Meta + Google
High-frequency. Short-attention. Direct-response. Excellent for the sale in front of you. Structurally bad at building the relationship that brings someone back.
01 · The Meeting
Three Things Boost Wants From This Room
01
Lower Cost Per Acquisition
02
Grow Sales
03
Build Brand That Lasts
Here's where the market already is — most competitors are cycling high creative volume at low cost per unit. Boost can compete on cost AND go somewhere they can't: retention.
Creatives in market · iSpot.tv brand scorecards · 30-day window 07/11/2026–08/10/2026
#5 / #8
Verizon / T-Mobile spend rank
#263
Boost spend rank
#570
Mint spend rank
14,848
T-Mobile national airings (30d)
Boost airings and all dollar spend figures are gated behind an iSpot enterprise seat. Ranks shown are directional. Source: iSpot.tv brand scorecards, data available through 08/10/2026.
Efficiency curve · airings per creative
4.7
Verizon
14.0
T-Mobile
13.2
AT&T
12.9
Metro
58.0
Mint
Boost unknown (airings gated). Computed: national airings ÷ creatives in market.
Mint runs 58 airings per creative on a #570 spend rank and still out-airs Verizon. They are not buying attention. They are earning it, then buying cheaply.
02 · The White Space
Where Brand Retention Actually Gets Built
TV + CTV + YouTube
20+ minutes of sustained attention. This is where retention and brand recall actually form. This is the underused lane.
Creator and long-form social content are a related, real opportunity, but outside the scope of this conversation.
This is the lane Vision Bridge and Perlocity are built to find and fill for Boost.
Directional framing based on category attention patterns, not a proprietary Boost audience study. Not a claim about Boost's specific customer data.
03 · The Unlock
One Closed Loop, Not Two Separate Systems
01Attribution finds the opportunity
02Perlocity tests AI-generated :15s/:30s inside it
03Holdout measurement reads the lift
04The winning signal feeds back into media
Vision Bridge and Perlocity aren't two vendor pitches. They're one system that gets smarter with every test.
04 · Proof, Not Promise
We've Run This Exact Play Before
Case study from a separate Vision Bridge engagement. Shown to prove the method — not Boost's numbers, not a guarantee.
37 test states vs. 13 holdout states
Clean geo-isolated holdout methodology
46 days · $139K spend
2.66x Per Capita iROAS at 99.7% confidence
90 days · $234K spend
~2x the spend and days of phase 1. Signal held: 1.59x Platform ROAS sustained every month.
12 days · $31K spend
Previously-dark holdout markets activated and immediately produced 1.52x ROAS — “they were thirsty for it.”
1,556 conversions
Vision Bridge validated a real ~4.5 day consideration cycle sitting inside a 7-day attribution window.
$250K–$300K/month modeled national CTV investment
Projected $4.35M–$5.22M in annual incremental revenue, up to $6.53M including a returning-customer retention halo.
This is the same holdout-and-attribution discipline we'd bring to Boost's white space.
05 · How It Feels
We Tag the Brain State Before We Ship the Creative
Every concept gets tagged for which nervous-system state it activates before it goes into test — prefrontal cortex, parasympathetic “calm and connect” vs. amygdala-driven “fight or flight.” We're not guessing which creative feels safe to a viewer. We measure for it.
Observed: category-typical opening frames, rapid skip
Salience & prediction error
+The brain allocates attention to violated expectation. Creative instruction: the first sub-second artifact must be unpredictable within its own placement context. A category-typical opening frame is discarded pre-consciously.
Observed: Mint's fixed face + joke structure
Processing fluency
+Repeated, easily-processed structures feel truer and safer. This is why a recurring character and a fixed format outperform a rotating library. It is the mechanism behind Mint's moat.
Observed: bill creep, teaser rates, fee stacking
Loss aversion
+Losses loom larger than equivalent gains. "Stop paying for what you don't use" outperforms "save money." The bill-creep territory is built on this.
Observed: prepaid status penalty in culture
Identity signaling
+People buy what confirms who they believe they are. Prepaid carries a residual status penalty; the creative job is to convert "what I can afford" into "what I'm too smart to overpay for."
Observed: 3,000 neighborhood store presence
Social proof & in-group
+3,000 stores means the brand is physically present in a neighborhood. Localized proof beats national proof.
Observed: plan grids, feature matrices, fee footnotes
Cognitive load & the decision cliff
+Every additional plan variable reduces conversion. "One price for life" is a load-reduction claim before it is a price claim.
Observed: "forever" pricing sold as future benefit
Temporal discounting
+Immediate certain benefit beats larger delayed benefit. This is why "forever" pricing must be dramatized in the present tense.
Observed: ads, store visits, first bills forgotten mid-flow
Peak-end rule
+Memory of an experience is dominated by its peak and its end. Applies to the ad, the store visit, and the first bill.
Tribe v2 · Character Casting for Boost
Villain
The contract
The contract-locking mega-carrier and the bill that grows while you sleep. Punch the mechanism, not the competitor's customers.
Hero
The customer
The person who took control of their telecom bill. Agency restored.
Sidekick
Sparky + the arithmetic
Sparky, and the arithmetic of a price that does not move. (Text treatment only — no fabricated mark.)
The villain is the contract, not the competitor's customers — punching at the mechanism, not at people, is what keeps it likable at high frequency.
These are applied findings from attention and behavioral decision research used as creative constraints. They are heuristics for making work more memorable, not deterministic levers. Every one of them still has to win in market against a holdout.
06 · The Brain
The Brain — Boost Encoder
Signal encoding → cross-modal fusion → high-resolution catchment prediction. Every holdout test, every neuro-tag, and every attribution signal feeds back into this. This is what gets smarter. This is what compounds. Click any node.
CLOSED LOOPExposure → household → catchment → activation
GOVERNANCEConsent gates every rung; brand rules never leave the boundary
OUTPUTCreative conditioned on the lever the signal licenses
07 · The Buying Layer
A Feedback Loop the Models Can Actually Learn From
Boost wants to buy on a CPA basis without cannibalizing other media. The mechanism is agentic buying — negotiating directly with publishers at scale, algorithmically, so every purchase feeds a precision signal back to the model. Vision Bridge is built to do exactly this.
Not cannibalization, negotiation.
Agentic buying negotiates directly with publishers on a CPA basis, so gains don't come at the expense of other channels' performance.
A precision feedback loop.
Every buy teaches the model precisely how and when to buy next — this compounds independently of any single test's dollar figure.
Portable by design.
What gets built and proven in this test isn't locked to one platform. It's an intelligence layer that can hand off to whatever advanced AI system Boost adopts next — a real head start, not a locked-in dependency.
Every test run compounds. Twelve months in, most competitors are still buying manually.
This test isn't just a CPA number. It's the start of an intelligence layer Boost owns.
Positioning claim about Vision Bridge's current capability, not a Boost-specific performance guarantee. Node diagram is illustrative of the buying/feedback mechanism, not a literal system architecture diagram.
08 · The Ask
Answer the Three Goals With One Pilot
Vision Bridge cleans up cross-channel attribution and puts spend behind a guaranteed CPA threshold. The full savings model is in Supporting Evidence.
Use attribution to find the white space, test :15s and :30s against clean holdouts, then feed the winning signal back into media allocation.
Tag creative for retention-friendly “calm and connect” states before it ships, then prove which work drives both business outcome and memory.
Three questions to shape the pilot
- What is your true guaranteed CPA target?
- Which TV, CTV, and YouTube inventory should enter the first white-space test?
- Can we get read access to store-level activation data for a single-market pilot?
Q3 pilot shape
Structural, not invented
Single market · one owned device · N creatives · guaranteed CPA · store-level readout. Reference allocation already on the table: $300,000 Q3 test. No additional dollar figures invented here.
Supporting Evidence
Appendix A · Vision Bridge
Vision Bridge — Guaranteed CPA
A dedicated, performance-driven test mapped entirely to a guaranteed cost per acquisition. 100% performance-based — spend governed by the guaranteed CPA threshold.
$300,000
Reference scope · Q3 test allocation
- RudderStackTagging deployed via RudderStack, establishing server-to-server data flow that bypasses client-side pixel restrictions so every acquisition is attributed in real time.
- Google CloudData infrastructure managed on the Myosin side and integrated back into Boost's Google Cloud environment via unified API. Closed-loop reporting; acquisitions verified automatically with no manual reconciliation..
- Creative delivery4–5 live creatives concurrently to prevent fatigue while giving budget density to optimize; :15 and :30 primary; every asset carries baseline URLs, dedicated QR codes, and short codes for cross-device continuity.
We provide the pipeline. You provide the endpoint. Real-time CPA validation directly into your existing data stack.
Savings & scale model · arithmetic only
Two inputs. Everything below recomputes. We assert none of your figures.
—
Acquisitions / month at guarantee
—
Annualized acquisitions
—
More / year than a $150 blended CPA buys
—
$115 is an opening ceiling we expect to drive down as the graph closes the loop — step it toward your own target. $150 blended CPA is a placeholder reference, not your number. Supply line assumes 4–5 live creatives per $100K/mo refreshed every 2 weeks, at modeled production costs of $8,000 (traditional) vs $45 (Perlocity) per finished asset. Under a guarantee, variance sits with Myosin, not Boost. Modeled illustration. Not a quote.
Guaranteed CPA is only affordable to underwrite if creative supply is cheap and the identity graph closes the loop. The engine is what makes the guarantee underwritable. That is why the whole deck hangs together.
Appendix B · 3,000 Stores
Boost's 3,000 Stores
The layer nobody else in the category has at this granularity for prepaid.
Every dot is a first-party sensor feeding the identity graph: consent-gated resolution runs from deterministic account keys down to contextual signals, and joins each activation to a household and a store catchment (section 04).
Real US Census state geography (Albers USA projection). 3,000 modeled points across 77 real metro areas, weighted by 2020 census metro population — not Boost's actual store addresses.
Store as sensor
Every activation, port-in, and upgrade is a first-party conversion event with a physical location and a timestamp.
Store as creative unit
Catchment-level creative variants — local device mix, competitive density, language mix, payday cycle. Generation makes per-catchment creative economically possible for the first time.
Store as proof
Closed-loop: national exposure → household graph → store catchment → in-store activation → attributed CPA.
Every other carrier measures a market. Boost can measure a block.
Appendix C · Speed & Volume
Speed & Volume Math
Traditional
brief → agency → concept → production → post → legal → traffic
0
6 to 8 weeks per campaign · tens of thousands of dollars per finished asset
Perlocity
brief → generation → swarm critique → brand check → traffic
0
15 to 30 minutes · cents to low dollars per asset
Volume without a fixed brand device is just more noise. The engine only earns its keep because sections 4 and 5 tell it what to hold constant.
Appendix D · Brand Room
The Perlocity Brand Room
Compose a brief from chips. Generate. Watch structured creative stream out offline.
Simulated session. Output shown is representative of live Perlocity generations.
00:00.0
concept → structured output
Select chips, then Generate.
00:00.0
Appendix E · The Stack
The Stack
01
Perlocity
Brand-room ideation, generation, swarm critique, asset library, brand governance.
REPLACES: slow agency concept cycles, unmanaged asset sprawl
02
The Identity Graph
Zero/first/second/third-party resolution, consent-gated, household + store catchment.
REPLACES: pixel-only attribution, market-level guesswork
03
The Neuro Framework
Tribe v2, Character Casting, the applied-lever library.
REPLACES: offer-led creative with no memory device
04
Vision Bridge
Guaranteed-CPA media with closed-loop server-side attribution.
REPLACES: brand spend without acquisition accountability
Boost stops selling the phone and starts selling freedom from the bill, through one owned device, in two languages, at a generation-engine cost structure, measured to a guaranteed CPA at store level.
Spanish-language beachhead
Boost already has meaningful Spanish creative in market and Metro is fighting hard there. Build the campaign world bilingual from concept, not as a translation layer.