state coverage, an actually supported request, declared timing, available capacity, required consent, and contactability. Qualification is an intake disposition, not a statement about borrower eligibility, approval, terms, or likely funding. Unsupported, duplicate, spam, uncontactable, and capacity-declined records need separate neutral dispositions.
What is the difference between a booked job, a completed job, an application, and a funded loan?
A booked job is a confirmed broker consultation; a completed job is that consultation marked held. An application starts only under the brokerage's documented LOS rule, and a funded loan appears only when the LOS records funding. Booking, attendance, application, submission, approval or clear-to-close, closing, and funding remain distinct events with separate timestamps and owners.
How should a mortgage brokerage measure SEO without stopping at traffic?
Measure mortgage SEO by joining Search Console impressions and clicks to privacy-approved site, call, form, CRM, calendar, and LOS events. Preserve page, query class, device, country, licensed geography, and brand/non-brand filters. Report the later qualified-enquiry, consultation, application, and funded-loan stages by acquisition cohort; traffic and average position alone cannot establish brokerage value.
How long should a mortgage marketing KPI window be?
Use a declared 28-day or calendar-month acquisition cohort for the approved formulas, then add the brokerage's observed qualification, booking, attendance, application, and funding lags. Compare only mature cohorts with identical filters. A short recent cohort may describe impressions or clicks while its funded-loan field remains pending rather than being treated as zero.
How should purchase and refinance enquiries be separated in reporting?
Separate purchase and refinance enquiries only when each is an actual supported offering and the classification has qualified approval. Give each record one controlled purpose value, keep unknown separate, and compare the same licensed market, source, cohort, and maturity. Do not infer national seasonality or combine cash-out, investor, or documentation-profile requests unless the brokerage explicitly supports and defines them.
How can a brokerage connect marketing data to its LOS without exposing borrower data?
Connect marketing records to the LOS with a compliance-approved opaque join key stored in controlled systems, not borrower names, email addresses, financial fields, application documents, or loan details in analytics. Restrict access, document retention and deletion, and expose only suppressed cohort counts to marketing. The compliance owner must approve the key, field map, and minimum cohort threshold.
Build the first governed mortgage marketing dashboard
Begin with definitions, not charts. Approve the operating card, stage dictionary, opaque join key, source map, formulas, cohort lags, suppression threshold, and failure register. Then populate one mature historical cohort from controlled records. Leave unavailable fields blank and labeled; a clean limitation is safer than false precision.
- Have compliance and operations verify supported jobs, licensed footprint, public-location treatment, required identifiers, and permitted fields.
- Have each system owner test one record through Search Console or campaign source, intake, CRM, calendar, LOS, and suppressed finance output.
- Reconcile counts at every handoff. Fix duplicates, missing sources, timezone differences, and reschedule logic before publishing rates.
- Publish each stage as its own tile with formula, evidence window, cohort lag, source, owner, exclusions, and quality flag.
- Record keep, change, or stop decisions with a limitation and next review date. Never promise that the decision will improve funding or business results.
Send the final dashboard, marketing language, disclosures, identifier treatment, privacy map, and state-specific statements to your compliance officer or CCO before use. Equal housing opportunity where applicable. Advertising is not a commitment to lend. Past performance is not indicative of future results.
Make the stage dictionary the brief for every future campaign. That gives marketing, intake, operations, compliance, and finance one auditable language from first impression to funded-loan record.
Sources & references
Blog SEO, Local SEO, and Social Media — one dashboard, no headaches.
Weekly local SEO teardowns
One practical email a week. Map Pack, GBP, AI Overviews — no fluff. Unsubscribe anytime.