Envio Revenue v2 — The Math, Updated With New Evidence

From: Kaustubh Agrawal — incoming Growth Engineer Source memos this updates: ENVIO_REVENUE_MATH.md (v1, the original math) · ENVIO_REVENUE_MODEL.md (the strategy memo) Evidence absorbed since v1: ENVIO_INDEXER_TEARDOWN.md · ENVIO_ONBOARDING_FORENSIC.md · ENVIO_CLICKHOUSE_TEARDOWN.md · ENVIO_CLICKHOUSE_DECK.md

v1 of the math memo was written before I'd read your customers' code, audited your onboarding flow as a fresh user, or seen the ClickHouse Sink announcement. Three significant pieces of evidence have come in since then, each of which changes the model. This v2 absorbs all three and lands a refreshed headline. Original v1 stays in the package as historical — I'd rather show evolution than overwrite the prior thinking.

Same disciplines as v1: every number is real (mine), industry benchmark, or an explicit placeholder labelled as such. Section 7 is the only place I do dollar math at Envio's business, and even there the formula is yours to plug into.

TL;DR — How v1 Changed

Three changes from the original math memo, in order of revenue impact:

1. A new sixth lever has appeared. The ClickHouse Sink launch this week introduces a Dedicated-tier expansion thesis that doesn't fit cleanly into any of v1's original five levers. It's a different funnel (named-account expansion, not free-trial conversion), a different ACV (Dedicated-tier, the highest in the pricing structure), and a different playbook (the launch deck). Modeled here as Lever 6.

2. The activation lever's confidence is higher. The onboarding forensic identified five concrete, fixable friction points with quantified fix costs (one working day for the Greeter tutorial fix; a few hours for the overview-page reorder). v1's recovery-rate assumption of 25% on non-activators was conservative; with the specific fixes scoped, 35–40% recovery is now the defensible base case.

3. The case-study lever's cadence is higher. The indexer teardown surfaced six free case-study seeds in plain sight in your customers' open-source code (the Sablier 0-alias caching trick, Velodrome's CLAUDE.md gotcha list, the codegen pipeline pattern, etc.). v1's case-study cadence assumption of 4 per year was paced for original-research interviews. With the seeds-in-plain-sight inputs, 6–8 per year is now the defensible base case for year 1.

The combined effect: the headline band shifts from v1's $140k–$570k to a v2 band of ~$250k–$880k, with a placeholder-value midpoint around $430k–$580k. The role's payback math improves correspondingly — base case ~3.5× against typical Growth Engineer total cost, high case ~5–7×.

1. Why a v2 — and What's NOT Changing

The two things v1 got right that this v2 keeps untouched:

• The five-lever structure for the original funnel (activation / migration / case-study / expansion / vertical content). Each lever's mechanic is unchanged. Only their input numbers move.
• The "plug your numbers in" methodology. Every number below remains an explicit placeholder. The team replaces them with internal data on day 1.

What v2 changes:

• Adds Lever 6 — Dedicated-tier expansion via ClickHouse Sink as a structurally distinct revenue line
• Refreshes Lever 1 (Activation) with the docs-audit findings → narrower sensitivity, higher base
• Refreshes Lever 3 (Case Study) with the indexer-teardown seeds → higher cadence, higher base
• Re-runs the combined math
• Notes which levers were unchanged (Migration, Expansion, Vertical Content) and why

The headline number changes; the methodology doesn't. That's the discipline.

2. Mirror Protocol Stats — Unchanged From v1

For continuity, the same lived-experience baseline anchors this document as the original:

This is the foundation. Everything below builds on the proof that I've executed the customer's side of this funnel end-to-end.

3. Lever 1 — Activation, REFRESHED

What v1 said

Monthly signups (N):                  ~300 placeholder
Non-activators (35% × N):             ~105
Recovered (25% × non-activators):     ~26/month
Paid conversion (5%, 12-mo window):   ~16 net-new paid customers/yr
Mid-market ACV placeholder:           $4,000
→ ~$64,000 incremental year-1 ARR

What changed

The onboarding forensic identified five specific friction points with concrete fix costs, four of which ship in under a week of focused work:

The first four fixes ship inside two working weeks. The total activation-rate lift is bounded by the lowest of the four (the Greeter time-estimate fix is the largest single jump). Aggregating cautiously: recovery rate of non-activators moves from v1's 25% assumption to a v2 base case of 35–40%.

Refreshed math

Monthly signups (N):                  ~300 placeholder *[unchanged]*
Non-activators (35% × N):             ~105 *[unchanged]*
Recovered (35% × non-activators):     ~37/month → 444/year *[was 26/mo, 312/yr]*
Paid conversion (5%, 12-mo window):   ~22 net-new paid customers *[was 16]*
Mid-market ACV placeholder:           $4,000 *[unchanged]*
→ ~$88,000 incremental year-1 ARR *[was $64,000 — up 38%]*

Why I'm confident in the lift. The four fixes have published industry-benchmark conversion-lift numbers behind them. The Greeter tutorial fix in particular is the most well-validated finding in dev-tools onboarding research — tutorials with explicit time estimates and success indicators retain roughly 2× as many users to completion. v1's 25% conservatism was warranted before the audit; with the audit done, conservatism is no longer the right frame.

4. Lever 2 — Migration Capture, MOSTLY UNCHANGED

What v1 said

5 completed migrations in year 1
~$18k migration-tier ACV placeholder
→ ~$90,000 incremental year-1 ARR

What changed

The indexer teardown gave me one new piece of conviction: migrations from sophisticated existing customers are likely larger than my v1 ACV estimate. Sablier's monorepo runs three separate Envio indexers, with a custom codegen toolchain — that's a Production Medium or Large customer profile, not a Production Small. The handful of migration prospects likely to engage in year 1 are not generic Subgraph-shaped customers; they're the ones whose stack already looks like Sablier's or Velodrome's.

Conservative ACV revision: $18k → $22k placeholder for migration-tier customers in year 1. Still well below Dedicated-tier; this is the mid-market end of migration.

Refreshed math

5 migrations × $22k ACV = ~$110,000 incremental year-1 ARR *[was $90,000 — up 22%]*

The lever's count is unchanged (5 migrations is still the realistic year-1 ceiling for one Growth Engineer running Migration Concierge). The ACV moved up modestly.

5. Lever 3 — Case Study, REFRESHED

What v1 said

4 case studies published year 1
~5 inbound enterprise inquiries per study (industry benchmark)
~20% close rate at $50k enterprise ACV
→ ~$80,000 year-1 recognised (40% of booked)

What changed

The indexer teardown surfaced six free case-study seeds in your existing customers' open-source code:

1. The Sablier 0-alias Effect cache optimization
2. Sablier's @sablier/devkit codegen pipeline pattern
3. Velodrome's CLAUDE.md gotcha list (which is itself six potential canonical Growth assets)
4. Velodrome's Aggregators/ + Snapshots/ separation as a production architecture pattern
5. Sablier's three-indexer-monorepo pattern (streams + airdrops + analytics)
6. The Sablier-runs-Envio-and-Graph-in-parallel pattern (proof of "comparison-then-commitment" customer behavior)

Each of these is a publishable post in the case-study factory's voice — customer-credited, technically precise, distributed in 5 formats. Even the original-research case-study interviews (the v1 model) become easier once these "starter" posts have established the cadence and the production rhythm.

Refreshed cadence assumption: 6 case studies in year 1, of which 3 are open-source-derived (low cost) and 3 are original-research interviews (the v1 model). That maps to roughly 1 published study every 8 weeks — well below the case-study factory's eventual ~1/month cadence.

Refreshed math

6 case studies × ~5 inbound inquiries × 20% close × $50k ACV = ~$300,000 booked
40% recognised in year 1 (timing) = ~$120,000 incremental year-1 ARR *[was $80,000 — up 50%]*

The factor of growth here is real but bounded: I'm cautious about over-promising on case-study production cadence in year 1 because customer-permission cycles introduce real lag. v2's number of 6 stays inside the credible range; 8–10 would start to be aspirational.

6. Levers 4 + 5 — Expansion + Vertical Content, UNCHANGED

Why these aren't being refreshed

• Lever 4 (Expansion) depends on internal account-by-account context I don't have. v1's number of $40k incremental year-1 ARR (4 expansions × $10k uplift) stays unchanged because the new evidence I've gathered doesn't sharpen this lever specifically. That's a place we'd recalibrate together in week one with the real customer list.
• Lever 5 (Vertical Content) is the back-loaded lever — its real payoff is years 2–3 from SEO compounding. v1's $12k year-1 estimate is conservative and intentionally so; refreshing it doesn't change the year-1 picture.

Discipline note: refreshing every lever just because I have the chance would be a tell that I'm pushing the headline number rather than honoring the evidence. Two of the five levers don't have new evidence behind them. Their numbers stay.

7. NEW: Lever 6 — Dedicated-Tier Expansion via ClickHouse Sink

This is the lever v1 didn't model because the feature didn't exist yet.

The thesis

ClickHouse Sink is structurally a tier-up trigger. It's exclusive to the Dedicated Plan. It targets the analytics-product persona — customers whose dashboards/leaderboards/BI workloads hit Postgres-aggregation latency at scale. The named cohort I'd outreach to first (from the launch deck): Velodrome, Sablier, Polymarket, LI.FI, Limitless — all already on Envio, all already past the analytical-workload threshold.

This isn't a free-trial-conversion play. It's a named-account expansion play with a different funnel shape than v1's five levers. Modeled separately on its own merits.

The funnel math

The 90-day launch playbook (the ClickHouse deck) maps to a year-1 revenue funnel:

Named outreach targets (year 1):     ~10 (the 5 from the deck + 5 second-wave)
Conversion to active Sink trial:     ~50% with white-glove engagement
                                   = 5 active trials
Trial → paid Dedicated upgrade:      ~60% (high — these are pre-qualified)
                                   = 3 net-new Dedicated customers
Dedicated-tier ACV (incremental):    placeholder *$40,000* per customer
                                     (the *delta* above their existing tier;
                                      total Dedicated ACV likely $60k–$120k)
→ ~$120,000 incremental year-1 ARR (placeholder values)

Why these placeholder inputs are defensible

• 10 outreach targets is conservative — I can name 5 from the public customer wall today; the wave-2 is plausible from continuing the same audit on remaining named customers.
• 50% trial conversion is high but realistic for a named-account play with the launch's "2 months free" Dedicated trial offer. Cold outbound to unrelated prospects would be 5–15%; warm intros to existing customers with structural fit are 40–60%.
• 60% trial-to-paid is high because the trial is the validation step — by the time a customer has actively run the Sink against their workload, the upgrade decision is largely made.
• $40k incremental ACV is the delta over their existing tier, not the full Dedicated ACV. Conservative; the full Dedicated ACV is plausibly $60k–$120k.

Why this lever has the longest LTV

Once a customer has plumbed ClickHouse Sink into their analytics product, switching costs are high — they'd have to migrate their dashboards, BI tool connections, and aggregation queries elsewhere. Net revenue retention on the Dedicated-tier cohort is structurally higher than the mid-market cohort, with industry benchmarks of 130–150% NRR for analytics infrastructure of this shape.

Sensitivity ranges

• Half case (5 targets, 30% conversion, 50% trial-to-paid, $30k delta): ~$22,500 — small but real
• Base case (above): ~$120,000
• 2× case (15 targets, 60% conversion, 70% trial-to-paid, $50k delta): ~$315,000

The half case is bounded below by zero only if the launch lands without a single named customer engaging — which is implausible given the architectural fit at Velodrome and Sablier alone. The base case is what I'd actually defend.

8. Combined Math — All Six Levers At Placeholder Values

v2 LEVER                                             Year-1 ARR
─────────────────────────────────────────────────────────────────
Lever 1 — Activation (REFRESHED)                     +$88,000
Lever 2 — Migration (slight ACV bump)                +$110,000
Lever 3 — Case Study (REFRESHED — higher cadence)    +$120,000
Lever 4 — Expansion (unchanged)                      +$40,000
Lever 5 — Vertical Content (unchanged, back-loaded)  +$12,000
Lever 6 — NEW: Dedicated-tier ClickHouse Sink        +$120,000
─────────────────────────────────────────────────────────────────
v2 TOTAL at placeholder values                       ~$490,000
v1 TOTAL at placeholder values, for comparison       ~$286,000
─────────────────────────────────────────────────────────────────
v2 lift over v1                                      +$204,000 (+71%)

Sensitivity at the lever level

If any single lever input is half what's placeholdered, that lever's contribution halves — but the v2 floor (half-on-everything) sits around ~$250,000, materially above v1's floor of $140,000. That's the real upgrade in the model: the floor moves up because two original levers tightened and a new lever was added.

If any single lever input is 2×, that lever doubles. The v2 ceiling (2× across the board) sits around ~$880,000.

Defensible v2 band: ~$250k–$880k year-1 incremental ARR.

Comparison table

The widening of the band is honest — adding a new lever with less established conversion data widens the uncertainty. The midpoint move is the headline.

9. Breakeven Math, Refreshed

Against a typical remote junior-to-mid Growth Engineer total cost of $120k–$160k annualised:

The role pays for itself across the entire band. The half-case-still-pays-back property is the single most important shift from v1, where the half-case was roughly breakeven.

10. Year 2–3 Trajectory — Updated

The biggest year-over-year compounding comes from three of the six levers:

• Lever 3 (Case Study) — multi-year SEO half-life on each published study; expect 2–3× year-1 contribution by year 3
• Lever 6 (Dedicated-tier ClickHouse) — high NRR (130–150% benchmark) means each year-1 Dedicated customer carries forward and expands; year-3 contribution from the year-1 cohort alone is plausibly 2× the year-1 number, before counting net-new year-2/3 acquisitions
• Lever 5 (Vertical Content) — back-loaded by design; year-3 contribution from year-1 work is 5–10× year-1 contribution

Implied year-3 attributable ARR run-rate: ~$1.2M–$2.5M, materially above v1's $1.0M–$2.0M projection. The Dedicated-tier lever is the largest single contributor to the upward revision.

11. What This v2 Doesn't Change

To stay honest about the model's limits:

• The numbers are still placeholders. Every input here is replaceable on day 1 with internal data.
• The role is still one Growth Engineer. None of the math assumes additional headcount or budget.
• The competitive landscape is still excluded from the docs. As in v1, no competitor names appear in this memo.
• Year 1 is still the year I'd want to be measured against. Year 2–3 trajectories are projections, not commitments.

12. What I'd Want To Do With This v2 In Week 1

Three things:

1. Replace every italicised placeholder with the real internal number — same exercise as v1 day-1, now with two more levers' worth of inputs to validate
2. Stress-test Lever 6 specifically with the team — the named-account funnel is the highest-confidence-loss zone in the v2 model, because the named accounts have internal context I can't see
3. Co-edit the year-3 trajectory — the case-study and Dedicated-tier compounding numbers are the ones I'd most want the team's roadmap-context input on, because they depend on Envio's own product investment in ClickHouse Sink + the case-study production cadence

Closing — Why a v2 Exists

The v1 math memo was honest about its sources and its limits. What I didn't say in v1 — because I hadn't done the work yet — is that v1 was the floor of the model. Three weeks of additional research (your customers' code, your docs, your latest feature launch) all moved the model in the same direction: upward. None of the new evidence weakened the model. Some of it tightened the bands; one piece (ClickHouse Sink) added an entirely new lever.

That convergence is the strongest single signal I can offer that the v1 model was directionally right and the v2 model is more accurate.

I'd rather walk into the planning meeting with both versions in the package — v1 as the historical anchor, v2 as the updated thinking — than overwrite v1 and pretend I always had this number. The lineage is the credibility.

— Kaustubh