29 guides on marketing
The metrics a campaign is judged on, and what each one hides. Acquisition cost, lifetime value, reach and return on spend — defined before they are computed, because most disagreements about a dashboard are disagreements about a definition.
Ranked by how many pages of this site link to them, not by traffic.
LTV/CAC says whether a customer is profitable eventually. Payback says whether you can afford to wait. The formula divides CAC by monthly gross profit — and the gross margin is the term nearly everyone drops.
Camille Laurent · 14 min read
Apple Mail Privacy Protection pre-fetches tracking pixels, so a share of your opens were never opened by a person. Open rate now moves with your audience's mail client. Here is what survives, worked on a single send.
Camille Laurent · 13 min read
Sample size is set by your baseline conversion rate and the relative lift you want to detect — and it explodes as that lift shrinks. Here is the formula, a full table at a 3% baseline, and why stopping early breaks the test.
Camille Laurent · 12 min read
CPM and CPC are not two prices for the same thing. They are the same auction seen from two sides, and click-through rate is the exchange rate between them — so the cheaper option flips as soon as CTR moves.
Camille Laurent · 12 min read
Percentage of promoters minus percentage of detractors. Passives sit in the denominator and nowhere else, the result runs from −100 to +100, and it is not a percentage despite looking like one.
Camille Laurent · 8 min read
Engagement rate is not one metric. Divide by followers, by reach or by impressions and the same post returns three different numbers — and every platform defaults to a different denominator.
Camille Laurent · 9 min read
There is no single per-view rate, and every page that quotes one is mixing four unrelated payment routes. Here is what each route actually pays, what moves you within the range, and why the fund is the smallest line for most accounts.
Camille Laurent · 8 min read
Between roughly $1 and $10 per 1,000 views, and the reason the range is that wide is arithmetic, not luck. Here is every step from advertiser spend to your account.
Camille Laurent · 7 min read
A hashtag makes a post findable in a query, which is a different job from making it spread. Model the visibility a tag actually buys and the answer falls out: a tag's worth is its viewers-per-post-published, not its volume — so a very popular tag returns almost nothing and a specific one returns twenty-five times more.
Camille Laurent · 12 min read
Published best-time tables average over audiences that are not yours, and on a ranked feed posting time matters far less than it did on a chronological one. Compute your own audience's activity across time zones instead — six zones flatten the timing lever from a 94% swing to a 22% one — and see how many posts a real timing test needs.
Camille Laurent · 12 min read
Platforms crop rather than letterbox, so an image is only safe if its subject sits inside the intersection of every ratio it will be shown at. That intersection has a closed form — the narrowest ratio divided by the widest — and it works out at 29.45% of the source for a realistic set. Geometry that does not expire, plus the pixel counts that do.
Camille Laurent · 12 min read
The famous email ROI multiples are self-reported survey figures with the labour left out of the denominator and revenue that would have arrived anyway left in the numerator. Build the honest version instead: incremental revenue against full cost, measured with a holdout — and the arithmetic of how big that holdout has to be.
Camille Laurent · 12 min read
The denominator should be delivered, not sent — which means defining hard and soft bounces properly. Unique and total opens are different numbers. And a reported open rate has a computable null point: past a certain share of prefetching clients it carries no information about humans at all. The instrumentation that replaces it.
Camille Laurent · 12 min read
Your bid decides whether you are eligible; the advertiser below you decides what you pay. Doubling a bid can move the price not at all, quality can halve it for the same position, and the average CPC in your report is a click-weighted mixture that describes no query. The auction arithmetic, worked.
Camille Laurent · 13 min read
A slug has two jobs that pull against each other: it is a permanent identifier and it is a piece of readable text. Length, hyphens, stop words, dates, non-ASCII characters and the identifier-plus-slug pattern that gets both properties — with the real numbers from a site that localises 1,736 tool slugs into six languages.
Camille Laurent · 13 min read
What Google's own documentation says about rewriting title links and about the meta description, rather than what SEO folklore says. Then the measurable part: titles are truncated by pixel width, so two titles of exactly sixty characters can render 204.55 pixels apart and only one of them survives.
Camille Laurent · 13 min read
Constant additions against a proportional loss rate do not grow forever — they converge on a ceiling equal to additions divided by the loss rate. Here is that ceiling computed for six loss rates, why a list can grow while its engaged half shrinks, and the feedback loop that makes a bought list self-defeating.
Camille Laurent · 11 min read
The overall rate is the product of the stage rates, which makes a percentage gain anywhere worth the same percentage overall — and a ten-point gain worth wildly different amounts depending on where you put it. Plus the lag that makes a growing business look worse than it is.
Camille Laurent · 11 min read
Impressions divided by reach is frequency — one identity that decides what a campaign actually is. The same impression budget split five ways, what a frequency cap really costs you, and why the three-exposure rule is a 1970s heuristic rather than a law.
Camille Laurent · 12 min read
A returned order is not a cancelled sale, it is a sale that cost you money. Here is the return rate at which returns eat the entire contribution of the orders that stayed sold, for three product profiles and five margin levels — plus what bracketing does to your order count.
Camille Laurent · 14 min read
A bounce was never a measure of dissatisfaction — it was a session that sent one hit. Google Analytics 4 replaced the definition entirely, and the new one is a configurable timer. Here is the same thousand sessions read five ways.
Camille Laurent · 13 min read
Average revenue per user is a mean over a distribution with no middle, divided by a denominator nobody defines. Here is the same month of revenue read five ways, and two opposite businesses landing on exactly the same ARPU.
Camille Laurent · 12 min read
CPL is spend divided by leads, and alone it means almost nothing, because a lead is whatever you decided to call one. Worked here: a channel with twice the CPL producing half the CAC, and the redefinition that moves CPL fivefold without moving CAC at all.
Camille Laurent · 14 min read
Density counted occurrences because retrieval once counted occurrences. TF-IDF, then BM25 with its saturation curve, then embeddings replaced it. Here is the same 800-word page scored three ways, and why the three disagree.
Camille Laurent · 13 min read
Schema.org markup buys eligibility for a rich result, never a guarantee and never a ranking boost. Here is which types still produce something visible in 2026, what each one requires, and the one rule that gets sites penalised.
Camille Laurent · 15 min read
Each of the five fields has one job, and attribution rarely breaks on the mechanism — it breaks on inconsistent values. Here are the naming rules, the case-sensitivity trap, and the internal-link mistake that destroys the original attribution.
Camille Laurent · 12 min read
K is invitations per user times the conversion rate of an invitation — two small numbers multiplied. Below 1 the loop is a finite multiplier worth 1/(1−K), and cycle time decides which loop actually wins.
Camille Laurent · 11 min read
Abandonment rate is one minus completed over created — and the figure quoted everywhere is an average across wildly different shops. The useful work is turning a percentage point of checkout completion into money, then discounting it for returns.
Camille Laurent · 10 min read
These tools multiply followers by an assumed rate per thousand, which makes them a CPM model under another name — and every input behind that rate is an assumption they never show you.
Camille Laurent · 10 min read