Scientific Advertising

1. 1Key every ad before it runs. Give each ad, headline, offer and publication its own coupon code, and record it next to the returns. The passages treat keyed advertising and traced returns as the way principles get proved, and note that brief ads are never keyed.
2. 2Choose the reply action that matches the product. Offer a sample, a book or a free package to draw direct replies. Where direct returns are impossible, compare one town with another and measure by cost of sales.
3. 3Lower the cost of the inquiry before you compare ads. If a sample must be requested, have coupons presented at a local store or taken by telephone instead of mailed. In the passages, mail inquiries averaged 70 cents each, store-presented coupons 18 to 22 cents, and telephone brought 70 percent of inquiries on one line.
4. 4Compute three costs for every keyed ad: cost per reply, cost per customer, cost per dollar of sale. Cost per reply only shows how much action the ad produced. One ad may bring many worthless replies and another fewer valuable ones, so final conclusions rest on cost per customer or per dollar of sale.
5. 5Change one variable per comparison and log it with the result. Run the identical ad with different headlines, then compare settings, sizes, arguments and pictures. The source cites a headline change multiplying returns five or ten times, and nearly two thousand keyed headlines on a single product.
6. 6Record the headline type beside every return. Over time this shows which appeal draws most readers (beauty, disease prevention, cleanliness). Keep the weaker appeals running if they still pay, but set the proportion of ads by proven pull.
7. 7Control sample waste and follow up fast. Print 'Only one sample to a home' and 'adults only' on the coupon. Have dealers send coupons in promptly and follow up while interest is fresh. Check repeaters only if checking costs less than the loss; the source found it did not on packages priced 10 to 50 cents.
8. 8Promote a result to a fixed principle only when one method invariably proves best across repeated tests. Then measure every departure from it, so you know what your preferences cost per customer.

The source describes a single product with keyed returns on nearly two thousand headlines. The stories in the ads were nearly identical, yet returns varied enormously, and a headline change alone can multiply returns five or ten times. Each keyed return is filed with the headline used. The record shows which quality readers want most: beauty, disease prevention, or daintiness and cleanliness. An appeal that pulls half the returns of another is still kept if it is profitable, but the share of ads given to each appeal follows its proven pull. In a parallel test, sample inquiries obtained by mail averaged 70 cents each, while the same ads brought inquiries at 18 to 22 cents when coupons were presented at a local store. Across these tests, cost per customer, not cost per reply, makes the final call.

• If two ads show similar cost per reply, then compare cost per customer or per dollar of sale before choosing, because worthless replies can hide in the cheaper one.
• If a sample must be requested, then route the coupon to a local store or telephone instead of mail. Mail brings minimum returns, and stores drew about four times as many presentations.
• If you want to depart from proven mail-order copy, then key the departure against the proven version and treat the extra cost per customer as the price of your preference.
• If an opinion comes from an advertiser who cannot show traced returns, then give it no weight in your decisions.
• If checking repeaters or cheats costs more than the samples they take, then stop checking and limit the offer instead ('one sample to a home', 'adults only').

• Judging by cost per reply alone: the tell is a low-cost coupon ad that produces few customers or little sales volume.
• Running brief or untraced ads and copying other advertisers' untraced ads: the tell is that you cannot say what any ad returned.
• Changing several elements at once so a result cannot be attributed: the tell is a winning ad whose cause you cannot name.
• Making the prospect pay for the sample, or requiring mail-in coupons: the tell is that three in four to nine in ten refuse, and inquiry cost runs far above the store-presented rate.
• Drawing conclusions from too few results or too soon, or scattering samples to homes uncounted: the tell is that you cannot trace the samples to any customer.

• ▸Ad key code: the unique code printed on the coupon or tied to the publication/town.
• ▸Variable tested: the single element changed (headline, offer, size, picture, argument).
• ▸Headline type/appeal: the category of appeal used (beauty, health, cleanliness, economy).
• ▸Ad cost: space plus art plus production for this keyed ad.
• ▸Replies received: the coupons returned, split by channel (mail, store, telephone).
• ▸Cost per reply: ad cost divided by replies.
• ▸Customers converted: replies that became buyers, traced back to the key.
• ▸Cost per customer / per dollar of sale: ad cost divided by customers, and ad cost divided by sales dollars.
• ▸Verdict: winner, loser or retest, with the proportion of future ads assigned to this appeal.

1. 1Define the one class of prospect your product interests, and the reasons it interests them. The headline's job is to pick those people out of a crowd, so write it to hail them only. A blind or clever headline may draw many times as many readers, but mostly impossible prospects.
2. 2Test the headline by the salesman's standard: would it work as the first words said to a stranger you want to stop in a crowd, like "Hey there, Bill Jones"? If it only amuses, or conceals what you sell, rewrite it.
3. 3Make the headline or the picture state the benefit the prospect wants: economy, beauty, labor savings, good things to eat and wear. Readers skip about three-fourths of what they pay to read, and they will not discover that your product interests them unless the headline or picture says so.
4. 4Write many headlines and discard most of them. The source's author spends hours on a single headline and throws away scores before choosing, because the whole return depends on attracting the right readers.
5. 5List every appeal the product supports (for the source's product: beauty, disease prevention, daintiness, cleanliness). Write a headline for each and run them in separate keyed ads with near-identical body copy, so the headline is the only variable.
6. 6Read the keyed returns and record the headline beside each result. Rank the appeals by response. Do not trust the ones you like best, because you do not know enough people to average up their desires.
7. 7Allocate headline types in proportion to what pays. A weaker appeal that still returns half as much as the best one stays in use. Vary the ads across publications, because circulations overlap and different readers respond to different approaches.
8. 8Set the body in ordinary 8-point type and make every picture earn its space. Large type and decoration raise the cost of your story without gaining worthwhile attention. Size a picture by its selling importance, as mail order advertisers do after comparative tests.

The source's author holds keyed returns on nearly two thousand headlines for a single product. The stories in the ads were nearly identical, yet returns varied enormously, and a headline change alone can multiply returns five or ten times. The product fostered beauty, prevented disease and aided daintiness and cleanliness, so each appeal became a headline family. Tracking showed which quality most readers sought, and the other appeals stayed in use because even one bringing half the returns of another was still profitable. The source gives parallel illustrations. On a soap, "Keep Clean" or "No animal fat" draws few, "It floats" interests more, and a beauty or complexion headline draws many times as many. On an automobile, a headline about a good universal joint falls flat because few buyers think of universal joints, while "The Sportiest of Sport Bodies" might out-pull it fifty to one. The lesson is to test the appeal before you spend on the story. The picture is held to the same test: a mail order man added a row of chickens in silhouette to an incubator ad, increasing his space 50 percent. The ad looked striking, but his cost per reply rose, since the text ads with the right headlines had already produced excellent returns.

• If a headline would not stop the typical buyer face to face, then rewrite it. Cleverness, slogans and blind headlines fail this test.
• If a headline appeals to a feature few buyers think about, such as a universal joint, then lead with the benefit buyers already want and move the feature into the body.
• If you cannot key the ad, then do not trust your favorite headline. Run keyed variants first, because liked appeals rarely prove best.
• If a picture does not sell, then cut it or shrink it. A larger or decorative picture that cuts returns per dollar costs you more per sale.
• If a weaker appeal still returns profitably, then keep it in rotation at a lower share, and spread appeals across publications because circulations overlap.
• If the only argument for large type or a bigger display is attention, then stay with 8-point body. People read all they care to in it, and louder type multiplies cost.

• Writing the headline to amuse or to be distinctive. The tell: it draws praise from colleagues but no keyed replies, and the readers it attracts are not buyers.
• Leaving the headline blind or vague to create curiosity. The tell: the right prospect has no idea the ad concerns something they want, like a woman never reading an article headed "Egyptian Psychology" that is about her.
• Choosing the headline by personal taste after drafting only one or two. The tell: no discarded drafts, and no keyed comparison against other appeals.
• Buying attention with space, type or borders. The tell: the ad grows, borders and white space appear, and cost per reply climbs while the story stays unchanged.
• Using decorative pictures that tell no selling story. The tell: you cannot say what benefit the picture promises to the buyer, as with the silhouette chickens that raised cost per reply.

• ▸Target prospect: one typical buyer, in a sentence, and the reason your product interests them.
• ▸Candidate appeals: every benefit the product supports (e.g. beauty, economy, labor saving).
• ▸Draft headlines: at least 10 to 20 across appeals; note which you discarded and why.
• ▸Face-to-face test: would this stop that buyer if spoken in person? Yes/No for each shortlisted headline.
• ▸Keyed test plan: which headline runs in which publication or key code, with body copy held constant.
• ▸Picture check: what benefit does the picture show, and is its size justified by its importance?
• ▸Returns by headline: replies and cost per reply for each key code.
• ▸Winning appeal and share: top performer plus a profitable weaker appeal, with the share of ads each gets.
• ▸Next round: the headline variants to test against the winner.

1. 1Replace the seller's question with the buyer's. Before drafting, ask: would this line help a salesman sell the goods if a buyer stood in front of him? Cut every slogan, clever conceit and flourish that fails the test. An ad is salesmanship in print, not entertainment.
2. 2Learn what buyers want before you write. Sell house to house, interview hundreds of possible customers, or send questionnaires. The maker knows manufacturing and dealers, and that knowledge misleads him about customers. Guesswork is expensive.
3. 3Build the offer on service, not on a request to buy. Give wanted information, cite advantages to users, and offer a sample, a first package bought for them, or goods on approval. Drop 'Buy my brand,' and often drop the price and the dealer list.
4. 4Remove the buyer's risk, and make the trial free of payment, not only refundable. 'Try it for a week, money back' is weaker than 'Try it a week, then pay if you like it.' A neighbor-dealer's signed warrant is stronger than a stranger's guarantee. Offer a trial with a demonstrator or a sample.
5. 5Replace generalities with specific, checkable facts. Swap 'abundant lather' for 'softens the beard in one minute' and 'creamy fullness for ten minutes.' Swap 'lowest prices' for 'our net profit is 3 percent.' Where value is hard to judge, state the cost, as with the $100,000 formula. Do not rest on cheapness; people want bargains but not cheapness.
6. 6Tell the complete story when the goal is an immediate sale. Mail-order experience says 'the more you tell the more you sell.' Give a reasonable story even in small space, and fill the space you buy. Do not impose a word limit on a prospect who is interested.
7. 7Make the headline show the interest, since only the interested read. Write many headlines, discard scores, and spend hours on one. Appeal to what readers actually seek (beauty, economy, labor saving), not what you like best.
8. 8Spend only what each element earns. Set type at ordinary 8-point size. Use pictures that sell the item and are sized by importance. Add a coupon as a reminder, because readers forget a decision to act within five minutes. Key every ad and keep the proof.

A motor maker found selling by solicitation difficult. On good advice he stopped asking for a purchase. His ad offered to send a motor to any home, through any dealer, for one week's use. A man came with it to show how to operate it. The ad said, 'Let us help you for a week without cost or obligation.' The passage reports that about nine in ten of the trials led to sales. The change was not louder copy or cleverer wording. The offer moved from 'buy' to service, and it put no risk or payment on the prospect. The same logic runs through the cigar makers ('Smoke ten, then keep them or return them') and through the horse parable: the seller who said 'come and pay me then' won the buyer over the one who offered money back.

• If the copy asks for the order before the reader has any proof, then restructure it around service: information, a sample, or a trial.
• If a claim could appear in any competitor's ad ('abundant lather,' 'lowest prices'), then replace it with a specific, measurable fact or a disclosed number.
• If a guarantee is common in the category, then strengthen it with a pay-after-trial offer or a named local dealer's signed warrant.
• If a layout element (large type, border, silhouette, clever line) cannot be shown to sell, then remove it and spend the space on the story.
• If you cannot say which headline appeal pulls best, then run the same ad with different headlines and key the returns before choosing, because your favorite is rarely the winner.

• Writing to please the seller: copy talks about the brand name, the company, and 'Buy my brand.' The tell is that the ad has no stated benefit to the user.
• Performing instead of selling: slogans, conceits, and amusement. The tell is that people praise the ad but the keyed returns are poor, or that it attracts entertainment seekers.
• Vague, common claims that carry no belief: every rival says the same thing. The tell is superlatives with no number, time, or cost.
• Decorating at the expense of the offer: the incubator ad added 50 percent more space for a chicken silhouette. The tell is that cost per reply rises by the same percentage with no added sales.
• Cutting the story short by theory, such as 'be very brief,' or leaving the space half used. The tell is a half-page of copy set in a page of space, which doubles the cost per return.
• Choosing appeals by personal taste. The tell is a headline on a minor feature, such as a universal joint, that few buyers think about.

• ▸Buyer want: what this reader seeks (economy, beauty, labor saving), and the evidence from interviews, questionnaires, or keyed returns.
• ▸Salesman test: would each major line help a salesman sell face to face? List any line that fails and cut it.
• ▸Service offer: the information, sample, or trial you give instead of asking for the purchase.
• ▸Risk reversal: money-back, pay-after-trial, or a dealer's signed warrant. Which one, and who stands behind it?
• ▸Specific claims: list each claim as a number, time, cost, or fact. Rewrite any generality.
• ▸Value proof: a stated cost, price, or disclosed margin that lets the buyer judge value.
• ▸Headline candidates: at least 10 headlines, each showing a different appeal, with the one chosen and why.
• ▸Story completeness: does the copy answer every question a prospect standing before you would ask, and does it fill the space bought?
• ▸Action reminder and key: the coupon or response device, plus the code that traces returns to this ad and headline.

1. 1Define the action before you write: name the single response you want (cut a coupon, request a sample, write in, buy) and write the ad backward from it. The passages measure ads by the 'amount of action which each ad engenders', so the action must be observable and countable.
2. 2Build the response device into the ad. Mail order ads usually carry a coupon because 'a large percentage of people who read an ad and decide to act will forget that decision in five minutes.' The coupon is a reminder the reader cuts out 'when ready to act'. Make it easy to find and fill in.
3. 3Offer something that induces a direct reply: a sample, a book, a free package. Put the offer where the reader decides, and give 'enough to get action' in the copy above it. Do not cap word count. Write as you would speak to a prospect standing in front of you.
4. 4Give the action a reason tied to the reader. Limit the offer to a class (veterans, lodge members, executives), because 'those who are entitled to any seeming advantage will go a long way not to lose' it. Where it fits, state that something already belongs to the reader, such as a book with his name on it, awaiting his request.
5. 5Remove risk from the request. Ship on approval or invite trial, since 'people are honest' and losses are small. Say 'Try our rivals' too' if substitution is bleeding your response.
6. 6Key every ad: give each headline, ad, and publication its own code on the coupon, so replies trace back to the version that produced them. Run the identical ad with different headlines. The record cites changes that 'multiply returns from five or ten times over'.
7. 7Judge by cost per customer or cost per dollar of sale, not raw reply count. One ad may bring 'too many worthless replies, another replies that are valuable.' Record the headline type beside every keyed return so you learn which appeal draws the most action.
8. 8Write for the new reader only. Treat each reader as a one-time reader and use the full ad to get action now. Spend no lines on present users unless the headline does it.

An incubator seller by mail ran type ads with the right headlines and got excellent returns. He then believed a striking picture would raise them, so he increased his space 50 percent to add a row of chickens in silhouette. The ad looked striking, but his cost per reply rose. The keyed returns showed the added space bought no more action. A second case: an automobile ad led with a headline about a good universal joint and fell flat, while the same ad headed 'The Sportiest of Sport Bodies' might outpull it fifty to one. In both cases the reader's action, counted through keyed replies, settled the argument that taste could not. In a third case, a books advertiser who was losing money added one touch, the buyer's name in gilt lettering on each book, with scarcely another change. The ads then sold some hundreds of thousands of books.

• If a creative idea would not help a salesman close a buyer in person, then cut it and rely on a reason to act.
• If an element (larger type, border, picture) cannot show a better cost per reply in a keyed test, then do not pay for its space.
• If two ads draw different volumes of replies, then choose by cost per customer or per dollar of sale, not by reply count.
• If your headline would draw people who cannot buy, then narrow it to the buyers you seek, even if a vague one draws more readers.
• If the offer is general and response is weak, then limit it to a defined class or tie it to something that belongs to the reader.

• Writing to entertain: the ad is admired and clever, yet replies come from amusement seekers rather than buyers. The tell is a high reply count with a poor cost per customer.
• No reminder device: interest forms and fades within minutes. The tell is readers who say they meant to respond but never did, and no traced returns.
• Spending on space that does not pull: large type, borders, decorative pictures. The tell is a doubled space bill with replies that rise by less than double, so cost per sale climbs.
• Not keying ads: you cannot say which headline or publication produced replies. The tell is relying on the appeals you like best, which 'will rarely prove best'.
• Aiming at current users or a mass audience: copy addresses existing customers or 'people in the mass'. The tell is blurred copy with no clear person to act.

• ▸Desired action: state the one response you want (coupon, sample request, letter, purchase).
• ▸Typical buyer: describe one individual likely to want this, not a mass.
• ▸Headline and appeal type: write the headline and name its appeal (beauty, curiosity, economy, etc.).
• ▸Offer and reason to act: state the sample, book, or approval offer and any class limit or 'belongs to you' hook.
• ▸Reminder device: describe the coupon or reply form and where it sits.
• ▸Key code: record the unique code on the coupon for this ad and publication.
• ▸Replies and cost per reply: log replies received and total ad cost divided by replies.
• ▸Customers and cost per customer / per dollar of sale: log actual buyers and cost per customer or per dollar of sales.
• ▸Verdict and next test: record whether to keep, kill, or alter one variable (headline, offer, space) in the next ad.

1. 1Define the ad's only job as making sales, then judge it as you would a salesman. Write down its cost and its actual result, and accept no excuse a good salesman would not offer. Reject 'general effect,' 'keeping your name before the people,' and 'aiding other salesmen' as justifications.
2. 2Key every ad so each sale traces back to its source. Use coupons, a distinct offer, or a distinct address for each ad. Where direct returns are impossible, compare one town with another and measure by cost of sales. Without a key you are paying for a guess.
3. 3Offer something that induces a direct reply: a sample, a book, a free package, or a product on approval. Record the action each ad produces. Let the offer carry the service, not a plea to buy.
4. 4Do not stop at cost per reply. Follow replies through to customers and dollars of sale. One ad may bring many worthless replies and another fewer valuable ones. Base final conclusions on cost per customer or cost per dollar of sale.
5. 5Compare variants against each other and record every result. Test headlines, settings, sizes, arguments, and pictures. Keep a record that pairs each keyed return with the headline used. A headline change alone can multiply returns five or ten times.
6. 6Cut cost wherever the keyed returns show waste. Trace space size, type size, and pictures. Doubling space by doubling type size may still be profitable, but traced returns show you are paying double for the sales. Use every line, drop decoration and entertainment, and size pictures by their selling importance.
7. 7Stop paying for lost decisions. Add a coupon as a reminder, because many readers who decide to act forget within five minutes. Measure how much it recovers.
8. 8Keep a weaker-appealing ad running if it still pays. One appeal may bring half the returns of another and still be profitable. Match the share of ads using each headline appeal to its proven yield. Vary ads across publications because circulations overlap and different readers respond to different approaches.

The source gives a five-dollar article sold by mail. One ad's replies cost 85 cents each. A second ad, which its author thought was better, brought replies at $14.20 each. A third ran for two years at an average of 41 cents per reply. At 250,000 replies a year, the gap between $14.20 and 41 cents is the difference between ruin and profit. Cutting the 85-cent cost in half was worth a great deal. Running the $14.20 ad without a key would have meant paying for sales at 2 to 35 times what they needed to cost. Because the replies were keyed, the ranking was visible at once. The practical move is to kill the $14.20 ad, keep the 41-cent ad, and then test whether those replies become customers and dollars of sale.

• If an ad has no key on its returns, then treat its cost as unknown and do not scale spend until you can trace each sale to its ad.
• If an ad wins on cost per reply but its replies do not become buyers, then judge it by cost per customer or cost per dollar of sale and discard the reply count.
• If one headline pulls far more than another, then shift most of the volume to it, but keep weaker appeals that still return a profit.
• If a design choice is added to please the seller (larger type, borders, clever slogans, entertainment), then keep it only if keyed returns show it earns its space.
• If you cannot trace direct returns, then compare one town against another and measure by cost of sales.

• Judging ads by general effect or name recognition. The tell: the justification cites awareness or 'keeping the name before the people' and offers no cost per sale.
• Trusting your own taste in appeals. The tell: you pick the headline you like best. The appeals you like best will rarely prove best, because you do not know enough people to average their desires.
• Stopping at cost per reply. The tell: a low cost per inquiry coexists with weak sales, because the replies are worthless.
• Spending large sums on a guess with no key. The tell: you cannot say which ad or headline produced a given sale. This is the habit that costs 2 to 35 times what sales need cost.
• Writing for the seller instead of the buyer. The tell: copy says 'buy my brand' or 'drive people to the stores' and offers the reader no service, sample, or information.

• ▸Ad / headline ID: name the ad and the exact headline, so every keyed return pairs with its appeal.
• ▸Key used: coupon code, address, or town used to trace returns to this ad.
• ▸Publication and space cost: total spend on this insertion, including space size.
• ▸Replies received: count of keyed replies traced to this ad.
• ▸Cost per reply: space cost divided by replies.
• ▸Customers / sales resulting: count of replies that became buyers.
• ▸Cost per customer: space cost divided by customers; this is the deciding figure.
• ▸Cost per dollar of sale: space cost divided by sales revenue.
• ▸Verdict and next test: keep, change, or kill, and name the single variable to test next (headline, size, picture, offer).

1. 1Name the habit you are asking people to break. State it as a behavior: 'uses a rival soap and is satisfied with it', 'bakes beans at home', 'does not brush teeth'. The passages state that people do not quit habits without reason, so you cannot plan until the habit is named.
2. 2Classify the field. A line that created a custom and typifies it with consumers is nearly impregnable. A field where every possible customer already uses a rival soap is only less difficult. Entering the first takes a convincing advantage or very superior salesmanship-in-print. Entering the second takes a strong appeal that wins people from long-established favor.
3. 3Size the true target. Canvass homes and dealers before you write. The pork-and-beans canvass found only 4% used any canned beans and 96% baked their own. A 'buy my brand' appeal reaches only that 4%. The real job was to win people away from home-baked beans.
4. 4Test the cost of conversion, with every ad keyed. Run small test campaigns and trace the cost per converted customer. The dentifrice tests put converts to the toothbrush habit at $20 to $25 each, partly because much of the advertising went to people already converted. Compare that figure with what a customer returns in sales. Your rule is that what cannot be done profitably on a large scale cannot be done on a small one.
5. 5Check who captures the new business. The toothbrush educator's new customers were shared by his rivals, so his sales did not rise in line with his spending. If you pay to change a category habit and rivals collect the gain, drop the objective.
6. 6Find the reason to switch. Imagine one typical person, such as a man on a Pullman using his favorite soap. Ask what you could say to him in person to make him change. If you have no exclusive feature, search for something impressive that others have not told, because you need at least a seeming advantage.
7. 7Answer the objection in full. A person wedded to a brand will not switch on a name or a sentence. Write the complete story, even if it takes columns. Aim at adoption of your brand, not a first trial package to please you.
8. 8Decide: attack, redirect, or abandon. Attack the entrenched field only with a tested advantage. Redirect to a different habit if canvass shows one is weaker. Abandon the objective if the traced cost per convert exceeds what the customer is worth.

A dentifrice maker wants to grow sales. The obvious move is to teach people to brush their teeth. Keyed tests in the source show that each such convert costs $20 to $25, because much of the advertising reaches people who already brush. That cost is 'unthinkable' since a lifetime of sales might not repay it. The maker who learned this from tests put 'not one line' in any ad toward the habit and built what the source calls a remarkable success on keyed ads. A second maker spent heavily to make toothbrush converts. His new business was shared by his rivals, and he wondered why his sales rose 'in no way commensurate with his expenditure'. The lesson is to price the habit change before choosing the objective. If the price is above the customer's value, sell to people who already hold the habit and win them from rivals with a stated advantage.

• If the objective requires changing a habit or custom, then get a traced cost per convert from small keyed tests before committing budget.
• If the traced cost per convert exceeds what the customer returns in sales, then drop the objective, because what fails small fails large.
• If the new demand you create is shared by rivals, then do not fund category education. Sell your brand to people who already hold the habit.
• If almost every prospect already uses a satisfied rival product, then enter only with a specific advantage and a full-length story, and test it on one typical person first.
• If a canvass shows most of the market uses a different habit, such as home-baking, then aim the appeal at that habit and not at rival brands.

• Funding altruistic habit education. The tell: sales rise far less than spend, and rivals' sales rise too.
• Writing 'buy my brand' to a market that does not use the category. The tell: the appeal reaches only a small fraction of people, like the 4% of canned-bean users.
• Assuming the product is 'much like others' and has nothing to say. The tell: you offer a name, a boast, or a brief claim to a person wedded to a rival. The source says there is nearly always something impressive that others have not told.
• Treating prospects as a mass. The tell: you cannot say what you would tell one Pullman soap user to make him switch.
• Judging by untraced ads or others' opinions. The tell: you cite brief ads as proof that short copy works, although those ads are never keyed.

• ▸Habit to break: describe in behavioral terms what the target does now and which rival brand or custom it involves.
• ▸Field type: custom-creating and near-impregnable, or crowded with satisfied rival users? Give your evidence.
• ▸Canvass result: share of the market using your category versus the share using an alternative habit.
• ▸Reason to switch: the specific advantage or untold fact you can offer, stated as you would to one person face-to-face.
• ▸Test design: the keyed ads, the test market, and the cost of the test.
• ▸Traced cost per convert: the figure from keyed results.
• ▸Customer value: expected sales per customer over time.
• ▸Spillover check: the share of new business that rivals will also capture.
• ▸Decision: attack, redirect, or abandon, with the one-line reason.

1. 1State the purpose before drafting: this ad exists to make profitable sales, and nothing else. It is not for general effect, not to keep your name before people, and not primarily to help your other salesmen. Write the sentence at the top of the brief and judge every later choice against it.
2. 2Pick one typical buyer, a single man or woman likely to want what you sell. Do not picture people in the mass, because that gives a blurred view. Assume the headline has already gained this person's attention.
3. 3Learn the buyer's side before you write, because the maker's knowledge of manufacturing and dealers misleads him about customers. Sell the article from house to house, interview hundreds of possible customers, or send out questionnaires. You are learning which arguments draw reactions and which factors don't appeal.
4. 4Run the two-question test on every element: 'Would it help a salesman sell the goods?' and 'Would it help me sell them if I met a buyer in person?' Apply it to slogans, clever conceits, length, type size, unusual style and dress. If you can't imagine a customer being impressed by the element face to face, cut it.
5. 5Set length by what the prospect needs, not by a word cap. You wouldn't limit a salesman with a prospect standing in front of him to a set number of words. Give readers enough to get action, since the only readers you have are people your subject already interests.
6. 6Strip out performance and keep plain talk: brief, clear, convincing. Fine writing and a unique literary style take attention from the subject and reveal the hook, and visible selling creates resistance. Don't be amusing, don't boast, don't show off. Do what a good salesman does with a half-sold person before him.
7. 7Price the salesman. Figure the ad's cost and its result, compare it with your other salesmen, and accept no excuses a good salesman wouldn't make. Per-word cost can run to $10, and a mistake multiplies across all your trade, so be more exacting than you would be with one salesman.
8. 8If you pay for something that doesn't sell, such as dress, pride or distinctiveness, do it knowingly and know its cost. If returns disappoint, go back to a good mail order ad as the model and cut the waste.

The brush maker with some 2,000 canvassers shows the frame at work. Selling from house to house looks very difficult, and it would be if the men asked housewives to buy. They don't ask. They go to the door and begin with a service offer instead of 'Buy my brand.' The passages treat this as plain salesmanship: the good salesman does not merely cry a name, he pictures the customer's side until buying is the natural result. The electric sewing machine motor maker applied the same test to his print ad. He stopped soliciting a purchase and offered to send a motor through any dealer for a week's use, with a man to show how to operate it, 'without cost or obligation.' About nine in ten of the trials led to sales. Neither seller was judged on cleverness or applause. Each was judged on what a good salesman would do with a prospect at the door, and on actual sales.

• If an element would embarrass or handicap a salesman standing in front of a buyer (loud voice, clever conceit, odd dress, word limit, slogan), then cut it from the ad.
• If you can't name the typical buyer and what he or she wants, then stop writing and go sell or interview first, because guesswork is very expensive.
• If the copy says 'buy my brand' or talks about your profit or your name, then rewrite it around service to the buyer: wanted information, advantages to users, a sample or trial without cost or risk.
• If a style choice is defended as pleasing, distinctive or admired, then ask whether it brings sales. Keep it only as a knowingly paid-for indulgence, with its cost known.
• If the ad is judged by applause, cleverness or the seller's pleasure, then judge it again by cost and sales result against your other salesmen.

• Performer's ad: the writer seeks applause instead of sales. The tell is that people praise the style while sales don't move, and any entertainment-seekers you attract aren't your buyers.
• Seller-centered copy: the ad says 'Buy my brand' or 'Give me the trade', or blazons a name. The tell is that the copy is full of 'we' and no reader benefit appears.
• Mass-audience thinking: the ad is written for 'the public', so the view is blurred. The tell is that you can't describe one buyer in a sentence.
• Imposed handicaps: arbitrary brevity, oversized type or shouting headlines, none of which you'd demand of a salesman. The tell is that the stated reason is 'people won't read' rather than 'the buyer needs this to act', and the cost of the story rises without a gain.
• Maker's-eye conception: the manufacturer's knowledge of production and dealers stands in for knowledge of customers. The tell is that nobody has interviewed or sold to a prospect before the ad is planned.

• ▸Sales purpose: write the one profitable sale this ad must make, not awareness or name display.
• ▸Typical buyer: describe one individual who likely wants this and what they want.
• ▸Buyer research source: record the house-to-house selling, interviews or questionnaires behind your picture of the buyer.
• ▸Face-to-face version: write what a good salesman would say and do with this half-sold prospect in person.
• ▸Two-question test: for each element (headline, slogan, length, type, style, illustration), answer 'Would it help a salesman sell?' Yes or no.
• ▸Elements cut: list slogans, conceits, show-off or boasting that fail the test.
• ▸Service offered: state the wanted information, user advantage, or sample or trial that serves the buyer instead of asking for a purchase.
• ▸Cost and result: record ad cost and actual sales, and compare with your other salesmen.
• ▸Knowing indulgences: list any pride or style items kept, with their estimated cost.

1. 1Start from the reader's selfishness. Write one line: 'What does this person get for themselves?' Drop any sentence that serves the seller, such as 'Buy my brand,' 'Give me the trade you give to others,' or 'Let me have the money.'
2. 2Choose the service form. The passages name these: wanted information, a stated advantage to users, a sample, an offer to buy the first package, or a product sent on approval. Pick the one that lets the customer prove your claims without cost or risk.
3. 3Remove the ask. The best ads ask no one to buy. Often they omit the price and omit the statement that dealers handle the product. Cut the demand to purchase and the name-blazoning. Let the offer carry the ad.
4. 4Make the offer concrete and bounded: what is sent, for how long, at what cost to the reader. Example: 'Let us help you for a week without cost or obligation,' or 'Smoke ten, then keep them or return them, as you wish.' Where possible, add the help the product needs, such as a man to show how to operate it.
5. 5Push the risk onto yourself. Free trial with the money paid after the week proved many times more impressive than 'try it, and we'll refund your money.' Sending goods with no money down ('Pay in a week if you like them') beats a refund promise. Losses stay small because people are honest.
6. 6Give the reader a way to respond now. Put a coupon in the ad as a reminder to cut out, because a large share of readers who decide to act forget within five minutes. Offer a sample, a book, or a free package to induce direct replies.
7. 7Put the offer in the headline. People skip about three-fourths of what they read and will not read business talk unless the headline shows it is worth their while. State the benefit and the offer where the eye lands.
8. 8Key the response and judge it by cost per customer or cost per dollar of sale, not by reply count alone. One ad may bring too many worthless replies while another brings valuable ones. Compare offers against each other and keep the winner.

The maker of an electric sewing machine motor found advertising difficult while he asked people to buy. On good advice he stopped soliciting a purchase. He offered to send a motor to any home, through any dealer, for one week's use, with a man to show how to operate it. The ad said, 'Let us help you for a week without cost or obligation.' There was no price pitch and no demand to buy. The offer was described as resistless, and about nine in ten of the trials led to sales. The same logic runs through the brush canvassers who say 'I was sent here to give you a brush' and let the housewife pick one, and the coffee wagon man who leaves a half-pound package and returns only to ask how she liked it. In each case the reader's self-interest, not the seller's, carries the sale.

• If your draft says 'buy,' 'ask your dealer,' or leads with the brand name, then rewrite around what the reader receives. The best ads ask no one to buy.
• If the product is hard to sell on claims alone, then offer a trial, sample, or approval shipment so the customer proves it at no cost or risk.
• If you are choosing between 'money back if dissatisfied' and 'pay after a week,' then choose 'pay after a week.' It proved many times more impressive, and losses are very small.
• If readers need a reason to stop and read, then make the headline state the benefit or offer. People will not read your business talk unless it is worth their while.
• If two offers both generate replies, then keep the one with the lower cost per customer or per dollar of sale, not the one with more replies.

• Writing the ad to please the seller. The tell: it talks about your interest, your brand, your profit, and uses phrases like 'Drive people to the stores.' People can be coaxed but not driven.
• Blazoning the name as if it mattered. The tell: the brand name is the biggest element and the reader has no reason to care about it.
• Making a risk-laden offer. The tell: the reader must pay first and trust a refund promise, while the stronger version, trial first and pay later, is left unused.
• Offering a free item with no response mechanism. The tell: no coupon or reminder, so readers who meant to act forget within five minutes.
• Judging the offer by reply volume alone. The tell: many inquiries but few customers, because cheap free offers attract worthless replies and cost per customer was never computed.

• ▸Reader's self-interest: what does the reader get for themselves, in one line?
• ▸Offer form: information, stated advantage, sample, first package bought, or trial on approval.
• ▸Offer terms: what is sent, for how long, and at what cost or obligation to the reader.
• ▸Risk transfer: who carries the risk (refund promise, or goods sent and pay later)?
• ▸Seller-centered lines removed: list any 'buy,' name-blazoning, or 'drive to stores' phrases cut.
• ▸Headline: does it state the benefit or offer so a skimmer sees it?
• ▸Response device: coupon or reply mechanism, and the key code.
• ▸Success measure: cost per customer or cost per dollar of sale, and the comparison offer.

1. 1Name the one kind of person who can want your product, and the reason they want it. The passage states it flatly: "What you have will interest certain people only, and for certain reasons. You care only for those people." Write that person and that reason in one sentence before you draft any headline.
2. 2Write the headline as a hail, the way you would say "Hey there, Bill Jones" to one person in a crowd. A reader in your target group must recognize at a glance that the ad refers to something they want.
3. 3Run the newspaper-headline test. Readers pick what to read by headline and skip whole pages. Ask whether your target reader would stop at this headline, or pass it the way the beautiful woman passes "Egyptian Psychology."
4. 4Reject any blind headline or clever conceit that draws a bigger crowd. It may attract "many times as many" readers, but they are "mostly impossible subjects." Count only readers who could buy, not total attention.
5. 5Draft many headlines and discard most. The source writer spends hours on a single headline, and scores are thrown out before the right one is chosen. Write at least a couple dozen, each hailing a different group or a different reason.
6. 6Match the headline to one appeal from the product's range. The cited product fosters beauty, prevents disease, and aids daintiness and cleanliness. Each appeal hails a different class of reader, so choose one per ad.
7. 7Key the headlines and compare them on identical copy. Returns can differ five- or ten-fold from the headline alone. Judge by cost per customer or cost per dollar of sale, not raw replies, because one ad may bring "too many worthless replies."
8. 8Keep a record of every headline beside its keyed return. Use the record to set what share of your ads should hail each class of reader. Keep weaker appeals that still pay, since one appeal may bring half the returns of another and remain profitable.
9. 9Vary the ads across media. With twenty magazines you may run twenty separate ads, so each headline can hail the circulation it reaches.

The source describes a writer with keyed returns on nearly two thousand headlines for a single product. The story in the ads was nearly identical, yet returns "vary enormously, due to the headlines." The product had several uses: it fosters beauty, prevents disease, and aids daintiness and cleanliness. Each headline hailed a different class of reader, and the keyed record showed which quality most readers sought. The writer then set the proportion of ads whose headlines should attract each class. Weaker appeals stayed in the mix because even half the returns of the best appeal still paid. The lesson is that selectivity gets measured: the headline picks the audience, and the returns show which audience is worth hailing.

• If a headline would attract readers who cannot buy, then drop it, however clever. Selectivity beats crowd size.
• If your target reader would skim past the headline in a newspaper, then rewrite it until it names or signals their interest plainly.
• If two headlines hail different groups, then key them on identical copy and judge by cost per customer, not by reply count.
• If a secondary appeal returns about half of the best one and still pays, then keep it in the mix and cap its share of the ads.
• If you run many media with different circulations, then write a separate headline for each circulation instead of one for all.

• Writing a blind or clever headline for broad attention. The tell: many replies or compliments, but few buyers and a high cost per customer.
• Letting a headline mislead. The tell: readers who click or cut the coupon are the wrong people, as in the woman who skips "Egyptian Psychology."
• Settling on the first headline. The tell: no discarded drafts, when the source expects scores to be thrown out and hours spent on one.
• Judging headlines by gut or taste. The tell: no keyed comparison, so you cannot say how much returns change, though a change can multiply them five or ten times.
• Using one appeal for all readers. The tell: the single headline is never varied by class or by medium, so you never learn which quality most readers seek.

• ▸Target reader: one sentence naming who can want this product.
• ▸Reason they want it: the one appeal (for example beauty, health, cleanliness, saving) this headline hails.
• ▸Who must be excluded: groups the headline may attract but who cannot buy.
• ▸Headline drafts: at least 20, each aimed at a different group or appeal.
• ▸Newspaper test: would the target reader stop at the headline, yes or no, and why.
• ▸Misleading check: does the headline conceal the interest it should reveal, yes or no.
• ▸Key code per headline: the identifier for tracing returns on identical copy.
• ▸Cost per customer or per dollar of sale: the result for each keyed headline.
• ▸Share of ads per appeal: what proportion of the ads each appeal should get, given the results.

1. 1Collect every claim you plan to make, then mark each as a generality or a specific. 'Best in the world,' 'Lowest price in existence,' 'Supreme in quality,' 'Pure,' and 'Abundant lather' are generalities. Readers discount them as 'claiming the expected,' and they make the reader doubt your other statements too.
2. 2Replace each generality with a measurable fact: a number, a time, a count, a percentage, a stated test. The source models the swaps: 'more light' becomes 'three and one-third times the light,' 'prices reduced' becomes 'reduced 25 percent,' 'quick shave' becomes 'a 78-second shave,' 'softens the beard quickly' becomes 'softens the beard in one minute.'
3. 3Use only figures you can stand behind, because a specific claim 'is either telling the truth or a lie.' Specificity carries weight because it signals that tests and comparisons were made. Run the test or pull the record first, then write the number.
4. 4When the claim is a competitive one that rivals can copy, such as lowest price, a guarantee, or an undersell promise, look for a fact that proves it indirectly. The mail order advertiser's 'Our net profit is 3 percent' made low prices evident without asserting them, and rivals could not echo it.
5. 5Mine your own process for concrete facts: what you paid, how many formulas you tested, what a hidden cost totals. Examples: '$100,000 for that formula,' 'testing and comparing 130 formulas,' hidden costs of $735 on a $1,500 car. State the fact flatly and let the reader draw the conclusion.
6. 6Where a quality can be shown, show the mechanism. The brewer who dropped the empty word 'Pure' pictured a plate glass room with beer cooled in filtered air and a filter of white wood pulp. Show the actual room, filter, or process in place of the adjective.
7. 7Test which specific claims draw readers by keying the ads and comparing returns by headline. Add every claim that proves its pull to a standing list, and include the full list in every ad, since the average reader sees you once.
8. 8Do a final read-through as the skeptical buyer. Strike any sentence that would sound the same in a competitor's ad. Keep only sentences that a competitor could not truthfully repeat.

A mail order advertiser selling women's clothing to poorer-class buyers used the slogan 'Lowest prices in America' for years. Rivals copied it. He then guaranteed to undersell any other dealer, and rivals copied that too, so both claims became commonplace. Under able advice he changed the statement to 'Our net profit is 3 percent.' This was a definite figure, and it proved very impressive. With his volume of business it was evident that his prices must be minimum, because no one could be expected to do business on less than 3 percent. The claim let the buyer infer low prices without being told to believe them. The next year his business made a sensational increase.

• If a claim could appear unchanged in a rival's ad, then it is a generality. Replace it with a number, a time, or a described fact before it runs.
• If a rival has copied your claim or could copy it, then move to a fact that proves it, as the 3 percent profit figure did for 'lowest prices.'
• If you have no tested figure, then run the test or find the record before writing. Do not round up or guess, since a specific claim is either true or a lie.
• If you are in a field where every maker uses the same stock phrases ('Pure,' 'Abundant lather,' 'Quick shaves'), then the first advertiser to state a measured fact wins the attention. Take that position.
• If you have several specific claims, then put them all in every ad, and use keyed headline results to decide which one leads.

• Superlatives dressed as copy: 'best,' 'finest,' 'supreme,' 'pure.' The tell: the line would read identically in every competitor's ad, and the bigger the type, the bigger the folly.
• Vague comparatives: 'more light,' 'prices reduced,' 'quick.' The tell: the sentence has a comparison or direction but no quantity, so the reader is left in doubt.
• Specificity about the wrong thing: a precise detail buyers do not care about, such as the universal joint in the automobile example. The tell: headline tests show weak pull, so check keyed returns and not your own taste.
• Copyable claims: a price or guarantee promise that rivals can match. The tell: competitors repeat it within a season and it turns commonplace, as the 'Lowest prices' and undersell guarantees did.
• Rationing claims across ads and assuming readers have seen the others. The tell: each ad omits some strong facts, though the average reader reads only one ad and what you leave out he may never know.

• ▸Original claim: copy the vague line exactly as drafted.
• ▸Generality test: could a rival print this unchanged? (yes/no; if yes, it must be replaced)
• ▸Hidden fact: what measurable fact lies behind it (time, quantity, percentage, cost paid, number tested)?
• ▸Proof source: the test, record, or invoice that backs the figure.
• ▸Rewritten claim: the sentence with the actual figure or described mechanism.
• ▸Copy-resistance: can a rival truthfully repeat it? If yes, what indirect fact (like the 3 percent profit) would prove the point instead?
• ▸Reader relevance: do typical buyers care about this fact? Note the headline or keyed-return evidence.
• ▸Placement: headline, subhead, or body, and confirm it appears in every ad.

1. 1Fix the reader before you write: picture one typical individual who is already half-sold, since your headline has gained his or her attention. Ask at every step, "Would it help a salesman sell the goods?" and "Would it help me if I met this buyer in person?"
2. 2Build the claim list from tests, not opinion. Run different headlines, compare the returns, and note which claims appeal most. In usual lines several claims appeal to a large percentage, so keep every claim that earns a place on the list.
3. 3Cover every phase of the subject in each ad. One fact appeals to some readers and another to others, so omitting any one loses the percentage who would have been convinced by it.
4. 4Put every listed claim into every ad, in the same ad. Do not spread claims across serials or run one claim per ad for brevity. The source warns that serials almost never connect and that people rarely read successive ads on one line.
5. 5Make each claim specific. Replace elastic generalities with definite statements; the passage contrasts "Used the world over" with "Used by the peoples of 52 nations." The two take the same room, but the definite one can be many times as effective. A generality carries no weight.
6. 6Size the story to the claims, not to a brevity rule. A sentence can complete the story for chewing gum or Cream of Wheat. A car, or a change of brand against years of habit, needs columns. Use any space you buy fully: half-page copy set in a page space doubles the cost without doubling the returns.
7. 7Spend space on selling, not decoration. Include a picture only if it is a better selling argument than the same space in type. Set body copy in ordinary 8-point type, and use a coupon so the reader who decides to act can do it before forgetting.
8. 8Key the ad and judge it by traced cost per reply or per sale. Brief ads are never keyed, while every traced ad that lasts tells a complete story. Ignore the practices of untraced advertisers.

A man wanted a personal car, cared little about price, and wanted something to take pride in, but he also wanted value. He leaned toward a Rolls-Royce and also considered a Pierce-Arrow and a Locomobile. Their ads were very short, and the makers evidently thought arguing comparative merits undignified. The Marmon told a complete story, and he read columns and books about it. He bought a Marmon and was never sorry. Later he learned facts about another car at nearly three times the price that would have sold him that car had he known them. The lesson: a buyer willing to read a volume will read it, and a maker who cries a name plus a few brief generalities loses the sale to the one who tells everything.

• If you are tempted to run one claim per ad or a serial, then put every proven claim into each ad, because the reader is unlikely to see the next installment.
• If an ad is short and unkeyed, then do not treat it as evidence for brevity; follow traced ads, which tell complete stories.
• If a claim is worth making, then state it in its most definite form ("52 nations," not "the world over"); drop any generality that carries no weight.
• If the product is a cheap, familiar line like chewing gum, then a sentence may complete the story; if it is a major purchase or a change of habit, then use columns.
• If a picture does not sell better than the same space in type, then cut it. If you enlarge the space, then fill it with selling copy, not blank room, borders, or bigger type.

• Serial or one-claim-at-a-time advertising: the tell is that each ad omits arguments that would convince some readers, and the series is expected to connect across issues.
• Name-crying with brief generalities, common in dignified, high-priced lines: the tell is an ad with no comparative facts for a buyer making a lifetime-scale purchase.
• Decorating with art: the tell is a picture added on a whim, as with the incubator ad, where 50 percent more space raised cost per reply by exactly 50 percent and brought no added sale.
• Mistaking your own boredom for the reader's: the tell is dropping claims because the ad feels monotonous to someone who reads them all, when a complete story is always the same.
• Taking advice from untraced advertisers: the tell is that the "brevity" or "loud type" rule comes from people with no keyed returns.

• ▸Typical buyer: describe one individual, not a mass, who is already interested.
• ▸Headline and attention claim: state what gains the reader's attention.
• ▸Tested claim list: list every claim that has proved its appeal through headline or return comparisons.
• ▸Claims in this ad: mark which of the listed claims appear; every one should.
• ▸Specificity check: rewrite each generality as a definite statement with a number or fact.
• ▸Phases of the subject covered: note any phase a certain percentage of readers would need that is missing.
• ▸Space and picture test: say whether each picture sells better than the same space in type, and whether all purchased space is used.
• ▸Response device: note the coupon or keyed offer that lets the reader act and lets you trace it.
• ▸Traced result: record cost per reply or per dollar of sale against the previous version.