← back to rankingZ · Zillow Group, Inc. Class C
Internet Content & Information · mkt cap $8.4B · calls: Q1 FY2026 vs Q4 FY2025
66.0 conviction · conf-adj 66
conf 2/10 partial
enthusiasm:27.0 · trend:8 · quantifies:0 · impact:0 · under_radar:14 · credibility:12 · business_impact:8 · disruption:-6 · commitment:0 · confirmation:3
Enthusiasm latest 9 / prev 7 (rising)
Zillow's AI thesis is vertical, not horizontal: proprietary housing content (rich media, listings), deep transaction context (search through closing across consumer and Follow Up Boss/agent workflows), and integrated actionability differentiate it from general LLMs. Q1 FY2026 marks a step-change versus Q4 FY2025 — a dedicated AI strategy section, live consumer AI mode at 5% rollout, named products (AI Assist, AI-powered Follow Up Boss), and a quantified internal productivity claim (40% more code per engineer). Management quantifies adoption/usage (5% audience, 7M AI messages) and internal productivity, but does not tie AI directly to revenue, margin, or conversion dollars yet; credibility rests on real shipped products and early engagement signals rather than financial attribution.
GROUNDED NEXT-FY IMPACT vs CONSENSUS
Grounded on actual base — revenue $2.6B · net income $0.0B · net margin 0.9% · diluted EPS 0.0905
These are next-fiscal-year annual uplift estimates, not next-quarter numbers.
Aggregate next-FY est. rev uplift: % · next-FY EPS uplift: % · vs analysts: inline · priced in: low (model's call-read: high; verdict above is the hard-data one used for ranking) · confidence: 2/10
| Claim | Figure | Arithmetic | Next-FY Rev % | Next-FY EPS % |
|---|
AI mode live for ~5% of audience (millions of users)
engagement · soft | ~5% of audience / 'millions of users' | Disclosed rollout share only: 5% of audience ≠ 5% of revenue ($2,583M). No ARPU, take-rate, conversion lift %, or $ bookings disclosed anywhere; mgmt only says these users are 'expected to be higher-converting' (forward, unquantified). 5% reach × (no $/user) cannot convert to revenue. | | |
Engineers ship 40% more code/engineer at same+ quality
productivity · soft | 40% more code per engineer | Productivity output rate, not a disclosed $ opex cut or headcount reduction. Engineering $ base E not in claims or financials (only implied total opex ~$1,947M). If E were known: saving = E×(1−1/1.40)=E×0.2857; after_tax=saving×0.79 — but E is unanchored. GAAP NI $23M (0.89% margin) would make any EPS% a thin-denominator artifact even if a base existed. | | |
Concept-to-launch in days (product/design cycle time)
productivity · soft | 'a matter of days' (qualitative) | Qualitative cycle-time claim; no % faster, headcount, or $ opex/revenue anchor → cannot size next-FY uplift. | | |
Follow Up Boss AI smart messages: >7M sent in 2025
engagement · soft | >7 million messages (FY2025) | Usage-scale metric. No $/message, attach rate, or FUB segment revenue disclosed → 7M × (undisclosed $) cannot be anchored. Feeds retention/engagement, not a stated incremental revenue figure. Not bookings. | | |
>10% of new For Sale listings have 3D tours/floor plans
engagement · soft | >10% of new listings (rich media / AI-training moat) | Content-coverage / data-moat metric, explicitly framed as making 'AI more capable over time'. No lead-gen lift %, take-rate change, or $/listing monetization → cannot map coverage % to the $2,583M revenue base. | | |
AI mode user: 16 conversations / 10 days -> under contract
engagement · soft | 16 conversations across 10 days (single anecdote) | Single illustrative user (n=1), explicitly qualitative depth 'not revenue'. No cohort size, conversion delta, or revenue → unanchored, cannot size. | | |
Assumptions: Earnings basis: consensus uses ADJUSTED EPS/NI (FY2025 adj NI ~$421.6M / EPS $1.659, 6 analysts), NOT GAAP (NI $23M, EPS $0.0905; OpInc −$32M) — GAAP is distorted, so any EPS% off it is a meaningless thin-denominator artifact (per GAAP/thin-margin guardrail). Default tax 21% (after_tax=saving×0.79) and incremental margin = current net margin WOULD apply to any $ claim, but no claim is dollar-anchored, so no default arithmetic was executed. Phasing: next-FY=FY2026; only the FY2026-attributable portion of any $ claim would be taken. All disclosed figures (5% reach, 40% code, 7M messages, >10% listings) are unit/usage/output/coverage metrics with NO accompanying ARPU, per-unit price, headcount, or cost base anywhere in the inputs — none clears the disclosed-base bar for conversion to $. No supplier-side AI revenue claims; all metrics adopter (own product/ops).
Top line: All topline AI signals are adoption/coverage/volume metrics with no monetization anchor: AI mode at ~5% audience, >7M Follow Up Boss smart messages (FY25), >10% of new listings with 3D/rich media. None carries a disclosed revenue-per-user, per-message price, or take-rate, and mgmt only says AI-mode users are 'expected to be higher-converting' (forward, unquantified). Sum of hard adopter rev_uplift_pct = 0 anchored points. Consensus already bakes +15.3% revenue growth into FY2026 ($2,583M -> $2,978M); these initiatives are plausibly the engine behind it but cannot be isolated as incremental, so est_rev_uplift_pct = null (real but unsizable, not zero).
Bottom line: The only bottom-line-typed claim — '40% more code per engineer at same/higher quality' — is an output/productivity gain, not a disclosed dollar opex cut or headcount reduction, and no engineering cost base is given anywhere. The GAAP base is loss-making at the operating line (−$32M) with GAAP NI just $23M vs consensus adj NI ~$421.6M, so any EPS% off the GAAP denominator would be a thin-base artifact. Per the earnings-basis and thin-margin guardrails, est_eps_uplift_pct = null.
[impact n/m (all claims soft/unanchored)] FY2025 actual revenue $2,583M ≈ consensus $2,579M (Δ +0.16%). Consensus already embeds FY2026 revenue $2,978M (+15.3% YoY) and adj EPS $2.277 (+37.3% vs FY2025 $1.659) / adj NI ~$579M (vs ~$422M), extending toward $3,365M / $2.988 EPS by FY2027 — a large earnings step-up vs GAAP NI $23M. AI is management's stated growth engine, but none of the six disclosed AI metrics translates into incremental dollars or an anchored next-FY %, so the AI uplift is embedded in that consensus trajectory rather than additive to it. The story is directionally bullish but already inside the numbers, with no quantifiable gap above consensus to underwrite until AI-mode monetization (conversion to agents/loan officers) is disclosed with a take-rate.
MODEL CONSENSUS (impact)
partial
Full agreement on all math (6 null/soft adopter claims), null aggregates, priced_in=high; differed only on verdict label and confidence.
Conflicts reconciled
- vs_analyst_expectations: X=unclear vs Y=inline -> used inline because Y cites FY2025 actual $2,583M ≈ consensus $2,579M (Δ+0.16%), better-grounded than X's bare 'unclear'
- confidence: X=3 vs Y=2 -> used 2 (lower per tie-break on category disagreement)
| Field | Opus 4.8 | GPT-5.5 |
|---|
| Rev uplift % | – | – |
| EPS uplift % | – | – |
| Priced in | high | – |
| vs analysts | unclear | – |
| Confidence | 3 | – |
| Top line | All topline AI signals are adoption/reach metrics with no monetization anchor: AI mode at ~5% audience rollout, >7M Follow Up Boss smart messages (FY25), >10% of new listings with 3D/rich media. None carries a disclosed revenue-per-user, per-message price, or take-rate, and management itself only says AI-mode users are 'expected to be higher-converting' (forward, unquantified). Consensus already bakes +15.3% revenue growth into FY2026 ($2,583M -> $2,978M); these initiatives are plausibly the engine behind it but cannot be isolated as incremental, so est_rev_uplift_pct = null (real but unsizable, not zero). | – |
| Bottom line | The only bottom-line claim — '40% more code per engineer at same/higher quality' — is an output/productivity gain, not a disclosed dollar opex cut or headcount reduction, and no engineering cost base is given anywhere. Separately, the GAAP base is loss-making at the operating line (-$32M) with GAAP NI of just $23M vs. consensus adjusted NI ~$421.6M, so any EPS% off the GAAP denominator would be a thin-base artifact. Per the earnings-basis and thin-margin guardrails, est_eps_uplift_pct = null. | – |
| Reasoning | Consensus FY2026 already assumes +15.3% revenue ($2,583M -> $2,977.95M) and +37.3% adjusted EPS ($1.659 -> $2.277; adj NI $421.6M -> $578.7M), extending to $3,365M / $2.988 EPS by FY2027. AI is management's stated growth engine, but not one claim is translated into incremental dollars, so the AI uplift is embedded in that consensus trajectory rather than additive to it — there is no quantifiable gap above consensus to underwrite. Until AI mode monetization (conversion to agents/loan officers) is disclosed with a take-rate, the AI story is directionally bullish but already inside the numbers. | – |
Rows highlighted where the two models disagreed.
QUANTIFICATIONS
AI mode audience rollout: ~5% of audience (millions of users) (Q1 FY2026, current, topline)
“This new way of engaging throughout our site is live for about 5% of our audience so far, which equates to availability for millions of users, and we plan to expand access this year as we continue to test, learn and refine the experience, consistent with how we approach all major product rollouts.”
Engineering productivity (internal AI): 40% more code per engineer (Q1 FY2026, current, bottomline)
“Our engineers are shipping 40% more code per engineer at the same or higher quality.”
Product/design cycle time (internal AI): Concept to launch in a matter of days (Q1 FY2026, current, both)
“Product and design teams are prototyping faster and taking features from concept to launch in a matter of days, and our employees are using AI to reinvent and streamline their workflows.”
Follow Up Boss AI-powered smart messages sent: More than 7 million (Full year 2025, topline)
“In 2025, Follow Up Boss smart messages scaled from a small pilot to a nationwide feature with agents sending more than 7 million of these AI-powered messages.”
Rich media listings feeding AI training/content moat: More than 10% of new For Sale listings include Zillow 3D home tours and interactive floor plans (Q1 FY2026, current, both)
“More than 10% of new For Sale listings on Zillow today include Zillow 3D home tours and interactive floor plans, and we expect rich media to become the standard that buyers and sellers demand for every listing. That substantial growing coverage is already making the consumer experience better. And over time, it will help make our AI more capable.”
AI mode user engagement example (qualitative depth, not revenue): 16 conversations across 10 days (Recent example cited Q1 FY2026, topline)
“a recent AI mode user had 16 conversations across 10 days researching neighborhoods in Sonoma County, California, comparing areas, tracking sold properties, asking for shareable maps to discuss with a partner and referencing their agent in Santa Rosa. They are now under contract to buy one of the homes they found through this robust experience.”
PAST (realized)
- Q4 FY2025 — Wacksman: 20 years of advanced tech including Neural Zestimate (machine learning), natural language search, personalized discovery, workflow automation, and generative AI.
- Q4 FY2025 — Wacksman: Follow Up Boss smart messages scaled nationwide in 2025; agents sent more than 7 million AI-powered messages.
- Q1 FY2026 — Wacksman: 20-year arc from Zestimate through mobile, computer vision, and beyond.
CURRENT (now)
- Q1 FY2026 — Wacksman: Consumer 'AI mode' live for ~5% of audience (millions of users); early signals show deeper conversations and more actionable engagement.
- Q1 FY2026 — Wacksman: Embedding AI across agent/loan officer tools; Follow Up Boss becoming an AI-powered workflow engine for coordination, prioritization, and outreach.
- Q1 FY2026 — Wacksman: Rentals AI powering personalized renter search; property managers using AI Assist for lead management, screening, and lease coordination.
- Q1 FY2026 — Wacksman: Internal AI-native ops — engineers shipping 40% more code per engineer; product/design prototyping from concept to launch in days.
- Q4 FY2025 — Wacksman: Zillow Pro beta includes AI-powered follow-up tools; bringing more AI and insights to non-Zillow client databases.
FORWARD (guidance)
- Q1 FY2026 — Wacksman: Plan to expand AI mode access this year while testing and refining.
- Q1 FY2026 — Wacksman: Rich media coverage (3D tours, floor plans) will over time make AI more capable.
- Q1 FY2026 — Wacksman: AI mode users expected to be higher-converting when reaching agents/loan officers; goal to build the best AI mode experience.
- Q1 FY2026 — Wacksman: Positioned to lead real estate in the AI era across products, professional tools, and how the company builds.
- Q1 FY2026 — Hofmann: Mortgage origination 'purpose-built for AI applications over time.'
- Q4 FY2025 — Wacksman: AI will make the vertically integrated transaction experience better — elevating professionals and removing busywork to convert more transactions.
TRACK RECORD — PROMISE vs DELIVERY
76/100 track record mixed 6 calls reviewed
Zillow makes relatively few quantified, dated AI claims; it beat the Enhanced Markets >35%-by-end-2025 integration milestone but has not yet delivered its flagship AI Showcase 5–10% listing penetration or linked revenue targets, while newer GenAI rollout goals remain too early to score.
AI-powered Zillow Showcase on 5–10% of all U.S. listings (intermediate term) — promised Q4 FY2024
partial Penetration rose from 1.7% at end-2024 to 4.3% of new listings in Q1 FY2026; management still cites the 5–10% goal but has not entered the stated range.
$150M–$300M Showcase revenue opportunity at 5–10% listing penetration (intermediate term) — promised Q4 FY2024
too-early Later calls track Showcase adoption and agent ROI (2% price lift, 30% more listings won) but do not report revenue in this band; prerequisite penetration not yet reached.
>35% of connections through Enhanced Markets by end of 2025 (AI-enabled Follow Up Boss / integrated workflow) — promised Q4 FY2024
delivered Reached 34% in Q3 FY2025 and 44% in Q4 FY2025, above the 35% target; 49% in Q1 FY2026.
At least 75% of connections in Enhanced Markets (midterm) — promised Q4 FY2024
too-early Repeated as a midterm goal through Q1 FY2026; share was 49% of connections in Q1 FY2026, still well below 75%.
Nationwide rollout of AI-powered Zillow Pro in second half of 2026 — promised Q3 FY2025
too-early Q4 FY2025 and Q1 FY2026 describe beta testing and planned expansion; no nationwide launch metrics yet.
Consumer AI mode live for ~5% of audience with broader access in 2026 — promised Q1 FY2026
too-early Disclosed at March 2026 AI Summit / Q1 call as early rollout with encouraging engagement; expansion still in progress with no later quantified adoption outcome in this transcript set.
PRICED-IN (REFINED)
LOW (room left)Est. revisions flat · Fwd P/E 21.1 · EV/Sales 3.1x
AI claim maps to Residential Revenue, Sales Revenue, Rental Revenue
Over the last six months, grades_historical is unchanged (1 strong buy, 1 buy, 5 holds each month) and lastMonthAvg equals lastQuarterAvg at 63.67, so there is no recent upward revision momentum despite a YoY cut from 72.2. At ~21x next-FY consensus EPS and ~3.1x EV/Sales, valuation is moderate rather than clearly rich for a mature internet name, and the stock trades well below consensus targets. AI-driven efficiency and monetization would most plausibly flow into Residential, Sales, and Rental revenue lines, where consensus has not been ratcheted higher recently, so AI upside does not appear fully priced in.
COVERAGE — ENTHUSIASM TRAJECTORY + CATALYSTS
4Q4 FY20247Q1 FY20257Q2 FY20258Q3 FY20258Q4 FY20259Q1 FY2026
AI enthusiasm across 6 calls — trend ↗ rising
AI shifted from light product mentions to named agent tools, ChatGPT and Showcase CV, then summit-backed AI mode and integration thesis.
RECENT AI CATALYSTS & NEWS
BUSINESS IMPACT - QUALITATIVE MATERIALITY
7/10 qualitative impact material medium-term · mixed evidence
Where AI matters: consumer AI mode + agent/rental workflow conversion
Zillow has moved beyond narrative to shipped vertical AI (AI mode for millions at ~5% rollout, nationwide Follow Up Boss smart messages, rentals AI Assist) atop decades of housing-specific ML, but management still cites engagement and productivity—not incremental revenue, margin, or conversion dollars.
Caveats: No anchored AI-to-revenue, take-rate, or conversion lift despite rising disclosure enthusiasm; Consumer AI mode still ~5% rollout with forward monetization claims unquantified; Generic LLM/search alternatives could deflate top-of-funnel traffic even if Zillow adopts AI aggressively; Key AI Showcase penetration and revenue targets remain behind or too early to verify
AI DISRUPTION / CANNIBALIZATION RISK two-sided · 4/10
Horizontal LLMs can commoditize generic listing Q&A and weaken search-dependent discovery, yet Zillow’s moat is proprietary inventory, rich media, regulated local transaction context, and agent/loan workflows that generic chat cannot close—so threat is real at the funnel top but not core to how they monetize connections and SaaS.
OPTIONS / MARKET STRUCTURE
option liquidity: fair
proxy inputs — dollar-ADV $139M · beta 2.042 · px $35.03
source: proxy (no options chain on FMP)
FMP /stable/ exposes no options-chain endpoint on this key, so ATM IV, bid-ask spread and open interest are unavailable. Liquidity below is a PROXY from dollar-ADV, beta and price level (a stand-in for option depth), not measured option-market data.
CONFIRMATION — INSIDERS · 13F · LANGUAGE
Mixed — insiders selling, institutions adding, management language 6/10 measured.
INSIDERS selling 51 open-market sell(s) vs 0 buy(s) — net distribution
INSTITUTIONS (13F) adding as of 2026-03-31: 75 new / 121 closed positions; 266 increased / 167 reduced; institutional ownership -2.08pp; -53 net 13F holders
MGMT LANGUAGE 6/10 measured Firm on internal metrics and 5% live rollout; leadership and scale hedged with belief, testing, and over time.
commit “Our engineers are shipping 40% more code per engineer at the same or higher quality.”
commit “we are seeing more actionable engagement as a result”
commit “This new way of engaging throughout our site is live for about 5% of our audience”
VERBATIM AI QUOTES
“Before I give more detailed updates on what's driving our results in both for sale and rentals, I want to spend a moment on our company strategy and on how we're using AI to accelerate it.”
— Jeremy Wacksman, Q1 FY2026
“We have been building advanced technology in residential real estate for 20 years from the Zestimate to the mobile revolution to computer vision and beyond.”
— Jeremy Wacksman, Q1 FY2026
“And over time, it will help make our AI more capable.”
— Jeremy Wacksman, Q1 FY2026
“That full advantage is what allows us to do more than answer generic listing questions. We can answer personalized questions, anticipate what a consumer needs to do next and actually help them take that action.”
— Jeremy Wacksman, Q1 FY2026
“At our AI Summit last month, we also debuted Zillow's new consumer-facing AI mode experience.”
— Jeremy Wacksman, Q1 FY2026
“This new way of engaging throughout our site is live for about 5% of our audience so far, which equates to availability for millions of users, and we plan to expand access this year as we continue to test, learn and refine the experience, consistent with how we approach all major product rollouts.”
— Jeremy Wacksman, Q1 FY2026
“Early signals are encouraging. Zillow users in AI mode are having deeper, more substantive conversations than they do in traditional search, and we are seeing more actionable engagement as a result.”
— Jeremy Wacksman, Q1 FY2026
“We are also empowering the professional at every step, embedding AI throughout the agent and loan officer experience to help them better serve customers and work more efficiently.”
— Jeremy Wacksman, Q1 FY2026
“Follow Up Boss, which top agents in the country rely on to manage their businesses, is becoming an AI-powered workflow engine that handles coordination, prioritization and outreach, so agents can stay focused on the judgment, advocacy, trust and human relationships that get deals done.”
— Jeremy Wacksman, Q1 FY2026
“For renters, it's powering more personalized search and helping surface the right next step, whether that's scheduling a tour, submitting an application or understanding financial readiness for a future home purchase.”
— Jeremy Wacksman, Q1 FY2026
“For property managers using AI Assist, it's streamlining lead management, application screening and lease coordination, reducing friction at every step of the transaction.”
— Jeremy Wacksman, Q1 FY2026
“We are rapidly becoming an AI-native company.”
— Jeremy Wacksman, Q1 FY2026
“Our engineers are shipping 40% more code per engineer at the same or higher quality.”
— Jeremy Wacksman, Q1 FY2026
“Product and design teams are prototyping faster and taking features from concept to launch in a matter of days, and our employees are using AI to reinvent and streamline their workflows.”
— Jeremy Wacksman, Q1 FY2026
“Now we are powered by AI across every layer of our company in our products, in our professional tools and in how we build.”
— Jeremy Wacksman, Q1 FY2026
“All that depth of capability positions Zillow to lead real estate in the AI era.”
— Jeremy Wacksman, Q1 FY2026
“what we're seeing is that they're having deeper, more substantive conversations than they do in traditional search.”
— Jeremy Wacksman, Q1 FY2026
“for the transaction, that's a higher intent customer at the end of the day, right? It's keeping that person on Zillow. That person is getting more value from Zillow and those folks find more of their needs from our platform. And then when they reach out to talk to a loan officer or an agent, they're ultimately going to be a higher converting customer.”
— Jeremy Wacksman, Q1 FY2026
“the structural advantage we have here over the long term is to build the best AI mode experience because of the content we have on our platform, the context of all those consumers and professionals using our platform and the fact that we span the full gamut of the platform.”
— Jeremy Wacksman, Q1 FY2026
“AI mode is just the start of our ability to take AI in the hands of consumers and really help them with that in an AI-native way.”
— Jeremy Wacksman, Q1 FY2026
“it also helps us build the most capable AI platform so that as we all begin to expect AI-enabled services in our verticals, you will expect the Zillow One to be the one that understands you the best, that helps supercharge the professionals the best and that helps provide the most friction-free and delightful experience.”
— Jeremy Wacksman, Q1 FY2026
“we think mortgage origination is purpose-built for AI applications over time, too.”
— Jeremy Hofmann, Q1 FY2026
“That combination makes real estate an especially difficult end vertical for general purpose AI to disrupt.”
— Jeremy Wacksman, Q4 FY2025
“Zillow has been applying advanced technology in this category for 20 years. From natural language search and the Neural Zestimate powered by machine learning to personalized discovery, to rich media and virtual touring to workflow automation and coordination and now generative AI.”
— Jeremy Wacksman, Q4 FY2025
“In 2025, Follow Up Boss smart messages scaled from a small pilot to a nationwide feature with agents sending more than 7 million of these AI-powered messages.”
— Jeremy Wacksman, Q4 FY2025
“And when AI comes into the category, we think it's going to help make that vertical experience even better, right? Elevating the professionals, taking away the busy works so they can convert more transactions, so they can be more efficient and delighting our shared customers.”
— Jeremy Wacksman, Q4 FY2025
“It includes all of our AI-powered follow-up tools, and it becomes a pathway to Zillow Preferred, right?”
— Jeremy Wacksman, Q4 FY2025
“But I think equally exciting will be for existing Follow Up Boss customers, it's going to help increase the usage and the efficacy of Follow Up Boss because we're going to be bringing more AI and more insights to the part of their business that is not Zillow customers.”
— Jeremy Wacksman, Q4 FY2025
ANALYST QUESTIONS ON AI
Q (Q4 FY2025, Nicholas Jones (BNP Paribas)): You had a great slide on your positioning for vertically-integrated AI. You're partnered with OpenAI today. Can you talk about how you see kind of a vertical AI future and Zillow's place in it, to the extent you can kind of share what you're excited about?
A: Wacksman: Real estate is highly regulated, local, and spans months/years — AI is an ingredient in Zillow's vertically integrated transaction platform, not a threat. Zillow's proprietary assets, industry software (ShowingTime, Follow Up Boss), transaction workflows, and professional trust position it to use AI to elevate professionals, remove busywork, improve efficiency, and delight shared customers.
Q (Q1 FY2026, Ronald Josey (Citi)): AI mode — live for 5% of the audience today. Just talk to lessons learned thus far and thoughts about rolling this out more broadly.
A: Wacksman: Still early at 5%, but millions of users provide strong signal. Users have deeper, more substantive conversations than traditional search; keeps users on Zillow, surfaces more needs, and should yield higher-intent, higher-converting customers when they reach agents/loan officers. Structural advantage is content + context + full-transaction coverage; AI mode is the start of consumer-facing, AI-native guidance through the housing journey.
Q (Q1 FY2026, Nicholas Jones (BNP Paribas)): As you build this end-to-end stack, are you getting more visibility into consumer behavior… and getting better and better data…?
A: Wacksman: Yes — decade-long down-funnel strategy (pre-approval, tours, messaging, offers, closing) plus Follow Up Boss on the agent side creates unparalleled transaction context that helps build a better platform and 'the most capable AI platform.' Hofmann: Solving renter+buyer inventory in one place builds context across multiple transactions over time.
Q (Q1 FY2026, Bradley Erickson (RBC) — mortgage margin question, AI in answer): Mortgage is performing well but lower margin — how should investors ascribe credit and think about longer-term margin potential?
A: Hofmann: Margins improving with better loan officer and processor productivity (11% LO productivity gain last year despite 40% headcount growth). Long-term purchase-mortgage opportunity is large; as scale builds, margin dollars are compelling given low CAC, and 'mortgage origination is purpose-built for AI applications over time, too.'