The Unit Economics Dashboard Every Growth Team Should Build
Sixteen months. That's the median time a B2B SaaS company now needs to recover its customer acquisition cost, according to the Aleph × Benchmarkit SaaS & AI Performance Benchmarks, drawn from FY2025 actuals across 342 companies. Top-quartile teams claw it back in six months or fewer. The bottom quartile sits at 24 months or more. When your median payback runs longer than a year and a quarter, "we'll figure out efficiency later" stops being a strategy.
A unit economics dashboard fixes that. Roughly 12 tiles across four groups — acquisition cost, payback, lifetime value, blended efficiency — each with an explicit formula and a named source system. Not vanity metrics. Numbers you can defend when a CFO leans forward.
I'll admit the first "unit economics" dashboard I built had about 40 tiles and zero I'd defend under questioning. Impressions. Click-through rate. A gauge chart of "brand sentiment" that nobody could source. It looked like a cockpit and worked like a screensaver. This piece is the version I wish I'd shipped instead.
Why this dashboard exists now (and why the blended median lies to you)
Unit economics used to be a fundraising checkbox. You dressed up LTV:CAC for the deck, closed the round, and got back to spending. Heading into 2026 that's gone. Acquisition costs climbed, retention compressed, and the money got expensive. Now the same numbers that impressed investors have to justify next quarter's budget.
Here's the caveat the whole piece rides on. Every blended median misleads enterprise and SMB teams in opposite directions. Read each tile inside an ACV band, never as a single point estimate.
The proof is in the payback split. Citing Benchmarkit's 2024 B2B SaaS Performance Metrics Report, G-Squared Partners found companies above $100K ACV had a median CAC payback of 24 months, versus 9 months for companies at $5K ACV or less. Same metric, nearly triple the number depending on where you sit. Report one blended figure and you've told your enterprise team they're fine and your SMB team they're in trouble. Or the reverse. Both wrong. Band it or don't bother.
The plumbing problem before the pretty tiles
No tile is trustworthy without clean joins. I mean that more than the usual disclaimer.
Two things broke the old measurement stack. iOS App Tracking Transparency and third-party cookie deprecation cut multi-touch attribution coverage to roughly 30–60% of 2020 levels, per Improvado's attribution research, which means data governance now matters more than whichever attribution model you argue about at the offsite. And LTV:CAC, the metric everyone quotes, is really a data-integration problem. As Improvado puts it, accurate tracking requires unifying marketing spend, CRM attribution, and billing data across systems that don't natively connect.
That's the real work. Three source systems feed every tile you're about to build: ad platforms for spend, your CRM for who converted and from where, and billing for what they actually paid and what it costs to serve them. They rarely share a clean customer ID. Until they join on one, every downstream number is a guess wearing a suit.
The 12 tiles, spec'd
This is the spine. Each tile gets a formula, the source system it pulls from, and the failure mode that quietly breaks it.
| # | Tile | Formula | Source system(s) | Known failure mode |
|---|---|---|---|---|
| 1 | Blended CAC | Total spend ÷ new customers | Ad platforms + billing | Hides paid vs organic mix |
| 2 | Paid CAC | Paid spend ÷ paid-sourced customers | Ad platforms + CRM | Attribution gaps overcount organic |
| 3 | CAC Payback (months) | (Margin-adj. revenue) ÷ monthly gross profit per customer | CRM + billing | Using raw ARR; not lagging S&M |
| 4 | Gross Margin % | (Revenue − COGS) ÷ revenue | Billing + finance | Excluding true delivery cost |
| 5 | LTV (margin-adjusted) | ARPA × margin × avg lifetime | Billing | Using raw ARR instead of gross profit |
| 6 | LTV:CAC | LTV ÷ CAC | All three | Joining across systems with no shared ID |
| 7 | Net Revenue Retention | (Start ARR + expansion − churn − contraction) ÷ start ARR | Billing | Snapshot instead of cohort; not banded by ACV |
| 8 | Gross Revenue Retention | (Start ARR − churn − contraction) ÷ start ARR | Billing | Confused with NRR; hides the churn floor |
| 9 | MER (blended) | Total revenue ÷ total marketing spend | Ad platforms + billing | Reading it without segment context |
| 10 | Payback by ACV band | Tile 3, cut by ACV tier | CRM + billing | Not building it at all |
| 11 | Contribution margin/customer | Revenue − variable cost, per customer | Billing + finance | Ignoring support and infra cost |
| 12 | Cohort curve | Retention/expansion over months since signup | Product + billing | Needs time; can't fake it early |
The three tiles that carry the argument
CAC Payback is where most dashboards flatter themselves. Aleph's guidance is blunt: use gross-margin-adjusted revenue, not raw ARR, because a payback period that ignores cost of delivery makes low-margin businesses look healthy. And lag the S&M spend. The customers who signed in March came from money you spent in January. Get those two things right and your 9-month payback might reveal itself as 14.
Gross Margin % gates two other tiles. G-Squared Partners flags 75%+ as the operating target they like to see, and that percentage flows straight into margin-adjusted LTV (tile 5) and payback (tile 3). Build it first or the tiles it feeds are decoration.
NRR, cohort-based. The median for venture-backed SaaS runs around 106%. SaaS Capital's 2025 private-company research recommends benchmarking NRR by ACV rather than by company age, revenue, or industry, since companies at a similar selling price have the most in common. So follow a real cohort forward in time, and cut it by contract size. A snapshot NRR blended across your whole book tells you almost nothing.
Worked example: reading three tiles for one fake company
Napkin math, deliberately round. Call the company Ledgerly. Trailing revenue of $2M, marketing spend of $500K. That's an MER of 4:1 — four dollars of revenue for every dollar of marketing. On a board slide it looks great.
Now the CAC and payback layer. Say blended CAC lands at $10K, gross margin is 80%, ARPA is $500/month. Monthly gross profit per customer is $500 × 0.80 = $400. Payback is $10,000 ÷ $400 = 25 months. Suddenly that clean 4:1 has a hangover.
Here's the split the blended numbers were hiding. Ledgerly has two motions. The SMB motion runs a CAC around $3K, ARPA $500/month, payback closer to 9 months, right in line with the sub-$5K ACV benchmark from Benchmarkit's data. The enterprise motion runs a CAC around $40K, ARPA $2,500/month, payback near 24 months, which also matches the >$100K ACV benchmark almost exactly. Both are normal for their band. The blended 25-month number describes a company that doesn't exist.
I once showed a CFO the 4:1 MER version of a slide like this. First question: "So what's the payback?" I had a blended number. Second question: "Broken out by segment?" I did not. The meeting went quiet in the way that costs you a headcount. Band your tiles before someone asks the second question, because someone always asks the second question.
Wiring the dashboard: where the tiles actually get built
Tool choice sits downstream of clean plumbing. No platform reconstructs a customer ID your systems never wrote down. With that said, here's an even-handed look at where these tiles get assembled and how each option handles the join problem and cohort tracking.
Warehouse + BI layer (Looker, Metabase). If your ad spend, CRM, and billing data already land in a warehouse and someone models them together, a BI tool renders every tile above. This is the most flexible and the most work. NRR cohorts and margin-adjusted LTV live in the SQL you write. Full control, full maintenance burden.
Attribution/integration layer (Improvado). The hard part of tiles 3, 5, and 6 is unifying spend, CRM attribution, and billing, and Improvado's own writing frames LTV:CAC as exactly that integration problem. This layer's job is stitching the three source systems before your BI tool ever draws a chart. Strong on the join, less about the visualization.
Product/behavioral analytics with a chat-first angle (Kixo). Kixo covers product analytics — events, funnels, retention, cohorts, user flows — plus session replay and mobile attribution including deferred deep links. Its distinct angle is chat-first: you ask for a tile in plain language and get an AI-generated chart with a visible reasoning trail, so you can check how the number was built rather than trusting a black box. Honest maturity caveat: the finance-side joins for margin-adjusted LTV — COGS, billing, contribution margin — still live in your warehouse or billing system, not a product SDK. It's strong for the cohort and retention tiles (7, 12) and the acquisition side. It's not where your gross margin comes from.
The pattern across all three: whoever owns the join owns the truth. Pick the tool that fits where your data already lives, then argue about chart colors later.
Build checklist (ship-it order)
- Define your ACV bands first. Everything downstream gets read inside them, so decide the cuts before you build a single tile.
- Confirm the three source systems — ad platforms, CRM, billing — actually join on one customer ID. If they don't, stop here and fix it. This is the whole game.
- Build Gross Margin % next. It gates payback and LTV, so nothing margin-adjusted is real until this tile exists.
- Layer CAC Payback with lagged S&M spend and margin-adjusted revenue. This is where you'll find the ugly number, and the ugly number is the useful one.
- Add cohort NRR last. It needs time to accumulate, so start collecting now even if the tile stays empty for two quarters.
The one number to put on the board slide
If you can only defend one tile, defend MER, banded by segment. Total revenue over total marketing spend, per the argument from Eightx that platform-reported ROAS turned unreliable after iOS 14 and the blended business-level number became the honest one. It survives the attribution wars because it doesn't depend on winning them.
Payback and LTV:CAC are the follow-up questions, and you should have them ready, banded, with gross margin baked in. Get MER right first, because that's the number that opens the meeting.
Last-click's funeral, I keep hearing, is well attended in theory. Somehow nobody's shown up.