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.
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.
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.
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.
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.
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.
Both methods write off exactly the same total cost. Only the timing differs — and timing is worth money. A full year-by-year schedule for one asset, the crossover year, the switch-to-straight-line convention, and the present value of the tax deferral computed at 8%.
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.
FIFO and LIFO are assumptions about which cost you attach to a sale, not about which box leaves the warehouse. Worked through identical purchases and sales, they move cost of sales, inventory, profit and tax — but operating cash before tax is identical to the cent.
CCC = DIO + DSO − DPO. It is the number of days your cash is out of your hands, and it is the reason a profitable, growing business runs out of money. Worked end to end, with the negative-cycle case that makes suppliers your cheapest lender.
EOQ = √(2DS/H) balances ordering cost against holding cost. Its most useful property is how flat the cost curve is around the optimum — and its four failure modes are quantity discounts, lumpy demand, a finite replenishment rate, and the two inputs nobody can measure.
WACC = E/V × Re + D/V × Rd × (1 − T). The tax shield makes debt genuinely cheaper, and the cost of equity comes from CAPM — whose beta and equity risk premium are estimates that move the answer by whole percentage points, and the valuation by a quarter.
The quick ratio removes inventory because inventory is the current asset that may not convert. Two companies with an identical current ratio can be five times apart on the quick ratio — and both measures are still blind to the one thing that actually stops a company paying: timing.
EBITDA adds back the two costs that differ most between companies, which is exactly what makes it comparable — and exactly why it flatters anyone who owns a lot of equipment. Here is the same profit walked all the way down, and the maintenance-capex floor the measure never shows.
Margin is measured on the selling price, markup on the cost. They describe the same profit from opposite ends, they are never equal, and reading one as the other quietly removes a large slice of your gross profit.
Cookies
This choice is about measurement. Google Analytics, which tells us where visits come from, stores two cookies and records nothing until you accept. Ads are shown either way, and the ad network stores cookies of its own: refusing does not remove them. What we measure