Glossary
Every abbreviation and bit of jargon used across these notes, grouped the way the site is. Where a concept has a note of its own, the term links to it.
Statistics
Section titled “Statistics”| Alpha (α) | The false-positive rate you’ve agreed to accept, conventionally 0.05 | → |
| Always-valid inference | Sequential methods whose p-values stay honest however often you look | → |
| Base rate | The share of tested ideas that contain a real effect. Sets how many of your “wins” are false | → |
| Baseline | The control’s current rate. One of the four sample-size inputs | → |
| Bayesian | Updating a prior belief with the data to get a probability the variant is better | → |
| Benjamini-Hochberg | A multiple-testing correction controlling false discovery rate. Less brutal than Bonferroni | → |
| Beta (β) | The false-negative rate. Power is 1 − β | → |
| Bonferroni | Divide alpha by the number of comparisons. Simple, and usually too strict | → |
| CI | Confidence interval. The range of effects the data hasn’t ruled out | → |
| Cohort analysis | Grouping customers by when they arrived and tracking each group forward separately | → |
| Covariate | A pre-test variable used to strip noise out of a result | → |
| Credible interval | The Bayesian interval, which does mean what people wrongly think a CI means | → |
| CUPED | Controlled-experiment Using Pre-Experiment Data. Uses pre-test behaviour to strip noise out of the result | → |
| Delta method | The variance maths that makes ratio metrics behave | → |
| Effect size | How big the difference is, as distinct from whether it’s significant | → |
| Expected loss | The average cost of shipping the wrong variant, weighted by how likely each wrongness is | → |
| Family-wise error rate | The chance of at least one false positive across a whole set of comparisons | → |
| FDR | False discovery rate. The share of your “significant” findings that are false | → |
| Frequentist | Asking how surprising the data would be if there were no effect. Produces p-values | → |
| H0 / H1 | Null and alternative hypothesis. H0 is the boring one | → |
| HARKing | Hypothesising After the Results are Known. Writing the prediction once you’ve seen the answer | → |
| Heavy tail | A distribution where rare large values dominate the average | → |
| M-type / S-type error | Right direction but inflated / wrong direction entirely | → |
| MDE | Minimum detectable effect. The smallest lift you commit to being able to see | → |
| Novelty effect | Early engagement inflated purely because the thing is new | → |
| p-hacking | Running comparisons until one crosses 0.05, then reporting that one | → |
| Point estimate | The single observed effect, before you look at the interval around it | → |
| Posterior / prior | Belief after / before seeing the data | → |
| Regression to the mean | Extreme early readings drifting back toward the truth, which looks like decay | → |
| ROPE | Region of practical equivalence. The band around zero you’d call “no meaningful difference” | → |
| RPV | Revenue per visitor. Conversion rate × AOV, and brutal to test on | → |
| Shrinkage | Extreme estimates pulled back toward zero, usually correctly | → |
| Significance | The verdict that a result is unlikely enough under chance that you’ll act on it | → |
| Simpson’s paradox | A trend in every subgroup reversing once you combine them | → |
| SPRT | Sequential Probability Ratio Test. The original sequential method | → |
| Zero-inflated | A distribution where most observations are zero, like revenue per visitor | → |
Marketing concepts
Section titled “Marketing concepts”| Advertorial | A long-form, editorial-styled landing page built for cold paid traffic | → |
| Ascension | A customer moving up to a more expensive rung | → |
| Awareness stages | Unaware, problem-aware, solution-aware, product-aware, most aware | → |
| Blended | Averaged across all channels or cohorts, usually hiding the variation that matters | → |
| BOFU | Bottom of funnel. Decision stage | → |
| CAC | Customer acquisition cost | → |
| CLV / CLTV | Customer lifetime value. Same thing as LTV, different house style | → |
| CRO | Conversion rate optimisation | → |
| Demand capture | Getting in front of buyers already shopping. Easy to measure, and finite | → |
| Demand generation | Creating want that didn’t exist. Hard to measure, and where the ceiling is set | → |
| JTBD | Jobs to Be Done. What the customer hired the product to accomplish | → |
| LTV | Lifetime value. Total revenue from a customer over the relationship | → |
| MOFU | Middle of funnel. Consideration stage | → |
| Offer | Everything the buyer is choosing between. Price, terms, bonuses, guarantee, urgency | → |
| Pre-sell page | A short warming page sitting between the ad and the product | → |
| TOFU | Top of funnel. Awareness stage | → |
| Traffic temperature | How much prior context a visitor arrives with. Cold, warm, hot | → |
| UTM | The tracking parameters bolted onto a URL to identify the campaign | → |
| Value ladder | A deliberate sequence of offers at rising price points | → |
Conversion psychology
Section titled “Conversion psychology”| AIDA | Attention, Interest, Desire, Action. A copywriting skeleton | → |
| Anchoring | The first number seen setting the reference point for every number after it | → |
| BNPL | Buy now, pay later. Klarna, Clearpay and similar | → |
| Charm pricing | Prices ending .99, leaning on the left-digit effect | → |
| Cialdini’s six principles | Reciprocity, commitment, social proof, authority, liking, scarcity | → |
| Cognitive load | The total mental effort a page demands | → |
| Compromise effect | The pull toward whichever option sits in the middle, regardless of dominance | → |
| Confirmshaming | Guilt-worded opt-outs. “No thanks, I don’t want to save money” | → |
| Curiosity gap | The pull to close a gap between what you know and what you want to know | → |
| Decoy | A third option added to make your target option look obviously better | → |
| Default bias | Whatever is preselected gets picked disproportionately | → |
| Endowment effect | Valuing something more once it’s yours. Why free trials and easy returns work | → |
| Fogg behaviour model | B = MAP. Behaviour needs motivation, ability and a prompt at once | → |
| Framing | Same fact, different presentation, different decision | → |
| Goal gradient | People accelerating as they get closer to a visible finish | → |
| Halo effect | One good impression colouring judgement of unrelated things | → |
| Left-digit effect | Reading £19.99 as “nineteen-something” before processing the rest | → |
| Loss aversion | Losses hurting roughly twice as much as equivalent gains please | → |
| Mental model | The expectations a user arrives with about how your category works | → |
| Negative halo | One bad signal contaminating perception of everything else. Faster than the positive kind | → |
| Open loop | An unfinished task or question the mind keeps wanting to close | → |
| Paradox of choice | More options producing fewer decisions, past a threshold | → |
| PAS | Problem, Agitate, Solution. A copywriting structure | → |
| Peak-end rule | Experiences remembered by their peak moment and their ending, not their average | → |
| Prestige pricing | Round numbers, signalling confidence rather than value | → |
| Processing fluency | How easily the brain decodes what it’s looking at. Fluent reads as true | → |
| Reactance | Pushing back because you’ve noticed you’re being pushed | → |
| Risk reversal | Moving the perceived risk of buying off the customer and onto you | → |
| Scarcity | Limited quantity. Urgency’s sibling, and the one that’s usually faked | → |
| Social proof | Looking to others’ behaviour to decide your own | → |
| Sunk cost | Continuing because of what you’ve already spent, not what’s ahead | → |
| System 1 / System 2 | Fast automatic thinking / slow deliberate thinking | → |
| UGC | User-generated content. Customer photos, video, reviews with images | → |
| Urgency | Limited time. Scarcity’s sibling | → |
| Zeigarnik effect | Unfinished tasks staying in mind in a way finished ones don’t | → |
Experimentation
Section titled “Experimentation”| A/B test | A randomised controlled experiment comparing two versions of something | → |
| AA test | A control-vs-control experiment, run to check the testing infrastructure is honest | → |
| Bandit | An algorithm that reallocates traffic toward the winner while the test runs | → |
| Bucketing | Assigning a user to a variant, usually by hashing their ID | → |
| Cluster randomisation | Assigning whole groups instead of individuals, when spillover makes user-level assignment invalid | → |
| Concierge test | The feature appears to work and is fulfilled by hand behind the scenes | → |
| Control | The existing version, held constant so the variant has something to be measured against | → |
| Epsilon-greedy | The simplest bandit. Mostly show the leader, occasionally explore | → |
| Experiment velocity | How many tests you get through. Compounds harder than per-test rigour does | → |
| Fake door | A painted door that admits the feature doesn’t exist yet | → |
| Friction analysis | Mapping where the user is working harder than they need to, before generating tests | → |
| Guardrail | A metric you watch for damage, not for wins | → |
| Heuristic review | Walking the site against a usability checklist, noting friction | → |
| HiPPO | Highest Paid Person’s Opinion, overriding the result | → |
| Holdout | A slice kept on control after launch, to measure the long-run effect | → |
| HTE | Heterogeneous treatment effect. The effect differing by segment | → |
| ICE | Impact, Confidence, Ease. A prioritisation score | → |
| Interference | One user’s outcome depending on another user’s assignment | → |
| Listwise deletion | Dropping any record with a missing field. The silent default in most tools | → |
| MCAR / MAR / MNAR | Missing completely at random / at random / not at random. Only the last one biases you | → |
| Metric shopping | Trying metrics after the fact until one crosses the line | → |
| Mutual exclusion | Carving traffic so two conflicting tests can’t overlap | → |
| MVT | Multivariate test. Several variables at once, one variant per combination | → |
| Orthogonal assignment | Independent bucketing per experiment, so concurrent tests don’t confound each other | → |
| Painted door | A button for something that doesn’t exist yet, to measure whether anyone wants it | → |
| Peeking | Checking results before the planned sample, which inflates your error rate | → |
| PIE | Potential, Importance, Ease. A prioritisation score | → |
| Pre-registration | Writing down metric, sample and stopping rule before the data arrives | → |
| Primary metric | The one number that decides the test, committed to beforehand | → |
| Proxy metric | A stand-in for something you can’t observe inside the test window | → |
| Randomisation unit | What you assign. It has to match what you analyse, or the variance is wrong | → |
| RICE | Reach, Impact, Confidence, Effort. A prioritisation score | → |
| Salt | The per-experiment seed that keeps bucketing independent between tests | → |
| Segment fishing | Slicing the data until something wins, then reporting that slice | → |
| Sensitivity analysis | Re-running the analysis under other reasonable choices to see if it holds | → |
| Smoke test | The lightest painted door. A button leading to “notify me” | → |
| Spillover | Treatment effects leaking into the control group | → |
| SRM | Sample ratio mismatch. The split isn’t what you asked for, so the test is invalid | → |
| SUTVA | Stable unit treatment value assumption. One user’s outcome doesn’t depend on another’s bucket | → |
| Switchback | Randomising time periods instead of users, when the whole system shares state | → |
| Thompson sampling | A bandit rule allocating by how uncertain each variant still is | → |
| Treatment | The variant carrying the change being tested | → |
| UCB | Upper confidence bound. A bandit allocation rule | → |
| Variant | Any arm of the test, including control in loose usage | → |
| Win rate | The share of tests producing a real win. Above 30% is a diagnosis, not a boast | → |
| Winner’s curse | Shipped winners overstating their true effect, because you only ship what looked good | → |
| Winsorise | Cap extreme values at a percentile instead of dropping them | → |
| Wizard of Oz | The user thinks they’re using software. It’s a person | → |
Growth engineering
Section titled “Growth engineering”| CAPI | Conversions API. Sending conversion events server-to-server rather than from the browser | → |
| CLS | Cumulative Layout Shift. How much the page moves about after load | → |
| CWV | Core Web Vitals. LCP, INP and CLS | → |
| Data layer | The structured object on the page that every downstream tool reads from | → |
| Deduplication | Matching the browser and server copies of one event so it counts once | → |
| Event taxonomy | The naming and property conventions your analytics events follow | → |
| Feature flag | A runtime switch deciding whether a user sees a feature, without a deploy | → |
| FID | First Input Delay. The Core Web Vital that INP replaced | → |
| Flicker | The control showing briefly before a client-side variant swaps in | → |
| FOIT / FOUT | Flash of invisible / unstyled text. Web fonts causing late text shifts | → |
| FOOC | Flash of original content. The same thing as flicker | → |
| GA4 | Google Analytics 4 | → |
| GDPR | The EU and UK data protection regime. Why declined consent means no cookie | → |
| Growth loop | A system where customers create the conditions for the next customers | → |
| GTM | Google Tag Manager. “Server-side GTM” means the container runs on your infrastructure | → |
| Identity stitching | Joining a user’s activity across sessions and devices into one record | → |
| INP | Interaction to Next Paint. Lag between input and visible response | → |
| Input metric | An operational metric that compounds into the north star | → |
| ITP | Intelligent Tracking Prevention. Safari’s cookie restrictions | → |
| Kill switch | Turning a feature off in seconds without a deploy | → |
| LCP | Largest Contentful Paint. How long the main element takes to render | → |
| Match quality | How well the identifiers you send let a platform tie an event to a person | → |
| Modelled conversions | Platform estimates of conversions it couldn’t observe. Never analyse a test on these | → |
| North star | The single company-level metric everything is meant to compound into | → |
| PII | Personally identifiable information. Hash it before it leaves your server | → |
| SPA | Single-page app. React, Vue, Next and similar | → |
| TBT | Total Blocking Time. Not a Core Web Vital, still worth watching | → |
| TTFB | Time to First Byte. Not a Core Web Vital, but upstream of all of them | → |
| Vertical slice | Shipping one feature through every layer it touches, rather than each layer broadly | → |
Business models
Section titled “Business models”| Activation | The point a new user first gets real value. What trial conversion actually depends on | → |
| AOV | Average order value | → |
| ARR | Annual recurring revenue. MRR × 12, and a projection rather than money in the bank | → |
| Churn | Customers leaving. Voluntary if they chose to, involuntary if a card failed | → |
| CM1 / CM2 / CM3 | Contribution margin after goods / after fulfilment / after acquisition | → |
| Contraction | Existing customers paying less through downgrades | → |
| CPL | Cost per lead. Misleading on its own, because it isn’t cost per customer | → |
| CRM | Customer relationship management. Where the closed-won data lives | → |
| Dark pattern | A design that gets the outcome by obstructing the user rather than persuading them | → |
| DTC | Direct to consumer | → |
| Dunning | Recovering failed subscription payments through retries, reminders and card updates | → |
| Expansion | Existing customers paying more through upgrades, seats or usage | → |
| First-order economics | The maths of the first order, which in DTC usually loses money on purpose | → |
| Freemium | A permanently free tier with paid features above it. No clock | → |
| Geo holdout | Turning a channel off in matched regions to measure what it actually caused | → |
| Ghost ads | Showing the control group an unrelated ad where yours would have run | → |
| GRR | Gross revenue retention. NRR with expansion taken out, so it can’t hide churn | → |
| Incrementality | Whether the spend caused the sale or just took credit for it | → |
| MER | Marketing efficiency ratio. Total revenue ÷ total marketing spend | → |
| MQL | Marketing-qualified lead. Someone who looked interested | → |
| MRR | Monthly recurring revenue | → |
| NRR | Net revenue retention. What a cohort’s revenue does over a year, before new customers | → |
| PQL | Product-qualified lead. Someone who has used the product and hit value | → |
| Progressive profiling | Building a lead’s profile across visits instead of one long form | → |
| PSA test | Showing a public-service ad to the control group to measure incrementality | → |
| Reactivation | A churned customer returning | → |
| Repeat purchase rate | The share of customers who buy again. The early read on LTV | → |
| ROAS | Return on ad spend. As the platforms report it, structurally overstated | → |
| SAL | Sales-accepted lead. The rep has agreed to work it | → |
| Save flow | What sits between the cancel button and the actual cancellation | → |
| Self-reported attribution | Asking the buyer how they heard about you, because tracking can’t see the dark channels | → |
| SLA | Service-level agreement. Here, the committed maximum response time on a lead | → |
| Speed-to-lead | How fast a new lead gets contacted. The curve is brutal in the first hour | → |
| SQL | Sales-qualified lead. Worked by sales and confirmed worth pursuing | → |
| Time-to-value | How long before a new user gets something useful. Sets the right trial length | → |
| Value metric | The unit you charge by. Upstream of every pricing-page decision | → |
| Win-back | Re-converting a customer who has already churned | → |
General
Section titled “General”| B2B / B2C | Business-to-business / business-to-consumer | |
| CPM | Cost per thousand impressions | |
| CTA | Call to action | |
| CTR | Click-through rate | |
| PDP | Product detail page | |
| SKU | Stock keeping unit. One sellable product variant |