Copywriting: Do the Research Before the Formula

The research comes first. A formula applied to a weak premise produces confident nonsense.

Search this term and you get the same page a dozen times. AIDA. PAS. FAB. One example each, and a closing line about practising daily.

The formulas are fine. They are also the last five per cent of the work.

A formula applied to a weak premise does not fail loudly — it produces confident nonsense. Copy that sounds persuasive and persuades nobody, because what it argues was never what the reader was worried about. Which pages and campaigns are worth writing at all is a planning question, and it lives in our content marketing strategy guide. This page is about the sentences.

One boundary, said once. Platform headlines, descriptions and character limits belong to ad copy. The wider discipline — web pages, email, product, long-form — is content writing. Taking a single post from question to published URL is blog writing. This page owns copywriting as a craft: writing whose job is to make someone act.

What copywriting is, and where it stops

Copywriting is writing built around a decision. Every line exists to move one specific reader a step closer to one specific action — click, reply, book, buy — and any line that does not do that is decoration, however true it happens to be.

That definition also tells you what to know first: who is deciding, what they are choosing between, and what holds them back.

What copywriting is not: clever writing, an accurate product description, or a page that hits a word count. A page can be all three and convert nobody.

Why a formula on a weak premise produces confident nonsense

A formula is a container. It arranges an argument in a sequence readers find easy to follow — but it has no opinion about whether the argument is correct, and it will carry a wrong one just as smoothly as a right one.

The failure in practice: a team writes a PAS page for a payroll product and agitates "payroll takes too long". The reader's actual problem was a compliance notice from three months ago that she is still paying for.

The page is fluent, on-brand and beside the point, and because it reads well nobody in the review calls it out. Weak premises survive reviews, which is what makes them expensive.

So the order runs the other way round from how it is usually taught: research, premise, formula, edit.

Find these four things before you write a line

Before drafting you need four things about one reader: what they already believe, the exact words they use for the problem, the objection they have not said out loud, and what made them start looking this week. Everything after that is arrangement.

The unsaid objection does the damage. Nobody writes in to say "I do not believe your case studies" or "I think this is a two-person company pretending to be twenty". They just do not reply.

Where to find all four without a research budget

You do not need a survey panel. Four sources you already own will give you the reader's beliefs, vocabulary and objections in one afternoon, and they are more honest than commissioned research because nobody was being asked.

1. Reviews — yours and your competitors'. Skip both extremes; three-star reviews are where people name the trade-off they accepted. Competitor reviews are the best free objection list there is.
2. Sales call notes and recordings. Listen for the question asked every time, and for how the prospect describes their situation before your salesperson reframes it. That sentence is your headline.
3. Support tickets, WhatsApp threads and DMs. Post-purchase confusion is pre-purchase fear with the mask off. If three customers asked whether the price included installation, your page has a gap.
4. The search box. Your site-search queries and Search Console data are literal transcripts of how people ask. Autocomplete, People Also Ask and the relevant subreddit or Quora thread fill the gaps.

One rule while collecting: copy the phrasing verbatim into a swipe file and do not tidy it. The moment you turn "it kept shutting down in the middle of billing" into "reliability issues", you have thrown away the sentence that would have worked.

Google's self-assessment for helpful content asks whether a page "clearly demonstrate[s] first-hand expertise and a depth of knowledge" (Google Search Central, last updated 10 December 2025). Research is how that shows up in a sentence.

Turn the research into one premise you can defend

A premise is one sentence that fixes what the copy argues, and nothing gets drafted until it exists. Write it in this shape: for a reader who believes X, our offer makes Y happen, and the thing standing in the way is Z.

Filled in, it reads like this: for a finance head who believes all payroll tools are basically the same, our service removes the compliance filings from her team's calendar, and what is in the way is her belief that migrating mid-year will break something.

Now the page has a job: prove the migration is safe. That is a very different page from one arguing that payroll is time-consuming.

Two checks. If two formulas both feel like a good fit, the premise is not sharp yet. And if you cannot say the offer in plain words, the problem is upstream of the writing — see how to build a value proposition.

Choose the formula by situation, not by habit

Formulas are tools, and each one assumes something about your reader's starting state. Pick the one whose assumption matches your premise; using your favourite regardless is how good research gets wasted.

Short copy — a button, a subject line, a WhatsApp message — usually needs no formula at all, just the premise said plainly. How it sounds is a separate, documentable decision: that is tone of voice, not structure.

Run the specificity pass: replace claims with consequences

A claim tells the reader what you think of your product. A consequence tells them what changes in their week. Readers discount claims automatically, because everyone makes them — so the first edit pass replaces every claim with the thing that actually happens.

The right-hand column only helps if it is true and checkable. Specificity you cannot stand behind is not better copy — it is a bigger liability, and in regulated categories a compliance problem.

Two habits speed it up: delete the adjectives and put back only the ones whose absence changes the meaning, and turn "we" sentences into "you" sentences.

The cut: the pass we run on our own copy

Most drafts get better by subtraction. This is the editing pass we run at Digital Hangover on our own pages and our clients' — it is a practice we follow, not a measured result, and no performance figure is attached to it.

1. Delete the first sentence. Read from the second. If nothing is missing, it was you getting started.
2. Read only the opening line of each paragraph. The argument should still hold. Where it breaks, that paragraph is hiding its point.
3. Strike every sentence that only restates the one before it. Restatement feels like emphasis while writing and reads as padding afterwards.
4. Count the asks. One page, one action. A second, competing call to action splits attention and wins neither.
5. Read it aloud at speaking pace. Anything you would not say to a client's face gets rewritten, not softened.
6. Halve it. Set the target before you start, not after.

The halving target is the one people resist and the one that works. Doing this to every page, every month, is the first thing a busy team drops — it is the work we take on in content marketing.

What copywriting cannot fix

Copy is a multiplier on an offer, not a substitute for one. It can make a good offer legible and urgent; it cannot make a bad offer worth buying. Rewriting the page is the wrong first move when the problem sits elsewhere.

• A bad offer. If the price, the terms or the risk are wrong for this buyer, better sentences only get more people to the point where they say no. Fix the offer.
• A mismatch between the ad and the page. Google Ads Help is blunt: "Choose a landing page that closely matches your ad and keywords", and it defines landing page experience as "How relevant and useful your landing page is to people who click your ad" (Optimize your ads and landing pages and About Quality Score for Search campaigns, Google Ads Help — neither page displays a last-updated date; both fetched 30 September 2026). If the ad promised a specific offer and the page opens with a generic pitch, the copy is not the failure — the continuity is.
• A product nobody wants. No sequence of words creates demand that does not exist. Persuasion works on a decision the reader was already open to making.
• The wrong traffic. Copy for a ready buyer, shown to someone months away, reads as pushy — a targeting problem in a copy problem's clothes.
• A page that gets in the way. An eleven-question form, a price that appears only at checkout, a button hidden on mobile — friction is a landing page design problem.

One honest limit on measurement: a copy change on a page with a handful of conversions a week cannot produce a readable test, so you get noise and read it as a result. Without the volume, A/B testing is the wrong instrument.

India: decision-maker copy and price-sensitive copy are different jobs

English-first copy for a decision-maker and copy for a price-sensitive consumer are two jobs with two different objections, and translating one into the other produces neither. It is the most common brief failure on Indian accounts selling to both.

The failure modes are symmetrical. A B2B page transcreated into Hindi keeps its English abstraction and lands as officious. A price-led line pushed at a procurement head reads cheap, and cheap gets negotiated on instinct.

Write each from its own research set: same offer, two premises, two vocabularies.

Hinglish works when it is the reader's actual register and fails the moment a marketer imitates one. If nobody says the sentence out loud, romanising it does not make it native.

Where to go from here

Take your worst-performing page and do the cheap version of this in one afternoon. Twenty quotes is enough to write from.

• Today: pull twenty customer quotes from reviews, tickets and sales notes into one document, unedited.
• Then: write the premise — believes X, we make Y happen, in the way is Z. If it needs two sentences, you have two pages.
• Then: draft in the formula that premise points to, run the specificity pass, and halve it.
• Before you ship: check that the ad or email sending traffic here still promises what the page now says.

If the new version does not beat the old one, you have learned something about the offer — a more useful finding than a better headline.