Keystone · The AI-Native Acquisition Engine

Acquisition gets more expensive as you scale.
That's not physics. It's amnesia.

Keystone is your acquisition engine, built on your own first-party data. It learns exactly who pays you, finds everyone who mirrors them, and spends only on the ones who will — so your cost to acquire holds profitable as you grow, and sharpens every cycle instead of starting over.

289M / 6M
people & companies in the identity graph your model runs across
~$500K
the AI hire you'd otherwise need to do this in-house (benchmark)
90 days
guaranteed pilot against a win condition you set up front

Your acquisition has no memory.

Every cycle teaches you the same lesson — who pays you and who never will. You pay full price to learn it, throw it away, and pay to be taught it again next quarter. Nothing compounds.

That's the enemy: activity without memory. It's why your CAC climbs exactly when you're trying to scale. More spend, more reach, same forgetting.

Running ads to people who'll never buy isn't a budget problem. It's a memory problem.

  • 01 You pay to learn who pays you — then delete it. Every won and lost deal is a lesson. Without a system to keep it, you re-buy the same lesson every cycle.
  • 02 You're spending on red. If your winners are green and your budget is reaching red, your CAC climbs by design. You're funding people who were never going to buy.
  • 03 More volume makes it worse. Pouring volume into the wrong people raises CAC — it doesn't lower it. The fix isn't more reach. It's better aim.
  • 04 Your agency optimizes clicks, not buyers. They run activity without memory, never model who actually pays, and nothing they build is yours.
  • 05 Doing nothing isn't free. The status quo is the expensive option. Every month you run on amnesia, the gap to a competitor whose engine is learning widens on its own.

Acquisition isn't a volume problem.
It's a green-from-red problem.

That's why we don't start with channels. We start with who actually pays you.

We tell green from red — at scale.

We model your real winners against your real losses and learn the pattern that separates who pays from who never will. Then we run that pattern across the whole market.

Most of your market is red — they were never going to buy. A slice is green — they mirror the people already paying you. We only spend on green. That's the entire reason your CAC can hold profitable while you scale.

Click through the panel. It's the same split we'll show you on your own customers — before you commit a dollar of media.

See the split on your data
Green — mirrors your winners Red — never going to buy

Is your acquisition amnesiac — or an engine?

Answer 8 quick questions and get your Acquisition Engine Readiness score: where you land from Amnesiac to a compounding Keystone engine, plus the single highest-leverage move to hold your CAC as you scale. Instant results.

Take the 2-minute assessment

Keystone: a populated, running engine — built on your truth, and yours to keep.

The technology is commodity. Keystone is the asset — your data, modeled and resolved against a 289M-person identity graph, run as a living system by a roster of agents like MarketWriter, its owned-email writer. Here's the pipeline Keystone builds for every client.

Step 01
Onboard your truth
Every winner, every loss, plus the calls, emails, and notes behind them. The model learns from your reality, not a generic playbook.
Step 02
Your vector database
All of it embedded into a living, machine-readable database of your product and your buyers. Yours, exportable.
Step 03
Win/loss model
We model your winners against your losses to learn the pattern that separates who pays from who doesn't. Green from red.
Step 04
Your seed pool
A growing library of confirmed winners and losers that trains the model — and gets sharper with every new win.
Step 05
Your TAM, across the graph
We run the model across 289M people and 6M companies to build the full universe of prospects who mirror your winners. Not "the industry" — people who look like the ones already paying you.
Step 06
SiteVisitor + InMarket
We resolve the identity of your site traffic and surface who's showing buying behavior right now — both filtered against your TAM, so we only ever work real, in-profile people.

Why it gets cheaper instead of more expensive.

Every win Keystone produces drops back into your seed pool. A bigger, cleaner pool sharpens the model. A sharper model builds a better TAM and better targeting. Better targeting produces more wins — which drop back into the pool.

Round and round, and every cycle the whole thing is more accurate and cheaper to run. That's the difference between us and everyone else: they do, we learn.

On the Full program, two channels — paid and owned email — feed the same pool, so a win on one makes the other smarter too. One compounding asset, cross-feeding itself.

You leave with the asset. Stated plainly.

Yours — forever

The asset you own and can export

  • Your first-party data, onboarded and structured
  • Your embeddings and vector database
  • Your win/loss model — the pattern of who pays you
  • Your seed pool of confirmed winners and losers

First-party, exportable, isolated to you. If you ever leave, you keep all of it.

What you rent

The match that turns a model into a market

  • The match against our 289M-person identity graph
  • Identity resolution + buying-intent infrastructure
  • Ongoing optimization and the running of the engine

No data lock-in. The moat is the populated, compounding instance plus the match — never holding your data hostage.

You've got four other options. We beat all of them.

The optionWhat you actually getWhy the engine wins
Your agency Activity without memory. They optimize clicks, never model who pays, and nothing they build is yours. We model who pays and keep it — the asset is yours.
Point tools
Clay, 6sense, Apollo, RB2B, Warmly
The parts — enrichment, intent, de-anonymization — handed over as an empty box to assemble and maintain. We deliver the populated, running version — and operate it.
In-house build Hiring data and ops talent to do in months what we do in weeks — and you still start from zero on the data. Live in weeks, already knowing your winners. No ~$500K hire.
Do nothing Keep spending into rising CAC. The status quo is the expensive option, and the gap widens every month. A 90-day pilot tests it now, low-risk, against your win.

The model learns from your truth. So we're honest about fit.

Built for you if…

  • You're US-based B2B doing $10M+ with a real sales (or B2B-ecom order) motion
  • You have a paper trail to learn from — roughly 50+ wins and 50+ losses, plus the conversations behind them
  • You have a defined offer and a CRM we can integrate with
  • Leadership is ready to act on what the model finds — including pointing spend away from loser-like segments

Not for you if…

  • You're pre-revenue or sub-$10M, still finding product-market fit
  • You have a handful of deals and no record of the conversations behind them
  • You want a one-off campaign, not a compounding asset
  • You won't let the data move the budget — or you're pure B2C

Same engine. The only difference is where it activates.

Both programs build the entire moat — data, vector database, model, seed pool, TAM, the compounding loop. One-time build + monthly engine. Media spend is yours, billed separately.

Starter
From $5Kper month + a one-time build  ·  media separate

The full Keystone engine, activated on your channels. We feed our model's targeting into your Meta, Google, and LinkedIn. You run the media; we supply the brains and the compounding.

  • First-party data + vector database
  • Win/loss model + seed pool
  • TAM by ICP across the identity graph
  • SiteVisitor + InMarket
  • Compound learning loop
  • CRM integration + reporting
  • Activate your channels — Meta, Google, LinkedIn
  • Owned email + MarketWriter
  • Dual-channel flywheel
Book a fit call

Media spend is yours and billed separately — we never bundle ad spend into our fee.

We take the risk, because we only spend on green.

Every engagement starts as a 90-day pilot against a win condition we set with you up front — booked calls, funded deals, orders, or a target CAC — and exactly how we'll measure it, agreed before we launch.

Hit the win, we continue. Miss it, and we keep running the engine at no monthly fee until you hit it — or you walk with your data, model, and seed pool and owe nothing further.

The clock starts at go-live, not at signature. We can offer this because the whole model only spends on prospects who look like proven winners. If we didn't believe it would hit, we wouldn't take the pilot.

We're early — so we lead with your data, not a logo wall.

6

live engagements today — 3 running, 3 onboarding. We add named results here as clients consent.

Day 0

You see the green/red split on your own winners in the first weeks — before you spend a dollar of media.

~160 bps

EBITDA improvement in 24 months — a transformation benchmark, not a Stibnite result.Industry benchmark

We don't show logos we haven't earned or attribute ROI to ourselves before a real, consented case exists. Benchmarks above are labeled as benchmarks. What we'll show you is the model's read on your data — the most honest proof there is.

You're trusting people to run your acquisition. Here's who.

No account managers. No black box. The two founders build Keystone and run it — the people on your fit call are the people in your engine.

MN
Matt Newbill
Co-founder & CEO
First employee at TSheets — the company Intuit acquired for $340M — where he was on the early growth team and saw, from the inside, what a compounding engine actually looks like. A software engineer by training and a data nerd at heart, Matt has spent 15 years building and running a growth practice, scaling SaaS, D2C, and manufacturing companies. He lives in the unglamorous machinery that makes acquisition compound: win/loss modeling, identity data, and email deliverability. In prior client work, he took pipeline from $0 to $1.5M in under 90 days.
Owns: the model, the data, and the engine — how Keystone gets built and run.
Matt on LinkedIn
TL
Tommy Liantonio
Co-Founder
In identity-based marketing since 2007 — before "identity resolution" and "intent data" were even categories — Tommy has built several identity graphs from the ground up. At ClickCertain, an early real-time-bidding platform, his tech touched tens of billions of daily impressions and ran white-label under other companies' brands; at LiftCertain, it resolved the people behind anonymous web traffic and surfaced them as in-market buyers. That backend quietly powered adtech and martech across 2,000+ agencies. He builds the technology himself and is a founding member of the 5x5 Co-op — and Keystone's SiteVisitor and InMarket, which catch buyers in the act, are the work he has perfected for nearly two decades.
Owns: the relationship and the win condition — your single point of contact, start to finish.
Tommy on LinkedIn

See the green/red split on your own customers.

Book a 30-minute fit call. We'll look at your win/loss data together, tell you straight whether you have enough to model, and if you do, show you who actually pays you — before you commit a dollar of media.

Book a fit call
30-minute call Straight answer on fit Your data stays yours

Not ready to talk? Take the 3-minute assessment and see where your acquisition stands.

What people ask before they reach out.

Keystone is a done-for-you acquisition engine, built on your own first-party data, that learns exactly who pays you and finds more of them across your channels. Instead of running acquisition that resets every cycle, the engine remembers — it models your winners against your losses, only spends on prospects who mirror proven winners, and gets sharper every cycle. The result: your cost to acquire holds profitable as you scale instead of climbing.

Agencies run activity without memory — they optimize clicks, never model who actually pays you, and nothing they build is yours. Point tools (Clay, 6sense, Apollo, RB2B, Warmly) sell you the parts and hand you an empty box to assemble and maintain. We deliver the populated, running engine — your data modeled and resolved against our identity graph — and we operate it for you, tied to a guaranteed win condition.

We need a real paper trail — roughly 50+ closed-won and 50+ closed-lost deals, plus the calls, emails, and notes behind them. The model learns from your truth, so thin data means a thin model. If you have it, we model it. If you don't, we tell you straight on the fit call before you commit anything.

No lock-in. Your first-party data, embeddings, win/loss model, and seed pool are yours — first-party, exportable, and isolated to you. The only thing you rent is the match against our identity graph plus our resolution and intent infrastructure. Leave, and you keep everything that's yours.

Two programs. Starter starts from $5K — the full engine on your own paid channels. Full adds Stibnite-run owned email with MarketWriter (the dual-channel flywheel) and is custom-priced; book a fit call for a number. Media spend is yours and billed separately; we never bundle ad spend into our fee. Both are sold against a guaranteed 90-day pilot.

Every engagement is a 90-day pilot against a win condition we agree on up front. Hit it, we continue. Miss it, we keep running the engine at no monthly fee until you hit it — or you walk with your data, model, and seed pool and owe nothing further. We can offer this because the model only spends on prospects who look like proven winners.

Tell us where your acquisition stands.

Where you are, where you want to be, and whether you've got the data to model. If we're the right fit, you'll know fast. If we're not, we'll tell you that too.

Got it.

We'll review and respond within 24 hours. If we're not the right fit, we'll tell you who is.