Writing From the Customer's Words: Voice-of-Customer Copy
The most persuasive copy in a small business is usually a cleaned-up version of what customers already say. Collect their words from reviews, searches and support, mine them for problems and outcomes, and write without inventing a voice.
The short answer
Voice-of-customer copy starts from material customers gave you: the words in reviews, the phrases in site search, the questions in support messages, the reasons in returns, and the language people use when they recommend you to someone else. Mine that material for three things: the problem as they name it, the outcome as they value it, and the objection as they phrase it. Then write the page in that vocabulary, in your own clear sentences. Do not copy a customer words into a claim they did not make, do not invent typical customers, and do not let the exercise become mimicry. The aim is recognition: a reader who feels the page was written about their exact situation.
Where the words live
Reviews are the richest source because they contain situation, expectation and verdict together. Search logs show the need in its shortest form, typed under pressure. Support messages show the points where the current site failed to explain, which are often the exact sentences the new copy needs. Returns reasons show the gap between what was promised and what arrived. Sales calls and counter conversations, noted honestly at the time, show objections in spoken form. Gather from at least two sources before trusting a phrase, because each source over-represents a different kind of customer: reviewers are the motivated, searchers are the urgent, support writers are the stuck.
Mine for problems, outcomes and objections
Read the material with three highlighters, in effect if not in fact. Mark problem language: what was not working, what they needed to do, what they feared. Mark outcome language: what changed, what they could do afterwards, what they noticed first. Mark objection language: what nearly stopped them, what they checked, what they asked. Keep exact phrases where they are vivid, and note frequency where it matters: a phrase used by many customers is a headline candidate; a phrase used once may be a lovely story and a poor general claim. This sorting is the research. The writing afterwards is comparatively straightforward.
Write in the words, not in quotation marks
Voice-of-customer copy is not a collage of quotations. Use the vocabulary and the framing inside your own sentences, which you are responsible for and which must be accurate for the product as sold. If customers call the outcome sleeping through the night, the headline can speak of sleep, provided the product can honestly be described that way for the customers it suits. If customers use a product for a purpose you had not centred, consider whether that use belongs in the copy, the category structure and the product data, following the same evidence logic as search log reading in the navigation cluster.
Limits: whose voice, and what it cannot prove
Your sources are your customers, who are not the whole market. Their words describe people who already found you, understood enough to buy, and stayed long enough to comment. Visitors who left confused are under-represented, which is why stranger testing still matters. Customer words also cannot prove performance: a phrase in copy must still describe something true, and any number a customer mentions becomes your claim only if you can support it as set out in numbers in copy. Keep the line clear between vocabulary, which you borrow freely, and evidence, which you must own.
Refresh the material as the business changes
New products, new audiences and new uses all change the vocabulary. Build a light routine: each quarter, pull a fresh sample of reviews, searches and support phrases for the pages that matter most, and compare them with the copy now live. Retire phrases for discontinued uses, add the language of the work you now sell most, and note where customer words have exposed a page that still speaks in the old internal terms. Keep the raw samples with your copy log, dated, so a future editor can see why a headline says what it says instead of replacing evidence with taste.
Test recognition, not cleverness
The test for this copy is recognition. Show the draft to a customer or a stranger in the target situation and ask what product this is, who it is for and what problem it addresses. If they answer in the words you hoped, the sourcing worked. If they paraphrase into completely different language, your material missed how the need is actually described. In live use, watch the same signals as other copy changes: movement onward from the page, support questions shifting from basic confusion to specific fit, and search phrases on the site beginning to match the words you now use. Recognition is measurable in behaviour, given time and a fair baseline.
Sources and evidence on the record
- CXL, CXL blog - practitioner writing on testing method, research process and optimisation programme discipline.
- Nielsen Norman Group, articles - usability research on reading behaviour, forms, response times and navigation.
- Microsoft, Microsoft Clarity documentation - heatmaps, session recordings and frustration signals such as rage clicks.
Evidence basis: Method guide. Any figure quoted is attributed to a named published source listed below; worked examples are labelled illustrations with round numbers, not measurements. Applies when: You are writing or rewriting any persuasive page and have real customer material to draw on