Field guide

The PPC math field guide for performance marketers

A practical guide to the math underneath paid-search work. The formulas, the assumptions, the recurring mistakes — for the operator who wants to internalize the math rather than offload it to a calculator.

Most paid-search work is downstream of math the operator never explicitly does. The formulas live inside ad platforms, inside calculators, inside vendor pitches — but the underlying math is simple enough that an operator who internalizes it sees the system more clearly. This guide walks through the math.

Section 1: The metric stack

The paid-search metric hierarchy, top to bottom:

1. LTV (lifetime value): Total revenue a customer generates × gross margin. For subscription, includes recurring revenue. For ecom, often a single transaction.
2. CAC (customer acquisition cost): All-in cost to acquire a customer.
3. Max CAC: The CAC ceiling derived from a target LTV:CAC ratio. CAC_max = LTV ÷ target ratio.
4. CPL (cost per lead): Cost per upstream conversion event — form fill, demo request. Max CPL = CAC_max × lead-to-close rate.
5. CPC (cost per click): Cost per click. Max CPC = CPL_max × click-to-lead rate.

Each step in the hierarchy is a multiplication by a funnel-stage conversion rate. The math is unforgiving: a small improvement at any stage compounds, and a small degradation at any stage compounds the other way. The Max CPC calculator automates this; the math is worth understanding by hand.

Section 2: True ROAS vs. reported ROAS

Reported ROAS = Revenue ÷ Ad Spend. The headline number on most paid-search dashboards. Almost always misleading.

True ROAS subtracts the costs revenue doesn’t actually keep:

For ecom, the gap between reported and true ROAS is usually 30–60%. The size of the gap depends on three things: gross margin (lower margin = bigger gap), return rate (higher returns = bigger gap), and shipping/fulfillment cost (higher = bigger gap). For categories like apparel and furniture, all three are unfavorable; for supplements and beauty, the gap is smaller.

Why this matters operationally

An account optimizing toward reported ROAS will systematically over-spend on traffic that produces revenue but not profit. The model is doing its job — it’s optimizing the metric you told it to optimize. The metric is just the wrong thing.

The fix: optimize toward true ROAS by exporting margin-aware conversion values to the ad platform via conversion-value rules in Smart Bidding (Google’s 2026 capability) or via offline conversion import with a contribution-margin value attached.

Section 3: Break-even ROAS

The ROAS at which an account exactly breaks even, ignoring fixed costs:

At 35% gross margin, 3% processing, 0% shipping: 1 ÷ 0.32 = 3.125×.
At 25% gross margin (electronics), 3% processing, 5% shipping: 1 ÷ 0.17 = 5.88×.
At 70% gross margin (supplements), 3% processing, 7% shipping: 1 ÷ 0.60 = 1.67×.

Profitable ROAS is typically 1.3–1.6× break-even, leaving margin for return-rate uncertainty and the cost of customer service. The target ROAS you set in Smart Bidding should be the profitable ROAS, not the break-even.

Section 4: The attribution math

Different attribution models distribute conversion credit differently across the touchpoints in a customer’s journey. For a journey with N touchpoints, paid Google Ads at position p, sales cycle of d days:

• Last-click: credit = 1.0 if p = N, else 0.0.
• First-click: credit = 1.0 if p = 1, else 0.0.
• Linear: credit = 1.0 ÷ N.
• U-shaped (40-20-40): credit = 0.40 if p is first or last; 0.20 ÷ (N−2) otherwise.
• Time-decay (7-day half-life): credit at position p = 2^{−daysAgo ÷ 7}, normalized so all positions sum to 1.
• Data-driven: proprietary to Google; approximates linear with mid-funnel boost.

The attribution simulator at the sister site visualizes all six models simultaneously. The operational point: the same revenue gets reported as wildly different ROAS depending on which model you pick. For B2B with multi-touch journeys, the difference between last-click and data-driven can be 50%+.

Section 5: Funnel arithmetic for lead-gen

For lead-gen accounts (B2B SaaS, services, professional firms), the math from click to customer:

The unit economics work backward: target LTV:CAC ratio (typically 3:1) implies maximum CAC, which combined with the lead-to-close rate implies maximum CPL, which combined with the click-to-lead rate implies maximum CPC.

The calculation produces a Max CPC that’s often substantially different from what the account is actually bidding. If observed CPC is above Max CPC, the account is over-bidding and unit economics will degrade as spend scales. If observed CPC is well below Max CPC, the account is leaving volume on the table.

Section 6: The compounding-improvement math

Small improvements at each funnel stage compound. Concrete example: an account with 3% click-to-lead, 20% lead-to-close, $10,000 deal value, 70% margin, 3:1 LTV:CAC target. Max CPC = $14.

Improve each stage by 10%:

• 3% → 3.3% click-to-lead
• 20% → 22% lead-to-close
• $10,000 → $11,000 deal value
• 70% → 71% margin

New Max CPC = $18.32, a 31% improvement. Each individual lever was 10%; the compound effect is 31%. The implication: spending operator hours on landing-page optimization (CTL improvement), sales-process improvement (LTC improvement), pricing review (deal value), and margin work compounds better than spending the same hours on bidding.

Section 7: The numbers that lie

Four metrics that systematically mislead:

• Average CPC. Branded clicks (cheap) lower the average; non-branded clicks (expensive) are masked. Always look at CPC by query intent.
• CTR alone. High CTR can mean great offer + great ad. It can also mean cheap, low-quality traffic clicking out of curiosity. Look at downstream conversion rate to interpret CTR.
• ROAS without attribution context. A campaign with 5× ROAS on last-click attribution might have 1.5× ROAS on data-driven attribution if it’s capturing branded queries. The number isn’t real until you specify the model.
• Quality Score. Increasingly opaque; not exposed for Performance Max. Diagnostic only, not a performance metric.

Section 8: When the math says the account isn’t the problem

If Max CPC is below $1, the funnel math doesn’t support paid acquisition. Three options:

• Improve close rate — sales process, qualification, demo conversion.
• Improve click-to-lead — landing page, offer, form length.
• Increase deal value — pricing, product, packaging.

None of these are PPC operator problems. They’re upstream of the ad account. The job of the operator in this case is to surface the math to leadership rather than to keep tuning bids that can never be profitable. The Max CPC calculator exists partly to make these conversations concrete.