How to Calculate ROI on Marketing

Marketing ROI is famously easy to overstate. The fastest path to a healthy-looking ROI is using the most generous possible formula — revenue instead of gross margin, excluding sales compensation, crediting all conversions to the channel that touched last. The honest version, measured the same way as the financial statements, tells a different story.

1. Start with a formula you can defend

The clean ROI formula:

Marketing ROI = (Gross-margin contribution − Fully-loaded marketing cost) / Fully-loaded marketing cost

Gross-margin contribution — not revenue. A subscription business with 75% gross margin acquiring $100k of new revenue contributes $75k toward covering CAC plus overhead. Using revenue instead of gross margin would overstate the return by 33%.

Fully-loaded cost. Every dollar you can reasonably tie to the marketing function: ad spend, content production, marketing software, marketing headcount, allocated share of sales compensation for channels with sales involvement, and events. In most companies, fully-loaded marketing cost is 1.3–1.5x the reported "ad spend" line^[2].

2. Compute fully loaded CAC

Customer Acquisition Cost (CAC) = Fully-loaded sales + marketing spend in a period / Net new customers acquired in the same period.

Fully loaded means: sales team salaries and commissions, marketing team salaries, all marketing spend, marketing tools, allocated CS time spent on acquisition (trials, demos), and sales tools. It does not mean: R&D, general overhead, or CS time spent on existing customers.

Worked example. A B2B SaaS spends $200k in a quarter: $90k on marketing team salaries, $40k on ads, $20k on content, $40k on sales commissions on new deals, $10k on tools. They acquired 50 new customers. Fully loaded CAC = $200k / 50 = $4,000 per customer.

Comparing this CAC against gross margin contribution gives payback. Against lifetime contribution gives LTV:CAC.

3. Use payback period as the operational metric

CAC Payback = CAC / (ARPU × Gross Margin), in months.

Continuing the example: CAC $4,000, ARPU $500/month, gross margin 75%. Monthly contribution = $375. Payback = $4,000 / $375 = 10.7 months.

Why payback beats LTV:CAC ratio as the operational metric:

• Payback is measurable in months of observed data. LTV requires projecting retention 24–36 months forward, which in early-stage businesses has wide error bars.
• Payback translates directly to cash flow. "We get our money back in 11 months" is a runway-relevant statement. "3.2:1 LTV:CAC" is a multiple that depends on assumptions.
• Short payback compounds capital efficiency. A company with 9-month payback can self-fund growth much earlier than one with 24-month payback, even at identical LTV:CAC.

Target benchmarks (2024 OpenView data)^[1]:

• SMB B2B SaaS: 12–18 months is healthy, 24+ is a concern.
• Mid-market B2B SaaS: 18–24 months is healthy, 30+ is a concern.
• Enterprise B2B SaaS: 24–36 months is healthy given contract values and retention.
• B2C subscription: 3–9 months is healthy.

4. Attribution: less-bad, not correct

Attribution is a fundamentally hard problem. No model is "correct" — they all make assumptions that fail somewhere. The question is which model is less-bad for the decision you need to make.

• Last-touch: Credits the channel of the final interaction. Easy to compute, systematically under-values top-of-funnel channels.
• First-touch: Credits the channel of the first interaction. Opposite bias — over-values awareness channels, under-values consideration and closing.
• Linear: Equal credit across all touchpoints. Arbitrary but not obviously wrong.
• Time-decay: Recent touches get more credit. Reasonable for short-consideration purchases.
• Data-driven / algorithmic (Markov chains, Shapley values): Learns weights from conversion data^[4]. Best accuracy if you have the data volume; unreliable at small scale.
• Marketing Mix Modeling (MMM): Econometric regression against spend by channel over time. Captures long-tail effects like brand and TV, which touchpoint attribution cannot^[2].

Practical advice: use last-touch for operational decisions (which ad to turn off this week), run periodic MMM analyses for budget allocation, and treat both as directional. The honest framing is "this channel is associated with X% of tracked conversions under Y attribution model," not "this channel produced X."

5. Channel-level ROI tells the real story

Aggregate marketing ROI hides that some channels are strongly positive and others are net-negative. The useful work is at channel level, and it changes decisions.

For each channel, compute: CAC, first-month retention (leading indicator of cohort quality), CAC payback, and 12-month LTV estimate. Rank by payback period. Channels with payback over 24 months should be triaged — either they have uncommonly strong LTV (rare), they need efficiency work, or they should be cut.

Attribution window matters. A B2B product with a 60-day sales cycle measured against a 30-day attribution window will systematically under-credit top-of-funnel channels. Set the window to at least 2x the median sales cycle^[3].

Marketing ROI done honestly is usually less impressive than the dashboard claims and more useful as a decision input. In the typical case, the discipline of computing fully-loaded CAC, gross-margin-based returns, and payback periods reveals that 2–3 channels drive the economics and the rest are spending to hit activity targets.

6. Brand and long-horizon marketing

Some marketing investments — brand advertising, thought leadership content, community building — don't produce trackable short-term conversions. They're often the highest-leverage investments a company makes, and they routinely get cut when ROI frameworks demand attribution that doesn't exist.

The honest treatment:

• Allocate brand budget as a percentage of revenue, not ROI target. Typical ranges: 1–3% of revenue for growth-stage B2B, 3–8% for consumer brands competing in crowded markets. This isn't ROI-based; it's a strategic commitment.
• Measure brand impact via lift studies, not attribution. Brand awareness surveys, share-of-voice analysis, and unaided recall tests over 6–12 month horizons show whether brand investment is compounding.
• Track proxy metrics. Direct traffic growth, branded search query volume, share of referral traffic from target media. These correlate with brand health even when direct attribution is impossible.

Marketing mix modeling can quantify brand effects retrospectively over several years of spend data^[2], which is why it's the right tool for mature companies with multi-channel budgets above $5M. For smaller companies, directional evidence from proxy metrics usually has to be enough.

7. The decisions ROI analysis should drive

Three decisions good marketing ROI analysis actually changes:

• Budget reallocation between channels. Quarterly rebalance toward channels with shorter payback and better retention; reduce or cut channels in the bottom quartile.
• Scope of paid acquisition. When paid CAC exceeds LTV or payback crosses 24 months, the question isn't "how do we optimise paid" — it's "should paid still be a major channel, or should we invest in organic/content/product-led growth?"
• Sales-marketing boundary. At scale, marketing and sales overlap on loaded-cost attribution. Clear boundaries — what does marketing deliver to sales, at what quality threshold — prevent the recurring blame-shifting when overall CAC rises.

The goal of marketing ROI measurement is not to produce a flattering number for the board deck. It's to surface where the economics work, where they don't, and what to do about it. That usually means less sophistication in the math and more rigour in the honesty of the inputs^[3].

8. Numeric worked example — headline ROI vs honest ROI

A SaaS reports a quarterly marketing ROI of 6:1 to the board. Rebuild the number using fully-loaded inputs and gross margin.

Headline math
  Revenue from new customers Q3      $300k
  Ad spend line item                 $50k
  Reported ROI                       6.0x

Honest math
  Ad spend                           $50k
  Marketing team salaries (loaded)   $60k
  Content production + agency        $18k
  Sales commissions on new deals     $24k
  Marketing + sales tools            $8k
  Fully-loaded S&M spend             $160k

  New-customer ARR (annualised)      $300k
  Gross margin (74%)                 $222k
  Honest ROI (first year)            ($222k−$160k)/$160k = 0.39x
  CAC (50 new customers)             $3,200
  Payback ($500 ARPU × 74%)          ~8.7 months

The headline "6:1 ROI" is the fully-variable-cost story the ad platform dashboard produces. The honest reconstruction drops the first-year ROI below 0.5x — the business is paying back in the second year, not the first^[1]. Neither number is "wrong," but only one matches the cash flow and the financial statements. When these two diverge by more than ~2x, it's usually the fully-loaded version that reflects the actual cash position.

9. Failure modes worth naming

• Attribution window shorter than sales cycle. A B2B product with a 75-day average sales cycle measured on a 30-day last-click window systematically under-credits top-of-funnel and over-credits brand/direct — then the team kills the SEO/content programs that were actually doing the work. Set the window to at least 2x the median sales cycle^[3].
• Organic treated as zero-CAC. "Organic" traffic usually reflects cumulative investment in SEO, content, or brand that has fully loaded costs — content production, allocated editorial salaries, backlink programs. A true zero-CAC channel is rare (genuine pure word-of-mouth).
• MMM claims at sub-$2M spend. Marketing Mix Models need several years of spend variation and meaningful weekly volume to separate channel effects^[2]. Teams running MMM on $150k annual marketing budgets are fitting noise. Use simpler geo-holdout tests or incrementality experiments until spend justifies the methodology.

As of 2026-Q2, OpenView benchmarks show SaaS CAC payback medians have lengthened by roughly 3 months versus 2021 highs^[1], driven mostly by rising paid-acquisition CPMs and tighter retention in lower-ACV cohorts. Marketing ROI calculations benchmarked against 2020–2021 numbers overstate current program health.