GTMOS decides which accounts to work, why now and what to say, then routes them, syncs the CRM and measures what happened. Sales teams lose pipeline in the gaps between those steps; this is the infrastructure that closes them. Every number below is computed live from the database.
Opportunity amount at creation · last 16 weeks
The current week is partial.
494 accounts first contacted in 90 days
| Stage | Accounts | Step | Cohort |
|---|---|---|---|
| Contacted | 494 | — | 100% |
| Engaged | 126 | 26% | 26% |
| Qualified | 71 |
Highest-intent A/B accounts that are not customers
| Account | Score | Intent | Latest signal | Owner |
|---|---|---|---|---|
| Kestrel Analytics AI/ML Platforms | A98 | 96 | Crossed the free-tier trace volume threshold 2d ago | Sam Okoro |
| Copper Networks AI/ML Platforms | A79 | 70 | Crossed the free-tier trace volume threshold |
From the GTM Stack Inspector
Last successful sync 7h ago · 0.2% record failures (7d)
85% of accounts fully enriched · 15% incomplete · 310 stale
0 A/B accounts unowned · 95 unmatched decisions (90d) · 13 rule conflicts resolved
18 duplicate contacts · 35 accounts missing employee count · 504 open issues
Last 90 days · no GTMOS decision uses open rates
Opens are shown for completeness only: Apple Mail Privacy Protection and image proxies make open tracking unreliable, so no GTMOS decision uses them.
Denominators, formulas and the caveat that stops each number being misread
Accounts whose first outbound touch happened inside the reporting window. Every funnel conversion below is measured against this fixed cohort, so later stages can mature after the window closes.
count(distinct account) where first stage transition to 'contacted' is within the window · denominator: accounts, not contacts
Caveat: A cohort funnel understates recent windows: accounts contacted yesterday have not had time to convert.
Share of the contacted cohort that ever reached a given stage, including after the window ended.
accounts reaching stage ÷ accounts reaching the previous stage · denominator: accounts in the cohort
Caveat: Stages can be skipped (a warm inbound can book a meeting without an 'engaged' transition), so step conversion is not strictly sequential.
Discovery calls that actually happened, not meetings booked. Bookings that no-show never count.
count of activities of type meeting_held in the window · denominator: activities
Caveat: Booked-to-held ratio is a separate operational metric and is not reported in V1.
| 56% |
| 14% |
| Meeting | 39 | 55% | 8% |
| Opportunity | 28 | 72% | 6% |
| Won | 4 | 14% | 1% |
Accounts whose first outbound touch happened in the last 90 days (494 accounts). Later stages are counted whenever they occurred, including after the window, so the cohort denominator stays fixed.
| Jordan Patel |
| Stratus Robotics Developer Tools | B71 | 62 | 6 open AI/ML roles 13d ago | Jordan Patel |
| Quill AI AI/ML Platforms | A76 | 56 | 5 open AI/ML roles 8d ago | Drew Martinez |
| Driftwood Platforms Fintech | B70 | 56 | Launched Driftwood Copilot 10d ago | Avery Brooks |
| Cobalt Robotics Data Infrastructure | A78 | 55 | Launched Cobalt Insights AI 7d ago | Jordan Patel |
| Tern IO Developer Tools | B67 | 55 | Viewed pricing 4 times in 7 days 24d ago | Casey Nguyen |
| Stratus AI Data Infrastructure | B62 | 55 | 15 open AI/ML roles 4d ago | Lena Fischer |
The fastest path through GTMOS
Last product event 24h ago · 6.1% webhook failures (7d) · 0 unmatched events
Incomplete for 18% of opportunities (180d)
7.7% of executed runs failed (30d) · 19 in dead letter · 21 steps recovered by retry
2.1% bounce rate (30d) · 66 invalid emails not suppressed
High-confidence signals · 14 days
Meeting rate by grade · contacted accounts, 180 days
Grades are current scores. Past outcomes also feed engagement points, so this view slightly overstates predictive power; a holdout backtest would remove that bias.
Accounts that reached the 'qualified' stage: a positive reply confirmed the pain and the buying context. This is GTMOS's sales-qualified lead equivalent, defined at the account level.
count(distinct account) transitioning to 'qualified' in the window · denominator: accounts
Caveat: Qualification is stage-based, not a score threshold: a high score alone never qualifies an account.
The 'why now' half of the score, normalised to 0–100: intent, timing and engagement points as a share of their combined budget, independent of fit.
(intent + timing + engagement points) ÷ their maximum × 100 · denominator: score categories
Caveat: A high intent index on a poor-fit account is still not a target; read it alongside the grade.
Hours between a qualifying signal being observed and the first outbound touch on that account. The operational SLA for signal-based selling.
median(first touch after signal − signal observed_at) · denominator: signals that triggered a play
Caveat: Demo history is simulated batch processing, so the distribution reflects the seed, not live latency.
Replies received per email sent. GTMOS reports the per-send rate because that is what a sequencer reports; the per-account rate is higher because most accounts receive several sends.
replies ÷ emails sent in the window · denominator: emails sent (not contacts, not accounts)
Caveat: Auto-replies and out-of-office are not separated in the demo dataset.
Replies a rep classified as interested, per email sent. This is the primary outbound quality metric because raw reply rate rewards provocative subject lines that generate negative replies.
positive replies ÷ emails sent in the window · denominator: emails sent
Caveat: Classification is human judgement; in production it needs an agreed rubric and spot audits.
Tracked opens per delivered email. Shown for completeness only and used in no GTMOS decision.
opens ÷ delivered · denominator: delivered emails
Caveat: Apple Mail Privacy Protection and corporate scanners pre-fetch images, so opens are inflated and not comparable across audiences. Never optimise on this.
Total amount of opportunities opened in the window, counted at creation, not at close.
sum(opportunity amount) where opened_at is in the window · denominator: opportunities opened in the window
Caveat: Amounts are the seller's estimate at creation and are never revised down in the demo dataset.
Amount currently in non-closed deal stages, regardless of when it was created.
sum(amount) where stage not in (closed_won, closed_lost) · denominator: open opportunities
Caveat: Not weighted by stage probability; a weighted forecast is a separate metric.
Share of decided opportunities that were won, over opportunities that closed in the window.
closed-won ÷ (closed-won + closed-lost) closed in the window · denominator: closed opportunities
Caveat: Open deals are excluded, so early-stage pipeline does not dilute the number.
Expected revenue per day implied by the current funnel.
open opportunities × win rate × average won deal ÷ median sales cycle (days) · denominator: open opportunities
Caveat: Combines four noisy inputs; useful for direction, not for forecasting. Unreliable below ~20 closed deals, which GTMOS flags.
Share of opportunity value that has at least one recorded touch inside the lookback window.
1 − (unattributed pipeline ÷ total pipeline) · denominator: opportunities in the window
Caveat: Anything GTMOS never saw (referrals, events, dark social) is unattributed by definition, not by failure. A 100% attributed number means your tracking is lying.
Share of accounts where every core field needed for scoring and routing is present (domain, industry, employee count, region, AI team size).
accounts with all core fields ÷ all accounts · denominator: non-merged accounts
Caveat: Coverage says nothing about accuracy; provenance confidence and staleness are tracked separately.
A metric an experiment must not damage, even if the primary metric improves: bounces, unsubscribes and negative replies.
count of guardrail events ÷ emails sent per variant · denominator: emails sent per variant
Caveat: Guardrails are evaluated for harm, not for lift: a variant that wins on replies but doubles unsubscribes should not ship.