██████╗ ██╗██╗ ██╗███████╗██╗ ██╔══██╗██║╚██╗██╔╝██╔════╝██║ ██████╔╝██║ ╚███╔╝ █████╗ ██║ ██╔═══╝ ██║ ██╔██╗ ██╔══╝ ██║ ██║ ██║██╔╝ ██╗███████╗███████╗ ╚═╝ ╚═╝╚═╝ ╚═╝╚══════╝╚══════╝ ███████╗████████╗██████╗ ███████╗███████╗████████╗ ██╔════╝╚══██╔══╝██╔══██╗██╔════╝██╔════╝╚══██╔══╝ ███████╗ ██║ ██████╔╝█████╗ █████╗ ██║ ╚════██║ ██║ ██╔══██╗██╔══╝ ██╔══╝ ██║ ███████║ ██║ ██║ ██║███████╗███████╗ ██║ ╚══════╝ ╚═╝ ╚═╝ ╚═╝╚══════╝╚══════╝ ╚═╝
# Plain-text mirror. Same words as the human page, minus everything a browser needed.
The techniques on this page still work, and they are no longer where the advantage is. A machine can give you clarity, brevity, an active verb and a clean AIDA structure in nine seconds. What it cannot give you is the one paragraph only you could have written: a number off your own invoices, a client you actually lost, a method you can name, a job you turn down. Learn the frameworks so you stop wasting words. Then spend the words you saved on something specific enough that nobody else could publish it.
Until this morning, this page opened with the words “Hey there! Have you ever wondered how some ads just make you want to buy something right away?” Forty-odd words of throat-clearing at the top of an article about persuasive writing. I am quoting it back rather than quietly deleting it, because that sentence is the exact failure the rest of the page is supposed to prevent.
I run Pixel Street, a design and branding studio in Salt Lake, Kolkata. We write copy for brands like Coca-Cola, ITC and Marico, and we use AI in every workflow we have, from design to code to content. That last fact is why this page needed rewriting. The craft has not changed since 2024. The economics have. Producing clear, correct, well-structured, entirely forgettable copy now costs about nothing, which means every tip below that was really a tip about production has quietly lost most of its value.
The techniques on this page still work, and they are no longer where the advantage is. A machine can give you clarity, brevity, an active verb and a clean AIDA structure in nine seconds. What it cannot give you is the one paragraph only you could have written: a number off your own invoices, a client you actually lost, a method you can name, a job you turn down. Learn the frameworks so you stop wasting words. Then spend the words you saved on something specific enough that nobody else could publish it.
Google put this in writing on 15 May 2026, in its guidance on how content surfaces in generative AI search. Its own worked example contrasts a piece called “7 Tips for First-Time Homebuyers” with one called “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line”. An article called Top 10 Copywriting Tips is on the wrong side of that comparison by default. I know. Hold me to the last section.
Clear, concise, compelling, credible. It is a checklist rather than a theory, and it is most useful as an editing pass on something you have already written.
Jargon is the usual culprit, but vagueness is the more expensive one. “Bespoke digital solutions” survives every grammar check and tells a buyer nothing. If a sentence would be equally true of your three closest competitors, it is not clear, it is just grammatical.
Brevity is a ratio, not a word count. The fastest way to find the fat is to read the draft aloud and stop at the first place you feel bored, because the reader stopped several seconds before you did.
Connect the benefit to something the reader already wants, and give them a reason to act now rather than later. This is the hardest of the four to fake. Urgency you invented is transparent, and it costs you the fourth C.
The one I would put first if the letters allowed it. Name the client. Give the real figure. Cite the source. Say what you are not good at. In a market where anybody can generate a flawless-sounding claim in seconds, checkable specifics are the only thing separating you from the generator.
Frameworks are scaffolding. They stop you staring at an empty page and they stop you burying the ask in paragraph nine. None of them is a law, and it is worth knowing that most of them come from working ad men rather than from research, so treat them as inherited craft rather than evidence.
The oldest of them, and a small lesson in checking the things everybody knows. Nearly every article on copywriting attributes AIDA to the American advertising man E. St. Elmo Lewis in 1898. Akinori Iwamoto of Kansai University went back through the primary material for the Japan Marketing History Review and concluded that “the dominant hypothesis that E. St. Elmo Lewis invented and formulated AIDA is unsupported or based on weak evidence”, with nothing to show Lewis invented any prototype of it in or around 1898. Iwamoto credits Frank H. Dukesmith and Arthur Frederick Sheldon instead. I mention it because the Lewis attribution is repeated confidently in a thousand places and traces back to almost nothing, which is the same failure mode as the marketing statistics I have spent this month deleting from the rest of this blog.
The framework itself is sound. Its real weakness is structural: it assumes a linear reader who arrives at the top and proceeds downward, which describes almost nobody on a website.
Attention: a headline that promises something specific, not a clever one.
Interest: evidence you understand the reader’s situation better than they expected.
Desire: the benefit made concrete. Not “save time” but “stop rebuilding this every quarter”.
Action: one ask, in plain words, with the next step visible.
[IMG: Copy Writing Tips - PAS]
Illustration: Piyaporn Daenphetploy on Vecteezy. The credit here used to be a Google image-redirect URL that carried a tracking string and no longer resolved to the artwork; both credits on this page now point at the source.
Name the problem, make the cost of ignoring it vivid, then present the fix. It is the most effective framework here and the most abused. The agitation step is where honest copy turns into manufactured anxiety, and readers in 2026 have been agitated at professionally for years. My rule: only agitate a problem the reader would recognise without your help.
A translation exercise more than a structure. The feature is what it is, the advantage is what it does, the benefit is what changes for the reader. Most weak product copy stops at the first column. Most overwritten product copy skips straight to the third and asks readers to take the middle on trust.
[IMG: 4P's of Copy Writing TIps for Content Writers]
Illustration: GetGenie AI.
Promise a result, paint the picture of it, prove it, then push for the decision. The third P is the one that earns the other three, and it is the one most copy skips. If you cannot supply proof, weaken the promise rather than dressing up the push.
Where they are, where they could be, and how you get them across. It is PAS with the anxiety removed, which makes it the better choice for a considered purchase where the reader is already aware of the problem and is comparing suppliers rather than discovering a need.
Not a persona document. The specific objection the last person raised on a sales call, in their words. A sentence a buyer has said out loud always outperforms a sentence you invented for them.
Short words, short sentences, one idea per paragraph. You will see the advice to write for a grade-eight reading level repeated everywhere as though a study established it; I could not find one, so I will not quote a grade. Your reader is busy, half-attentive and often reading in their second or third language, which in India is the normal case rather than the exception. Write accordingly.
A headline makes a specific promise the page then keeps. Curiosity gaps get the click and poison the next visit. One test: could a competitor put their logo on it? If yes, rewrite it.
“You must try the product” beats “The product must be tried by you”. Active voice is shorter and names the actor. It is a default, not a rule: passive is correct when the actor is unknown, irrelevant, or deliberately downplayed — and businesses reach for it most often in the third case, which is precisely when readers notice.
Persuasion is mostly evidence, arranged well. “Get started” and “see pricing” only work once the reader believes you, so spend more effort on the belief than on the button.
Stories work because they carry specifics that a claim cannot. Four years ago I made a cold call to ITC with a fancy deck and zero credibility. Today we design for them, and for Coca-Cola and Marico. That sentence describes exactly one agency in Kolkata. A generated sentence about “delivering excellence for leading brands” describes every agency in the country, which is another way of saying it describes none of them.
Lists help readers scan and hurt readers who need an argument. Use them for genuinely parallel items — features, steps, options — not to escape the work of connecting two ideas. Google’s 2026 guidance is explicit that there is no requirement to break content into tiny pieces for AI search, so fragmenting a good argument into bullets is a loss, not an optimisation.
Cut first, polish second, and do the cutting on a different day if you can. The three passes I run: delete every sentence that survives only because it sounds professional, replace every adjective doing the job of a fact, and check that the first sentence of each paragraph still makes sense if the reader skipped everything else.
A/B testing is real; most small-business A/B tests are not. At the traffic a typical Indian SMB page sees, a headline test needs weeks before it can separate a real improvement from noise, and calling a winner early is worse than not testing, because you then roll it out everywhere. If you cannot reach a decent sample, do the cheaper thing: read the copy to five actual customers and watch where they stop.
Read books rather than threads. Our reading list of branding books, with the current edition of each checked covers the messaging end, and the design reading list covers the page the copy has to live on.
We use AI on copy every day, so I would rather be specific than pious about it.
It genuinely helps with: the first ugly draft you will throw away, which is the part most people procrastinate over; compressing something you have already written by a third; generating twenty headline variants so you can see the shape of the space; turning meeting notes into a rough structure; adapting one proposition into the six lengths a campaign needs; and second-language editing, which for a Kolkata studio writing for global brands is not a small thing.
It reliably fails at: anything requiring a fact only your business holds. Ask a model for proof and you get a plausible statistic, correctly formatted, attributed to a real-sounding publication, that nobody ever published. Our rule is that every number goes back to a primary source before it ships, and this refresh is what that rule looks like in practice: more than thirty figures have come off this blog because the trail ran out. One of the most-quoted marketing statistics in existence turned out to rest on a law review footnote whose own citation reads, verbatim, “(citation omitted)”. Another is a real Harvard professor’s sentence about human thought, with the words “purchasing decisions” swapped in by a magazine headline.
The second failure is subtler and it is the one that should worry you commercially. A model trained on the whole internet writes the average of the internet. That is not a flaw you can prompt your way out of — it is what the tool is for. So the moment you use it unedited, you publish something indistinguishable from your competitors’ pages, at exactly the moment when being distinguishable is the only thing search and AI assistants reward. Google’s spam policies cover this under “scaled content abuse”, defined as “creating large amounts of unoriginal content that provides little to no value to users, no matter how it’s created”. The last clause is doing the work: the policy is indifferent to whether a human or a machine produced the page. Being written by a person is not a defence, and being written by a model is not the offence.
There is also a decent chance the speed is partly an illusion. In a randomised controlled trial published by METR on 10 July 2025, sixteen experienced open-source developers worked through 246 real issues in their own repositories and took 19% longer when allowed to use AI tools. They had expected to be 24% faster going in, and afterwards, having actually been slower, they still believed the tools had made them about 20% faster. Sixteen developers is a small sample and it is a coding study, not a copywriting one, so I am not going to pretend it transfers cleanly. But the shape of it matches what I see on our own work: the draft arrives faster and the editing bill arrives later, and only one of those two is visible while you are working.
So the split I would give any writer on my team. Use the machine for volume, variants, structure and compression. Do the specifics yourself: the number, the name, the date, the loss, the opinion you would defend in a meeting. If you strip your copy of everything a model could have written, whatever survives is your actual positioning. If nothing survives, that is the finding.
[IMG: Give the machine volume, variants, structure and compression — the ugly first draft, headline variants, notes into structure. Keep the specifics yourself: the number, the name, the date, the loss, the opinion you would defend.]
The production half is commoditised and the judgement half is worth more than it was. Deciding what to say, what to leave out, and which claim you can actually stand behind was always the job; the typing was never the bottleneck, it just looked like it.
Long enough to answer the question the visitor arrived with, and no longer. Length is an output, not a target. A page padded to hit a word count reads exactly like a page padded to hit a word count, and both readers and ranking systems have got good at spotting it.
Google does not require it, and it does not affect how a page is treated. I would decide it on brand grounds rather than search grounds. If your positioning is craft and judgement, publishing unedited machine copy contradicts the pitch whether or not you disclose it.
Replace the first paragraph. Almost every business page opens by explaining what it is about instead of saying the thing. Delete that paragraph and start at the second one. This page is a worked example: the version before this one spent its opening forty words asking whether you had ever wondered about advertisements.
If your copy is the bottleneck rather than the site it sits on, that is a different problem, and the warning signs that a site needs rebuilding are worth ruling out first. If it is genuinely the words, our branding practice does this work with clients, and we will tell you when it is not the words.
Every figure in this article traces to one of the following. Publisher and publication date are recorded so a claim can be checked against its origin, and re-checked when the origin changes.