Moneyball for B2B: stop buying pretty stats, start buying pipeline
Billy Beane didn’t win by signing shinier players. He won by pricing what the market ignored, using on-base percentage and building a system that compounded small edges.
B2B marketing has the same inefficiency today.
We keep optimising for good-looking numbers (CTR, MQL volume) that barely correlate with revenue.
I believe the 2025 edge goes to teams that buy undervalued signals and run their front office like… a front office.
The thesis
Modern B2B isn’t a battle for impressions; it’s a market for mispriced signals. The brands that win will:
define a win function tied to pipeline,
model expected value before conversions land, and
redeploy budget weekly into the most efficient “positions” (channels/creatives/supply paths).
Everything else is vibes.
1) Define the win function
Clicks don’t pay salaries. Opportunities do. Replace vanity metrics with a single north star:
Expected Pipeline Added (EPA)EPA = Σ_touches [ P(opportunity_in_30_days | touch) × Expected_Deal_Value(account) ]
Time-bound (30, 45, or 60 days—pick one and stick to it).
Calculated at the touch level (impressions, visits, CTV exposures, email opens).
Aggregated to creative / channel / supply path / account so you can trade budget like a GM.
If a tactic doesn’t move EPA, it’s off the roster.
2) Measure what actually moves EPA (your “on-base %”)
Build a catalog of High-Value Actions (HVAs) with clear intent:
Pricing/integration/security page views
Docs/API/implementation guides (new persona joining = buying committee signal)
Repeat visit from a net-new device or channel
Form re-open / meeting reschedule / chat intent
Deep attention (dwell + scroll + recency + repeat)
Then project value ahead of conversion:
xHVA: expected HVAs per 1,000 impressions/visits.
Attention Score: weighted dwell, scroll depth, repeat exposure, form engagement.
Surge: short-term lift vs. trailing baseline at the account level.
You’re now optimising to expected value, not yesterday’s last-click win.
3) Roster construction: channels are positions
Each position has a job. Don’t ask your shortstop to pitch.
CTV & Programmatic Audio → table-setters. High attention, broad reach into the committee, great for account surge.
Native & Quality Display → on-base machines. Cheap xHVA, superb for mid-funnel proof and segmentation.
Search, Direct, Retargeting → closers. Harvest HVAs and book meetings.
Email/Marketing Automation → player development. Turn signals into sequences and opportunities.
Keep a farm system (5–10% of spend) for experimental placements and creative formats. Promote winners quickly; cut underperformers even faster.
4) Buy market inefficiencies (Billy Beane’s edge)
You’re looking for cheap EPA.
Context niches: domain clusters with high attention but unfashionable CPMs.
Daypart/device edges: your ICP reads implementation content on desktop evenings; bid accordingly.
Supply Path Optimization: trim exchanges, prefer clean deals with high match rates and lower hidden fees.
Creative truth > polish: integration logos, quantified ROI, screengrabs of the product doing the thing.
Identity advantage: stronger account graph = better retargeting and exclusion = lower CPEP (see below).
5) Front office metrics (no more pretty averages)
CPEP (Cost per Expected Pipeline)
CPEP = Spend / EPA
Lower is better. Compare by channel, creative, supply path, and account segment.RARA (Revenue Above Replacement Account)
RARA(account) = EPA(account) − EPA(replacement_account_in_segment)
This tells you who deserves more innings (budget) this week.xHVA rate and Attention Score by placement and creative.
Match-rate health and Supply Path Quality (ad spend actually reaching your ICP).
If the number can’t influence a budget call, it’s not a Moneyball metric.
6) Decision cadence (how to “trade players” every week)
Daily (lightweight):
Reallocate 10–20% of budget to the top EPA deltas.
Enforce guardrails: frequency caps, max CPEP by segment, viewability/attention floors.
Weekly (heavy):
Roster review: which channels/creatives/supply paths are above average on EPA and RARA? Bench or promote.
Creative lab: 5–10 new micro-tests, each tied to one HVA hypothesis.
Account moves: expand high-surge accounts (more table-setters), add closers where HVAs are stalling.
7) The plumbing (your scouting network)
Account Graph: unify IPs, domains, hashed emails, ID partners into one spine. This is how you see cross-channel touches as a single account story.
HVA Listener: instrument forms, product pages, docs, and micro-engagements; stream events to CRM + DSP.
Goal Bidding: push goals (HVAs/attention thresholds) into your DSP; bid to expected value with shading, not clicks.
This is where most teams fail. Don’t be most teams.
8) 30-day starter plan (steal this)
Week 1 — Baseline & model
Backfill 6–12 months of touches and opportunities.
Label HVAs; train a simple model:
P(opportunity_in_30_days | touch).Calculate EPA and CPEP across your current mix.
Week 2 — Controls & guardrails
Define frequency caps, match-rate SLAs, attention floors, and max CPEP by segment.
Stand up the creative lab cadence.
Week 3 — Roster shifts
Move 15% of spend into the top three EPA+RARA clusters (channel × creative × supply path).
Cut the bottom quartile ruthlessly.
Week 4 — Scale the edges
Double down on proven contexts/dayparts.
Expand into lookalike account clusters with positive RARA.
Publish the first Moneyball Dashboard to the exec team (EPA trend, CPEP by position, top RARA accounts, creative leader board).
9) Plays that work (right now)
Proof Pack Creative: run variants that lead with integration logos, 1-line ROI, or a 15-sec product loop; map each to a specific HVA.
Mid-page Contexting: placements adjacent to docs/integration/security content—unsexy, wildly efficient.
Attention Floors: bid only into inventory that clears your attention threshold instead of blindly chasing cheap CPMs.
Retarget Narrow, Not Often: low frequency, high specificity creative tied to the last HVA observed.
SPO Trim: 80/20 your paths; fewer hops, better match, higher EPA.
10) Objections (and the answers)
“But CTR is up 40%!”
Cool. Did EPA rise? If not, you bought dopamine, not pipeline.
“We can’t see every HVA.”
You don’t need perfect coverage—just consistent, high-signal coverage. Start with pricing, docs, forms, and repeat visit logic.
“Attribution will miss half of this.”
Exactly why we optimize to expected value. EPA bakes in signal strength and deal size so you don’t need last-click crutches.
If I were your CPO, these are the five rules
Spend follows EPA deltas, not politics.
Every creative maps to a single HVA hypothesis.
Supply paths are a portfolio—pruned weekly.
Identity is a product, not a project.
Two review cycles without EPA lift? Bench it.
The closer
Moneyball wasn’t about data for data’s sake. It was about pricing reality better than the market. In B2B, reality says: attention and high-intent micro-behaviors predict pipeline—and most of your competitors still optimize for clicks. That’s your gap. Exploit it.
For subscribers (Pro)
EPA / CPEP / RARA worksheet & formulas (plug in your data, get your first dashboard in an hour)
HVA catalog with capture methods (analytics, GTM triggers, CRM activities)
Weekly roster meeting template (what to cut, what to scale, and how to justify it)
Creative lab hypotheses pack tied to HVAs
(Upgrade to get the toolkit and the dashboard spec.)