Module 7 — Measurement

Module time ~2 hours (reading + exercises) · ~3,500 words · 10 cited primary sources · Tool-independent · v1.0 · 2026-06-09

7.0 Measurement exists to change decisions

The last module of Foundations closes the loop the first one opened. Module 1 ended with the economic frame — LTV, CAC, the 60/40 split — and every module since has deferred its measurement questions here: the brief's field 7 (Module 2), the DoD's tracking check (Module 3), the attribution truce (Module 4), the finance defense (Module 5). This module pays those debts.

One principle organizes everything: a metric earns its place by changing a decision. Before adopting any number, name the decision it informs: if this goes up, we do X; if it drops below Y, we do Z. A metric with no attached decision is decoration — and decoration is not neutral, because every reported number consumes attention, invites gaming (§7.6), and crowds out the numbers that matter. The marketing dashboard of a well-run small company is short. The discipline of keeping it short is this module.

The corollary deals with marketing's oldest measurement lament — the line attributed (in slightly varying forms, and with contested origin) to retailer John Wanamaker: half the advertising budget is wasted, but nobody knows which half. The modern answer is not that the problem is solved — it is that the problem is now manageable, with three honest instruments (§7.4) that each see part of the picture. Knowing which instrument answers which question is the Foundations competence; pretending one instrument answers everything is the Foundations failure mode.

7.1 The unit economics — LTV, CAC, payback, with the math

Module 1 (§1.6) introduced the three numbers; here is the working math, in the form popularized for the operating community by David Skok's "SaaS Metrics 2.0" essays (freely available at forentrepreneurs.com) and standard across subscription and repeat-purchase businesses.

CAC — fully loaded, honestly

CAC = (all sales + marketing spend in period) ÷ (new customers acquired in period)

The two honesty rules: fully loaded means ad spend plus content costs, tools, freelancers, and the people-time (including the founder's — Module 2's Cell 8 discipline); and blended vs channel CAC are different instruments. Blended CAC (all spend ÷ all customers) tracks the whole machine's efficiency; channel CAC (channel spend ÷ customers attributed to channel) guides budget shifts — and inherits every attribution problem in §7.4, so treat channel CAC as an estimate with error bars, not a fact.

LTV — contribution, not revenue

The simple form for a subscription or repeat-purchase business:

LTV = average gross profit per customer per month × average customer lifetime in months

where lifetime, if you only know your churn rate, is approximated by 1 ÷ monthly churn rate (5% monthly churn → 20-month expected lifetime). The two honesty rules: gross profit, not revenue — LTV computed on revenue flatters every channel decision by your COGS percentage; and the simple form assumes stable churn, which young companies don't have. With under a year of data, treat LTV as a rough, conservative estimate and lean harder on payback (next), which needs less forecasting.

Payback period — the cash-reality check

Payback = CAC ÷ gross profit per customer per month — the months until a new customer has returned the cost of acquiring them. This is the metric that respects a bootstrapped company's actual constraint: cash. An LTV/CAC ratio of 4 sounds healthy; if payback is 30 months, the company may be dead before the L in LTV arrives. The rules of thumb from the SaaS finance literature — LTV/CAC above ~3, payback under ~12 months (consumer) to ~18 (enterprise) — are heuristics, not laws; bootstrapped businesses typically need payback far shorter than funded ones, because the marketing budget is recycled customer cash, not investor cash.

7.2 The channel metrics in context — CTR, CPC, CPM, CPA, email rates

The platform metrics exist to diagnose the unit economics, not to replace them. The catalog, with what each is for:

The diagnostic chain is the skill: a campaign with healthy CTR and bad CPA has a landing-page or offer problem, not a creative problem. Bad CTR with fine CPM has a message problem. Rising CPM with stable CTR means the audience is saturating or the auction got more expensive. Each combination points at a different fix — which is what "metric in context" means. Read columns alone and you optimize noise.

7.3 UTM hygiene — the unglamorous foundation

Every attribution claim in §7.4 is built on link tagging, and link tagging is where SMB measurement actually fails. UTM parameters — the utm_source, utm_medium, utm_campaign (plus utm_content and utm_term) appended to URLs, a convention dating to the Urchin software that became Google Analytics — are how your analytics knows where a visitor came from. The hygiene rules, learned by every team the hard way:

1. One written taxonomy. facebook, Facebook, fb, and meta are four different sources to your analytics. A one-page convention — lowercase always; source = platform, medium = paid/organic/email/referral; campaign = the brief's name — ends the fragmentation. This document is part of the Module 3 process layer.
2. Tag at creation, verify at QA. UTMs are applied when the link is made (a shared link-builder spreadsheet is sufficient tooling) and checked in the pre-launch QA (Module 3 DoD check #3 — "every tracking parameter present and correctly formed").
3. Never tag internal links. UTMs on links within your own site overwrite the visitor's true source with a fake one — the single most common self-inflicted analytics wound.
4. Accept the blind spots honestly. Dark social — links shared in private messages, untagged word of mouth, podcast mentions — arrives as "direct" traffic. Direct is not a channel; it is the bucket of everything you couldn't see. A simple post-purchase "how did you hear about us?" field recovers more of this truth than any tagging scheme, and its verbatims feed Module 2's research besides.

7.4 Attribution — MTA, MMM, and incrementality

Now the centerpiece. Module 4 (§4.4) promised the formal version of why attribution disputes are unresolvable at deal level; here it is, as the three instruments and what each can honestly say.

Multi-touch attribution (MTA) — the tracking approach

MTA follows individual users across touchpoints via identifiers and assigns conversion credit by a model: last-touch (all credit to the final click — the default in most tools, and systematically biased toward bottom-funnel channels like brand search), first-touch (all credit to the introduction — biased toward top-funnel), or position-based / data-driven models that split credit by rule or by algorithm. Two honest limits: the model choice is a belief about how marketing works wearing a math costume — switching models reallocates millions in apparent channel value without a single customer behaving differently; and MTA's visibility is degrading structurally, as privacy regimes (Module 6, §6.3), browser tracking prevention, and walled-garden platforms remove the identifiers it depends on. MTA remains useful for relative, directional reading of digital channels at SMB scale; it is no longer (if it ever was) ground truth.

Marketing mix modeling (MMM) — the statistical approach

MMM is the older econometric tradition, returned to favor precisely because it needs no individual tracking: regress sales outcomes on marketing inputs (spend by channel over time) plus controls (seasonality, pricing, promotions), and estimate each channel's contribution statistically. It captures what MTA can't see — offline channels, brand effects, long lags — and respects privacy by design. Its limits: it needs meaningful spend variation and history (typically two-plus years), delivers estimates with wide uncertainty at small budgets, and updates slowly. At Foundations scale, know what MMM is and when your company will graduate into it; full practice is Expert-tier material.

Incrementality — the experimental approach

The gold standard where feasible: did this marketing cause sales that would not otherwise have occurred? The methods — holdout tests (withhold the campaign from a random group), geo experiments (match similar regions, market in some), and platform lift studies — all share the experimental logic of a control group. The canonical cautionary finding comes from the large-scale eBay paid-search experiments (Blake, Nosko & Tadelis, "Consumer Heterogeneity and Paid Search Effectiveness," Econometrica, 2015): pausing brand-keyword search ads produced almost no sales loss — the clicks MTA credited to those ads came overwhelmingly from people who would have arrived anyway. Not every brand's result, but every brand's warning: attributed is not incremental.

The SMB-feasible version requires no statistics degree: pause one suspect channel for two to four weeks and watch total (not channel-attributed) sales; run geographic or audience splits where volume allows; ask the "how did you hear about us?" question and compare its answers to what the tracking claims. Crude instruments, honestly read, beat precise instruments misread.

7.5 Measuring the long of it — brand-equity tracking

The 60/40 allocation from Module 1 (§1.6) creates a measurement obligation: if roughly sixty percent of investment is long-horizon brand building, and none of §7.4's short-window instruments can see it, the brand work will be defunded by default — Binet and Field's own warning about over-reading short-term metrics. The brand-side dashboard, scaled to SMB reality:

• Branded search volume — people typing your name into a search engine, trended over quarters. The cheapest available proxy for mental availability (Module 1, §1.2), visible in free search-console and keyword-planning tools.
• Direct + organic arrival share — the fraction of traffic and of new customers arriving without paid prompting, trended. Rising unpaid arrival is compounding mental availability; flat unpaid arrival under rising paid spend is a treadmill.
• The "how did you hear about us?" distribution — the same field from §7.3, read longitudinally. When "a friend," "saw you mentioned," and "been following you" grow as a share, the brand reservoir is filling.
• Review and referral velocity — counts per quarter from the Module 5 (§5.6) proof supply chain.
• Survey-based awareness and consideration — the formal instrument (aided/unaided recall, consideration-set presence) when budget allows; the structured version of everything above, and standard practice from Expert tier onward.

None of these moves in a week. All of them should move in a year. The reporting discipline from Module 5 (§5.3) applies: present them as the explicit second half of the portfolio — "here is what the 60 is buying" — on a quarterly rhythm, so the long of it has a ledger entry and not just a hope.

7.6 Goodhart resistance — how every metric gets gamed

The module — and the curriculum — closes with the law that governs all measurement. Goodhart's Law, articulated by economist Charles Goodhart in 1975 and sharpened in Marilyn Strathern's now-standard phrasing: "When a measure becomes a target, it ceases to be a good measure." The moment a metric carries reward or punishment, optimization pressure flows toward the metric itself rather than the outcome it proxied.

Marketing's gallery of Goodhart casualties, each one a metric from this module:

The defenses generalize into three habits. Pair every target with a countervailing metric — one that the cheapest gaming of the target would damage (CTR with CPA, volume with quality, growth with retention). Keep decision metrics and performance targets separate where possible — the verdict table of §7.1 informs budget decisions; the moment a single line of it becomes someone's bonus, audit it. Re-derive, quarterly, what each metric was a proxy for — the retro question (Module 3, §3.6) applied to measurement itself: "what decision does this number serve, and is it still serving it?"

And with that, Foundations closes where it began. Module 1 defined marketing as the application of behavioral, empirical, and economic science to creating profitable mutual value with customers. Seven modules later, the definition has its full operating system: the theory (1), the workflows (2), the process (3), the judgment (4), the coalition (5), the governance (6), and the measurement that keeps the whole thing honest (7). The synthesis portfolio — and the role of Marketing Coordinator it certifies — is the demonstration that the system runs in your hands, on a real business, without any particular tool. That was the promise on the front page; this is where it's kept.