Marketing, oriented

Segmentation splits a market into groups whose needs differ enough that they would respond differently to the same offer. Groups that merely look different, but buy for the same reason at the same price, are a classification, not a segment. The bases trade cost against predictive power. Age and income are cheap and weak on their own, values and lifestyle explain more but are costly to measure, and behavior, meaning usage, occasion and the benefit sought, predicts best but needs data you may lack before launch.

A target has to pass four tests at once: you can measure it, it is big enough, you can reach it through some channel, and you can act on it. It also has to pay more than it costs to serve. A small budget facing large incumbents is usually best spent winning one segment completely, which gives you references for the next.

Positioning is the place the offer holds in the buyer's mind. Points of parity get you considered; points of difference win the choice, and a real one matters to the segment, can be delivered, and is hard to copy. The comparison buyers make often sits outside your category. A bookkeeping app for freelance designers may be weighed not against rival apps but against a shoebox of receipts and an accountant at tax time, so a campaign that argues against the rival app answers a question nobody asked.

The trap is targeting everyone. A message written for everyone has to drop every specific, and specifics are what make a claim believable. Broad copy is not broadly appealing; it is weakly appealing to each reader.

Rule: name the segment by what it would do differently, confirm you can reach it, and write the position against the alternative that buyer actually weighs.

The four Ps, product, price, place and promotion, are useful because they force four decisions to be made together. A premium product sold at a discount through a bargain outlet contradicts itself. Services add three more: the people who deliver, the process they follow, and the physical evidence a buyer can judge, because a service has nothing to inspect before purchase.

A product has layers: the core benefit someone buys for, the product delivered, and the augmented layer of support, warranty and onboarding, where rivals usually compete once the middle has converged. A difference counts only if a buyer can perceive it while choosing. Place means being available at the moment of need, not an address, and each channel layer takes margin in exchange for reach, control or credit.

Price begins with an objective, because one price cannot maximize share, margin and prestige at once. Cost-plus pricing ignores what buyers think the offer is worth and what the alternatives cost; value-based pricing captures the most and depends on knowing what the buyer compares against. Penetration pricing buys share quickly and works where scale lowers cost or switching later is hard; skimming takes early adopters first and teaches later buyers to wait. You learn elasticity by changing the price and watching volume, since stated willingness to pay is wrong in both directions.

The trap is the habitual discount. Say a candle maker runs 20 percent off every other weekend. Sale weekends sell well, but buyers learn that the list price is a ceiling and wait for the next sale, so full-price weekends thin out and the real price drifts down to the sale price.

Rule: check that all four Ps tell the same story, set price from an objective and the buyer's alternatives, and test it rather than ask about it.

A buyer's path is usually drawn in stages. Awareness has two jobs, making the problem known and making the brand known, and a market that does not see the problem needs education before any brand claim lands. Consideration is shortlist-building, so the work is being comparable on the buyer's criteria; content that pretends there are no alternatives reads as evasion. At the decision, what remains is usually risk, such as a contract, a migration or the fear of looking foolish, and removing a barrier converts better than adding a benefit. Retention is the cheapest revenue and the least resourced, because a customer who stays is a non-event. Advocacy is earned by the product, and it is the only stage whose output feeds the top.

That loop separates a funnel, which treats each customer as an exit, from a flywheel, which treats them as input to the next acquisition. A growth loop exists only when using the product produces something that attracts the next user, such as an invitation or a shared document. And a customer who never reached first value cannot be kept by any later campaign.

Adoption has its own pattern. Early adopters accept a promise, while the majority wants proof from people like themselves, so a campaign that worked early can stop working, and a product nobody can see in use spreads slowly however good it is.

The trap is the wrong-stage ask. Picture a payroll startup running "Book a demo" ads to people who learned the category existed thirty seconds ago. Few book, and the ad reads as pushy; a short explainer with a small next step would fit those readers.

Rule: identify the stage of the person in front of you, and ask only for the next step that stage can take.

A brand is the perception held in customers' minds, so a company with no brand strategy still has a brand, assembled from whatever buyers ran into. The logo is a trigger and the brand is what it triggers, which is why redrawing the mark rarely changes what people think. If what buyers believe differs in substance from the identity you project, the product is failing, and more communication makes it worse.

Brand equity is the premium a name earns beyond what the product does, and it shows up as a higher price, faster trial of new products and forgiveness after a mistake. Awareness splits into recognition, knowing the brand when you see it, and recall, naming it unprompted while buying; recall is what gets you onto the shortlist. Being thought of and being findable are separate failures: a brand people recall but cannot buy converts as badly as one stocked everywhere that nobody thinks of. Loyalty shows when a customer resists a competitor's discount, not in repeat purchase, which habit and inertia also produce.

Architecture decides how closely reputations are tied. One name across products shares both equity and risk; separate names contain both and pay for each build. An extension works when the brand's meaning fits the new category, and a poor fit blurs the parent as well as failing on its own.

The trap is refreshing for the sake of freshness. Imagine a juice brand known for its green bottle switching to white because the team is tired of green. Shoppers scanning the shelf for green no longer find it. Teams tire of their own assets long before the audience has noticed them.

Rule: hold the distinctive assets fixed, vary the execution, and judge any brand change by what customers would stop recognizing.

Primary research is data you collect for your own question; secondary research reuses data gathered for someone else's. Existing sources save time and suit general, quantifiable questions such as industry trends and household incomes, but their definitions, timeframe and population rarely match yours, so check all three before citing a number.

Interviews and focus groups explain mechanisms and cannot tell you how many; surveys and experiments measure size and cannot tell you why. So talk first, to learn what to ask, then count; surveying first fixes the answer options before you know what they are.

Every instrument shapes its answer. Question order, scale labels and the options offered all move results, and a leading question carries its conclusion inside it. The sampling frame decides what a result describes: surveying current customers about why people buy misses everyone who did not. People who answer differ from people who do not, so a low response rate warns about direction, not precision. Precision grows with the square root of sample size, so quadrupling the sample only halves the error. Let behavior, such as a price test, a preorder or an abandoned cart, outrank any stated intention. If you test demand by advertising a product you have not made yet, the Federal Trade Commission has advised that the ad must clearly say the product is only planned and may never ship.

The trap is counting from quotes. A founder interviews eight coffee shop owners, six say they would pay for a scheduling tool, and the pitch deck reports 75 percent demand. Eight interviews can reveal what owners struggle with; they cannot estimate a share of the market.

Rule: use conversations to find the question, numbers to size it, and observed behavior over stated intent whenever you can get it.

Attention comes in three kinds. Paid attention you control fully and it is trusted least; earned coverage you control least and it is trusted most; owned channels such as an email list sit between, with no algorithm between you and the reader. Paid search reaches people who have already stated a need, so it converts well and cannot create demand. Paid social reaches people who were not looking, so the creative has to establish relevance first. Retargeting re-reaches people already interested, which is why its reported conversions flatter it: many would have come back anyway.

A campaign needs one objective, because the creative, the targeting and the measurement all change with it. The same budget buys many people once or a few people often; recognition favors reach, and persuasion favors frequency.

Email is the owned channel with the clearest federal rules. Under Federal Trade Commission guidance, a commercial email must identify itself as an ad and include a valid physical postal address and a clear way to opt out, and the sender must honor an opt-out within 10 business days.

Borrowed audiences carry a duty. A material connection, meaning payment, free or discounted products, or a personal, family or employment tie, has to be disclosed with the endorsement itself.

The trap is a disclosure that technically exists. A skincare brand sends free serum to a creator, who posts a video and tucks "#ad" into a block of twenty hashtags in the description. Federal Trade Commission guidance says not to mix a disclosure into a group of hashtags, and to put a video's disclosure in the video, not only in its description.

Rule: pick the channel by the buyer's stage and purchase cycle, and place every disclosure where someone skimming would still see it.

Impressions count the times an ad was served, not seen, and dividing them by reach, the number of unique people, gives frequency. Click-through rate is clicks over impressions, cost per click is spend over clicks, and cost per mille is spend per thousand impressions.

Customer acquisition cost is the fully loaded spend, including salaries, tools and creative, divided by customers actually won. Lifetime value is the margin, not the revenue, expected over the relationship, and it rests on a forecast of how long customers stay. Their ratio says whether acquisition pays; the payback period says when, which is the cash question.

Work an example. A shop spends 2,000 dollars, gets 500,000 impressions and 5,000 clicks, and wins 50 customers. Click-through rate is 1 percent, cost per click is 40 cents, cost per mille is 4 dollars, and acquisition cost is 40 dollars on media alone. If each customer brings 15 dollars of margin a month, payback takes a little under three months.

Attribution models, such as first touch, last touch, linear and time decay, divide credit by assumption, so switching models reshuffles every channel's apparent performance. Only a holdout, a random slice deliberately left unexposed, answers whether the spending caused the outcome.

On a website, an engaged session is one that clears a set bar for duration, key events or pages viewed. Engagement rate is the share of sessions that were engaged and bounce rate is the rest, so broader, cheaper traffic can lower the rate while raising the count of engaged sessions.

The trap is reading a falling cost per click as progress. Suppose next month the same 2,000 dollars buys 8,000 clicks at 25 cents but wins only 30 customers. Cost per click fell by more than a third, and acquisition cost rose to about 67 dollars, because cheaper clicks are often less qualified ones.

Rule: judge every metric next to the one it trades against, and trust a channel's credit only after a holdout has tested it.

In the United States, the Federal Trade Commission reads an ad the way a reasonable consumer would, taking the words, phrases and pictures together. An express claim is stated outright; an implied claim is one a reasonable reader would draw, and you need proof for both. A water filter advertised as removing "the contaminants linked to illness" implies that its water keeps you healthier, without saying it. The proof has to exist before the ad runs: a reasonable basis, meaning objective evidence of the kind the claim calls for. Health and safety claims generally need competent and reliable scientific evidence, letters from satisfied customers usually are not enough, and a money-back guarantee is no substitute. Praise nobody takes literally, such as the world's best coffee, is puffery; the world's strongest coffee is a measurable claim.

A disclosure has to be clear, conspicuous and close to the claim it qualifies. Fine print cannot contradict the headline, so a banner promising no fees is not repaired by a footnote listing a monthly service fee. Customer testimonials must reflect typical results or say what buyers can expect, and "Your results may vary" does not do that. When a free item is tied to a purchase, the price of the purchased item should not be raised to pay for it.

Design and data follow the same logic: preselected boxes and hidden opt-outs obtain consent nobody meant to give, and sites aimed at children under 13 need verifiable parental consent before collecting their personal information.

The trap is believing that true sentences make a true ad. An ad built entirely from accurate statements can still leave a false overall impression, and the impression is what gets judged.

Rule: write down every claim a reasonable reader would take from the ad, including the implied ones, and hold the evidence for each before it runs.