Glossary
Every term the book teaches, in one line each, as the chapter that teaches it uses it, with a link to that chapter.
A · B · C · D · E · F · G · H · I · J · K · L · M · N · O · P · Q · R · S · T · U · V · W
A
API: A way for other systems to make links through the same rules, such as a CMS asking for a link per page. Every value is checked against the lists before anything ships (Chapter 9).
Approved values: The values a taxonomy allows in each slot or segment, and no others. They live in the Tracking Constitution (Chapter 3).
Attribution model: A rule for splitting one order’s credit among the touches in its path. The rule-based models are last click, first click, linear, position-based and time decay; data-driven learns its split from the data (Chapter 11).
Audit: Someone opening the report on a schedule to find what the rules missed: every distinct value, unfiltered, held against the list and against the links that were made (Chapter 8).
Auto-tagging: A platform writing onto your links by itself. An email platform writes UTMs; Google Ads adds a click ID, gclid, and GA4 then takes the source, medium and campaign from Google Ads (Chapter 2; the collisions in Chapter 5).
B
Beside the path: Where a rule lives when it’s written in a document: it holds when the person making a link remembers it. Inside the path, it’s built into the place links get made, where the wrong value isn’t offered (Introduction).
Break test: Trying to break your own rules in a tool, with your real lists, to see which mistakes it prevents, warns about or waves through (Chapter 9).
C
Cascading dropdown: A field whose value narrows the next field’s menu: once the medium is email, the source menu offers only the email sources (Chapter 9).
Changelog: The record of every change to a taxonomy, one entry per change: what changed, the date it takes effect, the mapping from old to new, and who made it. Each entry is a new version of the rules (Chapter 10).
Channel: Two things share the word: a dimension your team decides (paid social, email, paid search), and the analytics tool’s grouping of source and medium. They match only when the medium list is built to make them match (Chapter 3).
Channel grouping: The analytics tool reading the source and medium together and filing the visit into a channel, by its own rules. In GA4, offline matches no rule and lands in Unassigned unless you add a custom channel group (Chapter 2; GA4’s rules in GA4 Field Notes).
cid: A custom parameter that carries the whole letter in one value, cid=. Adobe Analytics classification rules split it back into channel, source, initiative, product and launch month (Introduction).
Click ID: An ID the ad platform adds to the link itself, like Google’s gclid. Analytics can turn it into a campaign only when the two are linked (Chapter 2).
Closed-loop reporting: The field’s name for the money that went out, the visits it bought and the leads that came back, joined on the same campaign value (Chapter 7).
Concatenation: Building a value by joining the values of several fields in order, a hyphen between each: initiative, product and launch month give spring_sale-runners-2025_03 (Chapter 4).
Controlled vocabulary: The list of allowed values for each slot. Closed, a value that isn’t on the list isn’t used; open, anyone types what seems right, though a tool can still hold its format (lowercase, underscores) (Chapter 3).
CRM: Customer relationship management system. It keeps one record per person and per deal: who they are, what they bought or might buy, and where they came from, if anyone told it (Chapter 7).
Custom parameter: Your own key on a link, like the cid. One slot holds everything, and every tool that reads it needs rules to split it (Chapter 2).
Cutover: The date from which a taxonomy change applies, for campaigns launching from then. From it the tool offers only the new values; links already in the wild keep sending the old ones (Chapter 10).
D
Dark social: Links pasted into messaging apps, work chats or personal email. They usually arrive with no referrer and no tags, and the visit files as direct (Cheat Sheet).
Data pipeline: What brings a system’s data into the warehouse on a schedule, one pipeline per system. Stage 5 of the tool map, beside the warehouse (Chapter 7).
Data warehouse: A database built to hold every system’s data in one place, where it lands on a schedule: the ad platforms’ spend, analytics’ visits and revenue, the CRM’s leads. BigQuery and Snowflake are two (Chapter 6).
Data-driven attribution: A model that learns how much credit each kind of touch gets by comparing the paths that converted with the paths that didn’t. (Chapter 11; GA4’s version in GA4 Field Notes).
Delimiter: The character that separates the pieces of a value. The book uses two: underscores join the words inside a segment, hyphens separate segments, so spring_sale-runners-2025_03 splits into three whole pieces (Chapter 3).
Dimension: One fact a report groups by, recorded with each click: channel, source, campaign, and others such as product, region or launch date. Chosen by working backwards from the reports you need (Chapter 1).
Direct: Analytics’ name for a visit it can’t place, one with no referrer and no tags (Chapter 2).
Drift: Values creeping apart as new spellings and habits come in. Auditing, on the index card, is there to catch it before it reaches a decision (Introduction; the audit in Chapter 8).
Dynamic value: A placeholder in a link that a platform fills in. Meta replaces {{campaign.name}} with the campaign’s name, so the link says whatever the campaign is called (Chapter 5).
E
Envelope: Anything that carries the letter: a tagged link, a cid, an ad platform’s campaign name, a CRM lead record. It has a fixed shape, someone else opens it, and it can be swapped or lost on the way (Introduction).
Export: A system’s record pulled out as a file. An ad platform’s carries spend, impressions and clicks per campaign, ad set and ad, with their names and no UTMs (Chapter 6).
F
fbclid: Facebook’s click ID, added to ad links and to ordinary shared links alike. Meta uses it for its own conversion tracking; GA4 doesn’t use it (Chapter 2).
Final link: The link in the visitor’s address bar when the page loads, the only one analytics reads. Whoever writes last decides it (Chapter 5).
First touch: Fields on a lead written once and never overwritten. They answer what brought this person in (Chapter 7).
Flat: One dimension per value: spring_sale, facebook, paid_social. Nothing to parse, and nowhere to put a second fact (Chapter 4).
Flight: One run of a campaign. A relaunch is a new flight with its own launch date (spring_sale-shoes-2025_09), which keeps the two flights apart in a report (Chapter 3).
Fragmentation rate: Distinct values in a parameter divided by the real number of things they name. 31 utm_source values for 8 real sources is 3.9x; 1.0x means no duplicates (Executive Summary).
Franken-spreadsheet: A sheet that started as a rulebook and grew a formula per column and a tab per team, until it became the link factory nobody maintains (Chapter 9).
G
GA4: Google Analytics 4, which replaced Universal Analytics in 2023. It reports a traffic source at three scopes (GA4 Field Notes).
gclid: Google Ads’ click ID, added by auto-tagging. GA4, linked to Google Ads, takes the source, medium and campaign from it; your utm_content and utm_term go to the Manual ad content and Manual term dimensions (Chapter 5; the Manual dimensions in GA4 Field Notes).
Governance platform: A tool above a shared builder that adds what a form can’t: approvals, a record of who made which link, workspaces, an API and webhooks. (Chapter 9).
H
Hidden form field: A form field the visitor never sees, sent with the form like any other. A script fills it with the values saved when the visitor landed, so they reach the lead (Chapter 7).
I
Incrementality test: An experiment that compares a group that sees the ads with a control group that doesn’t; the difference is the lift the ads caused. It measures cause, for what was tested, while it ran (Chapter 11).
Index card: The five parts of taxonomy governance, in order: dimensions, rules, workflow, auditing, automation (Introduction; read against the tool map in Chapter 12).
Inside the path: Where a rule lives when it’s built into the place links get made, where the wrong value isn’t offered. See Beside the path (Introduction).
J
Join: Lining up rows from two tables on a column that holds the same value in both, such as an ad platform’s spend beside analytics’ revenue (Chapter 6).
Join key: The column a join matches on, character for character. Spend and revenue share only the campaign name, paired with the platform (Chapter 6).
K
Key event: GA4’s name for an event that measures an action important to the business, created and reported the way its conversions used to be (Google, Conversions vs. key events, checked 2026-10-06; more in GA4 Field Notes).
Key-Value: Each segment labels itself with a key and a colon: init:spring_. Order doesn’t matter and a new key breaks no older name, at the price of a parser that reads keys in every tool (Chapter 4).
L
Last touch: Fields on a lead rewritten on every tagged visit. They answer what brought this person back the time they converted (Chapter 7).
Letter: The facts a campaign’s reports need, such as Stride’s channel, source, initiative, product and launch month, whatever envelope carries them. The letter plus its rules is the taxonomy (Introduction).
Link shortener: A redirect. A short link like stride.co/nyc25 sends the browser on to the address it was given, and the tags arrive only if they’re written into that address (Chapter 5).
Lookup table: A list that reads a recorded value and fills in what it implies: a campaign that promoted only lifestyle shoes means lifestyle (Chapter 1). Grown into a table, it holds every dimension behind an Opaque ID (Chapter 4).
M
Mapping table: The working half of a changelog: each old value or pattern, what it becomes, and how old data reads (remapped, a marked break, derived by lookup, or unknown) (Chapter 10).
Marketing automation: The system beside the CRM that runs the forms and the emails, such as a sign-up form and the emails that follow it (Chapter 7).
Marketing mix modeling: An estimate of each channel’s contribution from aggregated spend and sales over time, with no user-level data, so it sees channels nobody clicks. It needs about two years of weekly history for a robust estimate (Chapter 11).
msclkid: Microsoft Advertising’s click ID, for its conversion tracking (Cheat Sheet).
N
na: The value a Structured segment takes when its dimension doesn’t apply, so every segment after it stays in place. The other choice is an empty place between two hyphens (Chapter 4).
Naming pattern: One shape every campaign name follows. Stride’s is [initiative]-, as in spring_sale-runners-2025_03 (Chapter 3).
O
Opaque ID: The link carries one meaningless identifier, cid_4d71e09a, and every dimension lives in a lookup table. It survives any change and asks the most: one source of truth, every system joined to the table, and a table kept complete (Chapter 4).
Orphan row: What a failed join leaves: spend with no revenue, or revenue with no spend. No error is raised (Chapter 6).
Overlap: The phase of a taxonomy change when old and new formats run side by side, until the longest-lived link stops. Reports show one combined view or a marked break, never the two mixed unmarked (Chapter 10).
P
Parsing: Splitting a value back into its parts and reading each for what it is, in a spreadsheet, a warehouse or a report filter. Whatever does it is your parser (Chapter 4).
Path: The visits analytics ties to one buyer before an order, each under the values it arrived with. It’s all any attribution model sees (Chapter 11).
Pixel: A tag on your site that reports each sale back to an ad platform, which credits its own ads within a window set in the platform. Two platforms can each claim the same order (Chapter 6).
Post-hoc classification: Cleaning up in reports alone: a remap kept as a standing list instead of done once. It catches what it already knows about and nothing new (Chapter 8).
Q
QR code tracking: A printed link plus whatever redirect sits behind it. The code can’t change once it’s printed, so the redirect writes the tags (Chapter 5).
Query string: Everything between the ? and any # in a link: key=value pairs joined by &, which analytics reads. Tags after a # stay in the browser (Chapter 2).
R
Referrer: The browser’s note of where the visitor just was. It’s often missing (email apps, QR codes), often only the domain, and it never says why (Chapter 2).
Remap: A correction list a report reads recorded values through, such as FB read as facebook; the recorded click never changes. It works only when the value meant one thing (Chapter 1).
Retargeting: An audience, not a medium. A retargeting ad keeps its platform’s paid medium, like paid_social (Chapter 7).
ROAS: Return on ad spend, revenue divided by spend. $8,400 of revenue on $2,400 of spend is a ROAS of 3.5 (Chapter 6).
S
Segment: One piece of a packed value, between delimiters. spring_sale-runners-2025_03 has three (Chapter 3).
Source of truth: The one place all your data lands first, usually a data warehouse: the one copy everyone reads from. Opaque ID needs one (Chapter 4).
Structured: Segments in a fixed order, joined by hyphens, built by concatenation: spring_sale-runners-2025_03. Position carries the meaning, so every segment is present on every name, as na or an empty place when it doesn’t apply (Chapter 4).
Survey: Asking the buyer, as in “How did you hear about us?” on the order or sign-up form. It catches any touch the buyer remembers and names, clicked or not (Chapter 11).
T
Taxonomy: The letter plus its rules: which dimensions you record, the approved values for each, what each value means, and who may add one. The taxonomy is the asset; UTMs are one delivery mechanism (Introduction).
Taxonomy governance: Defining a taxonomy, encoding it, and keeping it enforced (Introduction).
Taxonomy owner: The one person who makes each change to the rules: approves new values, runs the audit and keeps the changelog. No owner, no taxonomy (Chapter 10).
Tool map: The route a click’s values travel, in six stages: plan and make links, distribute, land and record, capture and connect, store and model, read (Chapter 2; more examples per stage in The Tool Landscape).
Touch history: One row per tagged visit, kept in the browser or sent to the warehouse, tied to the lead when the form is sent. It’s the only way the middle of a journey survives (Chapter 7).
Tracking Constitution: The one shared document of a team’s rules: the approved values, the pattern with examples, who approves a new value, and how to ask (Chapter 3).
Two-stage workflow: One team defines each campaign value once, by concatenation; then each channel or regional team picks it from a dropdown of the defined campaigns, and chooses source, medium and content. It fits where several teams share one campaign; where one group runs each campaign, links are made in one stage (Chapter 9).
U
URL builder: A form where you paste the address, fill in the values and copy out the tagged link. It assembles one link correctly, and refuses a wrong value only where its fields are closed (Chapter 9).
URL rewrite: Something changing a link on its way: a wrap (the click routed through a platform’s redirect first), an add (a parameter such as fbclid), or a drop (the query string not carried forward) (Chapter 5).
UTM parameters: Five standard tags on a link, read by nearly every analytics tool. UTM stands for Urchin Tracking Module, and the names are Urchin’s convention; GA4 gives them force by building its channels from source and medium and its campaign reports from campaign (Chapter 2).
utm_campaign: By convention, why the link exists: the initiative, often packed with more facts by a naming pattern, as in spring_sale-runners-2025_03 (Chapter 2; the pattern in Chapter 3).
utm_content: By convention, which link or creative: hero-image, video-testimonial-15sec. An ad platform’s ad level matches it (Chapter 2; the levels in Chapter 6).
utm_id: One of the extra utm_ keys GA4 reads beyond the five, for a campaign’s ID (Chapter 2; what GA4 does with it in GA4 Field Notes).
utm_medium: By convention, how the click reached you: paid_social, cpc, email, from one closed list. GA4 reads it with the source to choose the channel (Chapter 2; the list in Chapter 3).
utm_source: By convention, where the click came from: the platform or property, lowercase, such as facebook or newsletter (Chapter 2; the list in Chapter 3).
utm_term: By convention, which keyword or audience: running_shoes, retargeting-cart_abandoners. An ad platform’s ad group or ad set level matches it (Chapter 2; the levels in Chapter 6).
V
Validation: A check of each value against the rules at the moment a link is made: on the list, fits the pattern, lowercase, safe characters. It stops a value that breaks a rule, and only for links made where it runs (Chapter 8).
ValueTrack parameters: Google Ads’ dynamic values, such as {campaignid}, filled in at click time. There’s none for the campaign name (Cheat Sheet; the {campaignid} key in Chapter 6).
Versioning: What changelog entries add up to: anyone can read a recorded click by the rules in force when its link was made, by its date or, after a change, by its shape (Chapter 10).
W
Webhook: The platform telling another system each time a link is made, so a launch sheet or a team’s chat channel hears about it without anyone checking (Chapter 9).
Workspace: One team’s or client’s part of one master taxonomy. It can narrow the master list, never widen it (Chapter 9; choosing the set-up in Chapter 10).