THE DATA
The research behind every claim we make.
We don't quote performance numbers we can't defend. Here's the underlying research that shapes how we build.
SECTION 01
The cost of slow response.
21x more likely to qualify when contacted within 5 minutes
Source: InsideSales.com / MIT Lead Response Management Study, Oldroyd
Average B2B response time clusters around 42–47 hours
Source: Drift, InsideSales
78% of buyers work with the first agent who responds
Source: NAR Home Buyers and Sellers Generational Trends Report
51% of leads are never contacted at all
Source: InsideSales Lead Response Research
SECTION 02
Subscription economics.
7–12% of MRR is at risk from failed payments
Source: Recurly, Stripe research
7.2% of subscribers at risk monthly from involuntary churn
Source: Recurly Subscriber Retention Benchmarks (1,200 subscription sites, 2023)
Top performers using AI-driven dunning recover 70–85% of failed payments
Source: Slicker, Baremetrics, Churnkey
20–40% of subscription churn is involuntary and therefore preventable
Source: Churnkey State of Retention 2025
SECTION 03
Industry-specific operational data.
48% of law firms unreachable by phone
Source: Clio 2024 Legal Trends Report — secret shopper study, n=500 firms
Firms using client intake technology see 51% more leads
Source: Clio 2024 Legal Trends Report
Median B2B response time exceeds 24 hours
Source: Drift Conversation Trends Report 2024 (Salesloft)
SECTION 04
Creative effectiveness.
Creative drives 49% of incremental sales lift
Source: NCSolutions, Five Keys to Advertising Effectiveness, 2023, ~450 CPG campaigns
Following creative best practices drives 1.2–7.4× short-term sales lift
Source: Meta + Nepa Marketing Mix Modeling, 2020, n=1,660 campaigns
Ads following Google's ABCD creative framework see +30% short-term sales lift and +17% long-term brand contribution
Source: Google + Kantar Link AI, 2021, 11,000+ ads
Creator content via TikTok Spark Ads drove +159% engagement at the same CPM
Source: TikTok Data Science, Feb 2024–Jan 2025, North America
Sub-3-second viewers drove up to 47% of total campaign value
Source: Facebook + Nielsen BrandEffect, n=173 studies
SECTION 05
Creative cost benchmarks.
Home-services average cost-per-lead $90.92 — Roofing $228.15, HVAC $127.74, Plumbing $129.02
Source: LocaliQ 2025, n=3,211 home-services search campaigns
UGC creator video market rate ~$150–$212 per video
Source: inBeat, 2024
Agency video production runs $1,000–$10,000 per finished minute
Source: Clutch
SECTION 06
Real estate visual performance.
41% of buyers rated photos as the most valuable website feature
Source: NAR, 2024 Profile of Home Buyers and Sellers
83% of buyers' agents said staging helped buyers visualize the home
Source: NAR, 2025 Profile of Home Staging, n=1,266 Realtors
29% of agents reported staging led to a 1–10% increase in dollar value offered
Source: NAR, 2025 Profile of Home Staging
69% of buyers searched for homes on a mobile device
Source: NAR, 2024 Profile of Home Buyers and Sellers
Median professional staging cost was $1,500 in 2025
Source: NAR, 2025 Profile of Home Staging (multi-year trend)
SECTION 07
Automotive visual performance.
Multiple custom photos vs. stock increased used/CPO VDP clicks by 349%
Source: Cox Automotive, Telling Your Value Story Through Photos, 2018
VDP views per listing are 64.8% higher with 40+ photos
Source: Cox Automotive, same study
40% of consumers said they'd buy a vehicle based on online images alone
Source: Cox Automotive, The Power of Pictures Study, 2016/2018
SECTION 08
Professional headshot impact.
LinkedIn profiles with a professional photo receive up to 21x more views, 9x more connection requests, and 36x more messages
Source: LinkedIn platform data
Trustworthiness judgments from faces form in approximately 100 milliseconds
Source: Willis & Todorov, Psychological Science, 2006
SECTION 09
Short-form video & social performance.
Reels make up 50% of time spent on Instagram; users share 3.5 billion Reels daily
Source: Meta platform disclosure, 2024–2025
TikTok's average engagement rate reached 3.70% in 2025 — more than 7x Instagram's 0.48%
Source: Socialinsider 2026 Social Media Benchmark Report, 70M+ brand posts
Maybelline FOOH campaign generated 76.1 million views and approximately 3 million likes
Source: Confirmed by Maybelline VP Marketing Fernando Febres, Marketing Brew interview, 2023
SECTION 10
E-commerce product visuals.
56% of online shoppers' first action on a product page is exploring images — before reading title, description, or any other content
Source: Baymard Institute product page UX research
62% of leading e-commerce sites have mediocre or worse product page UX
Source: Baymard Institute, 2025
U.S. retail returns projected to reach $890 billion in 2024
Source: National Retail Federation / Happy Returns, December 2024
SECTION 11
Med spa & local business marketing.
The average medical spa invests approximately 7% of revenue in marketing
Source: AmSpa 2024 Medical Spa State of the Industry Report
The U.S. medical aesthetics industry exceeded $17 billion in 2023, growing by more than $1 billion per year
Source: AmSpa 2024
97% of consumers read reviews when researching a local business; 41% "always" read reviews in 2026
Source: BrightLocal Local Consumer Review Survey 2026
SECTION 12
AI content & disclosure.
76% of Americans say it is extremely or very important to be able to tell if content was made by AI
Source: Pew Research Center, 2025, n=5,023
68% of consumers globally say brands should disclose AI use
Source: YouGov, published via Statista, January 2024, 17-market survey
SECTION 13
Website presence & local search.
Users form an opinion about a website's visual appeal in 50 milliseconds
Source: Lindgaard, Fernandes, Dudek & Brown, Behaviour & Information Technology, 2006, peer-reviewed, Carleton University controlled experiments
The single most-cited website credibility criterion is design look — named by 46.1% of participants in a 2,684-person study
Source: Stanford Persuasive Technology Lab / Consumer WebWatch, Fogg et al., 2002, n=2,684
When mobile page load goes from 1s to 3s, bounce probability increases 32%; from 1s to 5s, 90%
Source: Google/SOASTA Research, 2017, 900,000+ mobile landing pages, 126 countries
46% of all Google searches have local intent
Source: Google, Secrets of Local Search, 2018
76% of people who search locally on their smartphone visit a related business within 24 hours
Source: Think with Google, 2014/2016
42% of searchers click on a result inside the Google Maps Pack for local queries
Source: Backlinko Local SEO Statistics, 2024
Customers are 2.7× more likely to consider a business reputable when it has a complete Google Business Profile
Source: Google official documentation
62% of consumers would avoid using a business if they found incorrect information about it online
Source: BrightLocal Local Business Discovery & Trust Report, 2023
SOURCING DISCIPLINE
Stats we don’t believe — and why we won’t cite them.
Most agencies in our category sell on three numbers: 917 minutes, 47.6%, and ‘30-50% of leads are lost.’ We won’t cite any of them, and here’s why.
The 917-minute real estate response time. This number is everywhere — usually attributed to ‘Inman 2025.’ It isn’t. The original is a 2014 WAV Group / Weichert Lead Network study using a sample of real estate inquiries from a previous decade. The market has changed twice since. We use Drift’s Conversation Trends Report 2024 (median B2B response over 24 hours) and Oldroyd’s MIT/InsideSales.com Lead Response Management Study (2007) — and we tell you the second one is the 2007 source so you can decide whether to weight it.
47.6% recovery without AI dunning. This figure circulates as a Recurly benchmark. The closest verifiable Recurly figure is the 7.2% monthly subscriber-loss exposure from involuntary churn (Subscriber Retention Benchmarks, 1,200 sites, 2023). When we model your dunning lift, we’ll use Recurly’s published industry benchmarks, not the 47.6% number.
30-50% of inbound leads lost in roofing. This is repeated on a hundred vendor blogs and traces to no published contractor survey. Invoca’s platform data (1,200+ home-services contractors) puts the number at 27% of calls unanswered, with fewer than 3% leaving voicemail. That’s the number we work from.
The Mordor Intelligence ‘$43B disaster restoration market.’ This conflates global figures with adjacent categories (debris removal, environmental remediation). The IBISWorld U.S.-specific figure is $7.2B with 60,020 establishments (April 2025 update). We use the U.S.-specific number when we model U.S. operators.
The “test N creatives per week” number. You’ll see vendors promise a specific weekly creative count as if it were the optimal cadence — three a week, five, seven. No rigorous published source defines that number for the verticals we work in, so we won’t cite one. We describe the work as diversity of angles per cycle, scoped to the engagement, and report what shipped.
The ‘70% of med spa patients find providers through social media’ claim. This stat circulates widely in med-spa marketing content but does not appear in any AmSpa publication or independently sourced dataset. We won’t cite it.
The ‘75% of users judge credibility by website design’ stat. This figure circulates widely across marketing blogs attributed to Stanford. The actual Stanford Web Credibility Project (Fogg et al., 2002, n=2,684) found 46.1% cited design look — the largest single factor, but not 75%. We use the real number.
If we publish a number, you can click through to the primary source. If we can’t link to a primary source, we don’t publish the number.
ARCHITECTURE
How a Brensa engagement is built — websites, ads, and AI.
A master orchestrator agent routes to specialized worker sub-agents based on task type. All sub-agents share state via a central data layer. All actions flow to an observability layer. Your CRM is the single source of truth.
Master
Orchestrator Agent
Voice agent
DM agent
Dunning agent
Outbound caller
Booking agent
Support agent
State
Shared data layer
Observability
Action log
Truth
Customer CRM
Why sub-agent architecture beats monolithic agents.
Most AI agencies build one giant “do everything” agent. We don't. Specialized sub-agents are more reliable, easier to debug, easier to update, and produce better outcomes per task. A voice agent and a dunning agent have completely different skill requirements, and forcing one model to do both produces mediocre results in both.