Glossary

What happens over the period if nothing changes. Every result is measured against it.

Watch for: A baseline that quietly assumes growth or decline moves every result with it.

An input that produces the result, such as price, volume, or churn.

Example: Revenue = units × price, so units and price are the drivers.

A stated belief about how a driver responds to the change.

Example: 20% of Payments + Tax accounts add Radar through the bundle.

Watch for: The result is only as good as its weakest assumption, so each one carries a source.

Where an assumption comes from: a test, past data, a benchmark, or an estimate with no data behind it.

Watch for: An estimate with no data behind it is allowed. It is always labeled, which keeps it from being read as evidence.

How much the result moves when one assumption moves and the rest stay put.

Watch for: Moving one input at a time hides cases where two inputs move together.

The value of an assumption at which the change exactly matches the baseline.

Example: At $7 the cohort is worth the same as today if monthly churn reaches 6.9%.

Watch for: It measures the room for error. It does not say how likely that value is.

What the model said would happen, by line, for a period.

What did happen, by line, for the same period.

What one customer pays per period, or what one unit costs.

The count of units sold: customers, subscriptions, or transactions.

Customers who start paying in one period.

The share of visitors (or trials) who become paying customers.

The share of paying customers lost in one period.

Example: 5% monthly churn means an average customer life of 1 ÷ 5% = 20 months.

Watch for: Monthly and annual churn are not interchangeable. Keep the period the same as the price.

What it costs to deliver the product to one customer for one period, or one unit.

Watch for: It is usually fixed in dollars, so raising price widens the margin percentage. Costs that scale with price, like processing fees, are entered separately.

Cost that rises with price, as a share of price.

Example: A 3% payment fee on a $7 plan is $0.21.

What is left from one customer's price after the cost to serve them, per period.

Price − Cost to serve − ( Variable cost rate × Price ) = Contribution margin

Example: $7 price − $1 cost = $6.

Watch for: It counts only the cost to serve. Sales commissions, support, and fixed costs such as salaries are left out, so it is not profit.

The contribution margin one customer brings in over their whole time as a customer.

Contribution margin ÷ Churn rate = Lifetime value

Example: $6 margin ÷ 5.5% monthly churn = $109.09.

Watch for: Assumes churn stays constant and the price holds for the customer's life. It leaves out expansion revenue and does not discount future dollars.

Also called: LTV, CLV

The lifetime value of everyone who joins in one period.

New customers × Lifetime value = Cohort value

Example: 9,200 new customers × $109.09 = $1,003,636.

The rise in price as a share of the old price.

( Price − Price ) ÷ Price = Price increase

Example: $5 to $7 is a 40% increase.

The drop in price, or the discount, as a share of the old price.

The share of volume you can lose after a price increase before revenue falls.

Price increase ÷ ( 1 + Price increase ) = Revenue breakeven

Example: A 40% increase breaks even at a 28.6% loss of volume.

Watch for: It tests revenue only. Use margin breakeven when cost to serve matters.

The share of new customers the model assumes the higher price turns away.

1 − New customers ÷ New customers = Assumed loss of new customers

Example: 9,200 new customers against 10,000 is an 8% loss.

Watch for: The change is safe on this measure only while the assumed loss stays under the breakeven loss.

Contribution margin as a share of the old price.

Contribution margin ÷ Price = Margin share of price

Example: $4 margin on a $5 price is 80%.

The share of volume you can lose after a price increase before contribution falls.

Price increase ÷ ( Margin share of price + Price increase ) = Margin breakeven

Example: A 30% increase at an 80% margin breaks even at a 27.3% loss.

Watch for: Assumes cost to serve stays fixed in dollars.

The extra volume a price cut or discount must win to keep revenue flat.

Price cut ÷ ( 1 − Price cut ) = Breakeven volume gain

Example: A 20% cut needs 25% more volume. A 17.5% bundle discount needs 21.3%.

Watch for: The gain needed is always larger than the cut. It is a revenue test. When each unit has a cost to serve, keeping contribution flat needs a larger gain.

The rule that turns usage into a fee: what is counted and what each unit costs.

Also called: meter

How many units of a product one customer uses in a year: settled transactions, screened transactions, or a count you enter such as seats or API calls.

A year of one product's fees for this customer under one pricing model.

The share of each transaction's amount charged as a fee.

Example: 0.50% of a $20 sale is $0.10.

A fixed fee for each unit, whatever the transaction amount.

Example: $0.07 per screened transaction.

The flat amount added to a percentage fee on every transaction.

Example: 2.9% + $0.30: the $0.30 is the fixed fee.

Watch for: On small tickets the fixed fee is most of the price.

A fee charged every month, whatever the usage.

Also called: platform fee, flat pricing

The number of units each month that the monthly fee already covers.

Watch for: A customer under the included amount pays the same as one who uses none.

The fee for each unit used beyond the included units in a month.

Tiered pricing where the units that fall in each tier are charged at that tier's price.

Example: 7,000 units at $0.10 to 1,000, $0.08 to 5,000, then $0.05: 1,000 × $0.10 + 4,000 × $0.08 + 2,000 × $0.05 = $520.

Watch for: Crossing into a cheaper tier lowers the price of the next units only.

Also called: tiered pricing

Tiered pricing where every unit is charged at the price of the tier the monthly total reaches.

Example: The same 7,000 units all pay the third tier's $0.05: $350.

Watch for: The total bill can drop when usage crosses into a cheaper tier.

Also called: volume-based pricing

The most a percentage fee charges on one transaction.

Example: 0.50% capped at $0.50 stops growing at a $100 sale.

Watch for: Above the cap the effective rate falls as tickets grow, with no change in price.

The dollar value of payments a customer settles in a year.

Also called: TPV, payment volume

The average dollar amount of one settled transaction.

Watch for: A single average hides the spread. Real accounts have a distribution of tickets.

Also called: AOV, average order value

Payments that complete.

Annual volume ÷ Average ticket = Settled transactions

Payment attempts that are screened, including ones that are blocked or fail.

Settled transactions ÷ Settle rate = Screened transactions

Watch for: There are always at least as many screens as settled payments.

Settled transactions as a share of screened transactions.

Example: At 92%, every 100 settled payments took about 109 screens.

Total fees divided by payment volume.

Watch for: It moves when ticket size or product mix moves, even if no price changed.

What the customer pays for every product in the bundle bought separately at published prices.

Also called: list total

What the customer pays for every product together at bundle prices.

What the customer pays at published prices for every product in the bundle except one.

Example: In the example, the bundle without Radar is Tax alone: $2,500,000.

The bundle must never cost less than a smaller set of its products at list price.

Example: In the example at a $100 ticket, the bundle is $2,326,087 and Tax alone is $2,500,000, so the rule fails.

Watch for: It has to hold at every ticket size. A check at a few sample tickets can miss a failing range.

Also called: two-product floor

New revenue when an account that buys the other products adds one more through the bundle.

Bundle total − List total of the other products = Attach gain

Example: In the example, a Payments + Tax account adding Radar: $3,630,435 − $2,500,000 = $1,130,435.

Watch for: It nets off the discount the account now gets on the products it already had.

The cost per unit above which attaching a product through the bundle loses money.

Attach gain ÷ Annual units = Breakeven cost to serve

Example: $1,130,435 ÷ 27,173,913 screens = $0.0416 per screen.

Watch for: It is the zero-contribution line. A required margin sets a lower ceiling.

How much less the bundle costs than the same products bought separately.

1 − Bundle total ÷ A-la-carte total = Bundle discount

Example: 1 − $3,630,435 ÷ $4,402,174 = 17.5%.

A group of accounts that responds to the bundle in the same way, defined by which products they already use.

The share of a segment's accounts that move to the bundle.

Revenue lost when an account already paying list price for every product moves to the bundle.

A-la-carte total − Bundle total = Give-up

Example: $4,402,174 − $3,630,435 = $771,739 a year.

Watch for: It assumes the account pays list today. A large account with a negotiated discount gives up less.

Also called: cannibalization, dilution

The share of an account's contribution that the bundle discount removes, for an account already paying list for every product.

Give-up ÷ ( A-la-carte total − Cost to serve ) = Contribution given up

Example: $771,739 ÷ ($4,402,174 − $1,043,478) = 23.0%, against a 17.5% discount.

Watch for: The cost to serve does not fall with the price, so the share of contribution is always larger than the discount. It is only as good as the cost estimates.

Existing full-price revenue that a new offer replaces with discounted revenue. The give-up, summed across accounts.

An account adopting a product it did not use before.

Also called: cross-sell

New attaches needed for each cannibalized account to keep revenue flat.

Give-up ÷ Attach gain = Attach breakeven ratio

Example: $771,739 ÷ $1,130,435 = 0.68 Radar attaches per three-product account.

Watch for: It needs no account counts. It is a revenue test, and contribution also depends on cost to serve.

The cut in the processing rate that would cost the same as the bundle discount, in basis points.

Give-up ÷ Annual volume × 10,000 = Processing-rate equivalent

Example: $771,739 ÷ $500,000,000 = 15.4 bps.

Watch for: One basis point is 0.01% of payment volume. Net revenue after network costs is a small share of volume, so the same dollars are a much larger share of net revenue.

The change in the share of quoted deals won because the bundle is on offer.

The share of eventual take-up reached by each quarter.

Watch for: Existing accounts can move at once while new attaches build slowly, so the early quarters look worst.

Attach gains and new deals, less the give-up, across all segments for a year.

Net revenue impact less the cost to serve the newly attached products.

A range of transaction sizes treated as one line, such as $50 to $100.

Watch for: Rates still vary inside a band. Narrower bands give a cleaner split.

Forecast revenue divided by forecast units, across all lines.

Forecast ÷ Volume = Forecast average price

Watch for: It is a blended price. It changes when the forecast mix changes.

Actual revenue minus forecast revenue.

Actual − Forecast = Total variance

Example: $58,900 actual − $64,400 forecast = ($5,500.)

Total variance as a share of forecast revenue.

Total variance ÷ Forecast = Variance against forecast

Example: ($5,500) ÷ $64,400 = −8.5%.

The size of variance, as a share of forecast revenue, at or above which a miss needs a written explanation.

Example: At a 5% threshold, an 8.5% miss is material and a 2% miss is not.

Watch for: The threshold is a judgment set by the team that owns the forecast. A dollar threshold usually sits beside the percentage, and either one can trigger the explanation.

A revenue variance is favorable when actual is above forecast and unfavorable when it is below.

Watch for: The labels flip for costs, where coming in under forecast is favorable. A favorable driver can hide an unfavorable one of the same size.

A table or chart that starts at forecast revenue, adds each driver's variance, and has to end at actual revenue.

Example: $64,400 forecast − $3,220 volume − $1,330 mix − $950 price = $58,900 actual.

Also called: waterfall, walk

The part of the miss from selling more or fewer units in total, at the forecast average price.

( Volume − Volume ) × Forecast average price = Volume variance

Example: (9,500 − 10,000) units × $6.44 forecast average price = ($3,220.)

Also called: price-volume-mix (PVM), rate/volume analysis

The part of the miss from selling a different blend of lines than forecast, at forecast prices. For each line: (actual units − total actual units × the line's forecast share) × the line's forecast price.

Example: More annual plans and fewer monthly plans than forecast.

Also called: price-volume-mix (PVM)

The part of the miss from each unit selling at a different price than forecast. For each line: (actual price − forecast price) × actual units.

Watch for: Calculated last, at actual units. Calculating price first, at forecast units, moves dollars between price and mix. The total does not change.

Also called: rate variance, price-volume-mix (PVM)

Price variance caused by a change in the published price. For each line: (actual list price − forecast list price) × actual units.

Price variance caused by discounts and promotions off the published price.

Price variance − List price variance = Realization variance

Watch for: It is everything in price variance that a list price change does not explain. Currency effects and a shift in ticket size inside a band land here too.

Also called: discount leakage

The check that the lines add up to reported actual revenue, and that volume, mix, and price variance add up to the total variance.

Volume variance + Mix variance + Price variance = Total variance

Watch for: Volume, mix, and price always add up to the total, because all four are calculated from the same lines. The check that can fail is the one against reported actual revenue. A difference there means a line is missing, double counted, or mispriced.

Also called: tie-out, control total

Actual revenue for the period as finance reports it, taken from the ledger and not from the lines.

Watch for: Match the period and the scope of the lines. Refunds, credits, and revenue timing can make the ledger differ from units × price.

New customers gained or lost because visits differed from forecast.

( Volume − Volume ) × Conversion rate = Traffic effect

New customers gained or lost because the conversion rate differed from forecast.

( Conversion rate − Conversion rate ) × Volume = Conversion effect