E-commerce, oriented

Start with one identity. Average order value is revenue divided by orders, so revenue equals sessions times conversion rate times average order value: 20,000 sessions converting at 2% with a $50 average bring in $20,000. If revenue falls while sessions and conversion hold steady, the change is in order value, and a checkout redesign would target the wrong number. Name the denominator, too: the same 120 orders are 3% of 4,000 sessions but 4% of 3,000 users.

To find where shoppers leave, divide each funnel step by the step before. Suppose 10,000 shoppers view a product, 1,200 add it to the cart, 600 start checkout and 450 buy. The step rates are 12%, 50% and 75%, so the move from product page to cart loses the most people. Counts show where, never why. The reasons come from interviews and from usability tests, in which a few people think aloud while trying a set task.

An A/B test shows each visitor one of two versions at random and compares a success measure chosen in advance. Its sample size is fixed in advance too, and halving the smallest difference worth detecting roughly quadruples the visitors needed.

The common mistake is to watch a running test and stop the moment it looks significant, which inflates the chance of declaring a difference that is not there. Significance at 0.05 means a true no-difference case is flagged 5% of the time, not that the difference is large. Overall results can hide segments as well. If desktop converts at 4.0% on A and 4.8% on B, and phones at 2.0% and 1.8%, B wins overall on equal traffic, 3.3% to 3.0%, yet loses on phones.

Rule: name the factor that moved and the step that leaks before changing anything, then read a fixed-size test once, after the full sample is in.

An ad is judged by the overall impression its words and pictures create together, including what they only imply. So a tight close-up that makes a small item look large can mislead, and fine print elsewhere cannot fix a picture it contradicts. Every objective claim, stated or implied, needs a reasonable basis, meaning evidence in hand before the ad runs, and a money-back guarantee is no substitute. Leaving facts out can deceive when the gap gives a wrong impression, and some facts are required outright: an online listing for a textile product must say whether it was made in the USA, imported or both.

Behind the page sits product data. Each size or color is its own variant with its own ID, price and stock, and when the same attributes drive variant pages, filters and search listings, one wrong value can show up in all of them.

Urgency has to be true. FTC staff list as dark patterns a countdown clock that resets when it hits zero, inventory said to be low when it is not, and items added to the cart without permission or by a pre-checked box. Showing a case beside a phone leaves the choice with the shopper; preselecting it does not.

The common mistake is to trust a star average on its own. A 5.0 from three reviews is a far shakier estimate than 4.6 from 900, which is why a rating should travel with its review count. Say a store withholds every review under four stars: its average now misrepresents what buyers think. The FTC's consumer reviews rule bars claiming or implying that the reviews shown are all or most of those submitted while suppressing low-rated ones. Withholding on neutral grounds applied to every review, such as abusive content, is still allowed.

Rule: read the page as one claim, and check that each photo, stock note and rating matches the real item, the real inventory and what buyers actually wrote.

Shoppers give up on forms that look long, complex or directionless, and the payment step is a frequent breaking point. Every field is work for the shopper, so the guidance is to settle what the order truly needs and drop the rest: one name box instead of three, and a billing address that defaults to the delivery address. Guest checkout removes a separate sign-up task, and FTC staff list forced registration among dark patterns; an account can be offered after payment instead, linked to the order just placed. On a phone, the right input type brings up the right keyboard, and autocomplete values let the browser fill a saved address or card.

To find the leak, log each step as its own event. Suppose 1,000 shoppers begin checkout, 700 add shipping details, 600 add payment and 540 buy. The shipping step loses 300 people, more than the other two steps combined, so that screen gets studied first. Count with care, since in Google Analytics the add-to-cart count rises with every item added, not once per cart.

Fees are where the rules come in. Drip pricing, as FTC staff use the term, advertises part of a price and adds mandatory charges late, such as a convenience fee that first appears at checkout. Staff say unavoidable, mandatory fees belong in the upfront price. The common mistake is to read the FTC's fee rule as covering every store. It covers only live-event tickets and short-term lodging, and its total price may leave out shipping and government charges, so the staff warning and the rule are two sources with two different reaches.

Rule: ask of every field, screen and fee whether the order needs it, find the leaking step from the counts, and check who a rule covers before applying it.

Gross margin is gross profit divided by net sales: an item sold for $50 that cost you $30 leaves $20, a 40 percent margin. Markup divides by cost instead, so a pen bought for $5 and sold for $7.50 carries a 50 percent markup but only a 33 percent margin. Contribution margin goes further and subtracts every cost that rises with each sale.

Follow one invented order through. Add $6 of shipping and $2 of payment fees to that $50 order, and variable cost reaches $38, leaving $12 of contribution, or 24 percent. Break-even units are fixed costs divided by contribution per unit, so $6,000 a month of fixed costs needs 6,000 / 12 = 500 such orders a month. Lifetime value estimates the profit a customer brings over the whole relationship: at an expected 2.5 orders it is $12 times 2.5, or $30, while counting revenue instead, $125, overstates it more than fourfold. If winning that customer cost $40, the customer acquisition cost (CAC), each one returns $30 of contribution on $40 of spend. Count CAC against first-time buyers only, because dividing by every buyer that month, repeat customers included, understates it.

The common mistake is to read return on ad spend (ROAS) as profit. ROAS is the revenue ads bring in divided by what they cost, counted before the cost of goods, shipping, fees or the ads themselves. Spend $1,000 on ads that bring $3,000 of sales at a 20 percent contribution margin: the sales leave $600, so the campaign loses $400 despite a ROAS of 3. Break-even ROAS is 1 divided by the contribution margin, so the 24 percent order above needs a ROAS of about 4.2 just to cover its ads.

Discounts bite the same way: 20 percent off the $50 item leaves $40 against the same $30 cost, halving gross profit to $10, so the store must sell twice the units to earn the same total.

Rule: put every revenue-based figure, whether ROAS, order value or lifetime value, back into contribution before you judge it.

Paid ads are billed by the platform, for example per click or per thousand impressions. Organic search placement is earned, not bought: Google says it does not accept payment to rank pages higher. An owned channel, such as your own email list, reaches people you can contact directly, but you have to attract them first.

Attribution decides which touch gets credit, and the common mistake is to read that credit as cause. Under last click, the final channel clicked before a purchase gets all of it. Say a shopper clicks a social ad, later a search ad, then buys: last click credits search alone, while a linear model gives each half. Retargeting shows the trap. Its ads reach people who already visited your site, and last click credits them whenever their click comes last, even for shoppers who were coming back anyway. Incrementality asks the causal question by comparing actual sales with an estimate of sales without the campaign: $150,000 against an estimated $125,000 means $25,000 was incremental.

A marketplace supplies the shoppers and sets the rules, while a store you run yourself controls the relationship but has to draw its own traffic. Marketplace costs stack in layers. Invented example: a $40 item with a 15 percent referral fee ($6), $5 of fulfillment and $4 of ads costs $15 to sell, 37.5 percent of the price. Read the FTC's 2023 complaint against Amazon as allegations, not findings. It alleges that nearly 98 percent of purchases go through the Buy Box, where one featured offer gets the buy buttons, and that offers lose eligibility when Amazon finds a lower price on certain rival sites.

Rule: treat an attribution report as a credit rule, not a cause, and price a marketplace sale with every fee layer, not the commission alone.

The U.S. commercial email law sorts messages by primary purpose. An email that only confirms a purchase the buyer already agreed to is transactional, exempt from most of the law but still barred from misleading routing information. A message mainly promoting a product is commercial: among other things, it needs an accurate subject line, a working opt-out and a postal address, and an opt-out must be honored within 10 business days.

The common mistake is to assume a cart reminder is transactional because it concerns the shopper's own cart. The transactional category covers a transaction the recipient has already agreed to, and the FTC reads it narrowly. A shopper who left a cart has not completed the purchase, so the reminder is judged by its primary purpose like any other message. Texts face a higher bar: marketing texts sent with an autodialer need prior express written consent.

Retention figures need their own care. Churn is the share of customers lost over a period. If 120 of 2,000 subscribers active on March 1 cancel in March, March churn is 6 percent, and reading a monthly rate as annual can badly understate yearly cancellations.

An auto-renewing subscription is a negative option: the seller treats silence or failure to cancel as acceptance. For online sales, federal law requires disclosing all material terms clearly and conspicuously before taking billing details, getting express informed consent before charging, and offering a simple way to stop charges. The FTC's 2024 click-to-cancel rule was vacated by the Eighth Circuit in July 2025, reinstating the older rule, whose text the FTC restored in February 2026. The FTC's Negative Option Rule itself covers only prenotification plans such as book clubs.

Rule: judge each message by its primary purpose, not by what triggered it, and compare retention rates only when their period and definition of a lost customer match.

The FTC's Mail, Internet, or Telephone Order Merchandise Rule requires a reasonable basis for any ship time you state or imply, at the moment you make it. If you state none, you need a basis to ship within 30 days, or 50 if the buyer applies to you for credit, and the clock ends at shipment, when goods reach the carrier. Outsourcing does not move the duty: FTC staff say the seller is liable when its fulfillment house or drop shipper breaks the rule.

When an order will miss its date, you must tell the buyer reasonably quickly and by that deadline. The notice gives a definite revised date or says you cannot, and offers a way to cancel, at your expense, for a full and prompt refund. Silence can count as consent only to a first delay of 30 days or less. When the rule requires a refund, store credit does not count.

The common mistake is to treat a speed claim as an average. FTC staff say a seller claiming it ships in 48 hours most of the time must ship, or send a delay notice, within 48 hours on every order, because shoppers read the claim as a promise.

Inventory math keeps such promises keepable. The reorder point covers demand during the lead time plus safety stock. Selling 20 units a day with a 10-day supplier lead time and 60 units of safety stock, you reorder when on-hand plus on-order stock falls to 20 x 10 + 60 = 260. Returns travel the other way, inspected and then routed to resale, repackaging or disposal, and federal law bars selling a recalled consumer product.

Rule: state only a ship time you have a basis for, meet it on every order, and remember that the notice duty stays with you whoever packs the box.

Accepting cards obliges a merchant to meet the card industry's data security standard, which the card brands enforce, and compliance does not guarantee against a breach. The simplest defense is holding less: the FTC's data-security guide says not to keep account numbers or expiration dates without an essential need, since data not in your system cannot be stolen. Tokenization replaces card details with a randomly generated number that stands in for them, and HTTPS belongs on every page, not only at checkout.

Fraud screening is where base rates mislead. Invented example: of 10,000 orders, 100 are fraud, and a filter catches 80 of them but also flags 1 percent of the 9,900 good ones. It blocks 99 real buyers, so fewer than half its flags are fraud, while approving everything would score 99 percent accuracy and catch none.

The legal side rewards reading each source for what it is. The FTC's consumer reviews rule makes it unlawful to write or sell fake reviews. Nexus is the link to a state that lets it make a seller collect and remit its sales tax; the Supreme Court once required physical presence, but overruled that in South Dakota v. Wayfair (2018), finding sufficient nexus under a law reaching only sellers above a sales or transaction minimum. And the Justice Department's 2022 guidance that Title III of the Americans with Disabilities Act covers what businesses open to the public offer on the web says itself that it has no binding legal effect.

The common mistake is to treat an FTC complaint as a finding. Against Ecommerce Empire Builders, the FTC alleged that promised monthly profits from done-for-you stores never materialized; it is the 2025 settlement order that bars the defendants from unsupported earnings claims.

Rule: before acting on a federal statement, name its kind, whether rule, guide, staff report, complaint or guidance, and read it only within the scope and dates it gives.