How to write a roundup that does not read like every other roundup.
“Best X” posts all sound identical because they are all written from each other. Six things break the pattern — and five of them are about what you leave out, not what you add.
The short version
- Segment by use case, not by rank. “Best for small kitchens” beats “#3” because nobody identifies with a number.
- Name what you rejected and why. The products you left out say more about your judgement than the ones you included.
- State your selection method before the list. It converts a ranking into a recommendation.
- Put the answer in the first hundred words. Nobody scrolls 800 words of preamble, and Google stopped rewarding it.
- Cut the buying guide. The generic “what to look for” section exists to pad word count and everybody skips it.
Read five “best cordless drill” articles and they are the same article. Same nine products in roughly the same order, same section headings, same eight hundred words of introduction, same conclusion that recommends the most expensive option.
That is not laziness so much as convergence. Everyone researches the same way — by reading the pages already ranking — so everyone arrives at the same list and the same shape. The result is a format where being different is worth more than being thorough, because thorough is what everyone already is.
Why they all converge
Three mechanisms, and it is worth understanding them because each one suggests its own escape.
Everyone reads the same sources
You research by reading page one. Page one becomes your outline. Your article becomes page one for the next person. The format inbreeds.
The bestseller list picks the products
If you choose from what Amazon promotes, so did everyone else. Identical inputs, identical lists.
Ranking is a weak organising idea
A numbered list implies one right answer for everybody, which is never true and forces every article into the same shape.
And now, generated sameness
Content written from the same summarised sources converges harder still. The pages that lost most traffic in 2024–2026 were the ones a summary box could replace.
The six things that break the pattern
Organise by use case, not by rank
The single biggest changeReplace “#1, #2, #3” with “Best for most people / Best if you are on a budget / Best if you will outgrow the cheap one”.
Ranking asks the reader to accept your ordering. Segmenting lets them find themselves, which is what they came to do. It also removes the dishonest part of most roundups — the pretence that one drill is objectively third-best for everyone.
The practical test: a reader should be able to skip straight to their own section and stop reading. If your structure requires them to read all nine entries to work out which applies, it is organised for you rather than for them.
Say what you rejected, and why
Almost nobody does thisEvery roundup is a list of winners. A short section naming three products you considered and excluded — with the reason — does more for credibility than any amount of praise.
“The Ryobi is the obvious budget pick and we left it out: the battery platform is fine but the chuck slips under load, and it comes up repeatedly in owner reviews.”
Two things happen. The reader learns you actually evaluated rather than assembled, and they get information they cannot find on the pages that only list winners. This section is also, reliably, the part people quote when they link to you.
State your method before the list
One paragraphHow did these nine become these five? Say so plainly: what you tested, what you read, how many owner reviews, what disqualified a product.
This is the same “how we chose” paragraph that makes a research-based review defensible — see reviewing a product you do not own. On a roundup it does extra work, because it is the difference between a curated recommendation and a list somebody typed.
Include something only you could know
The durable advantageOne photograph of a product on your own desk. One sentence about the thing that annoyed you in week three. One measurement you took. One detail no spec sheet contains.
This is the part that cannot be researched, rewritten or generated, and since the 2024–2026 updates it is the clearest line between content that holds its rankings and content that drifts. → how to photograph products · what experience looks like on a page
Answer in the first hundred words
Costs nothing, changes everythingThe convention of eight hundred words before the first product is an artefact of an era when length was a ranking signal. It no longer is, and readers never liked it.
Open with the answer: “For most people it is the Gaggia Classic. If your counter is small, the Bambino. If you know you will upgrade, buy the Silvia now and skip a step.” Then spend the article justifying it. Readers who trust you leave satisfied; readers who want the reasoning keep reading. Both outcomes beat a bounce at paragraph four.
Recommend against something
The credibility shortcutTell a group of readers not to buy at all. “If you drill four holes a year, none of these are for you — buy the cheapest corded drill and spend the difference on bits.”
It costs a small number of conversions and buys the trust of everyone else, because it is visibly against your own commercial interest. It is also usually true, which is the main argument for saying it.
A structure that follows from all six
What to cut
Roundups get long by accretion. These four sections are almost always dead weight.
| Cut this | Why |
|---|---|
| The generic buying guide | “Consider your budget and your needs.” It is on every competing page, it is skipped, and it exists to hit a word count. If a specification genuinely decides the purchase, explain it where the products are. |
| The history of the category | Nobody choosing a drill wants a paragraph on the invention of the drill. |
| Products six through nine | Past five, entries are there for length. Cutting four weak products makes the remaining five more persuasive and the page easier to maintain. |
| Restated specifications | Copying the listing adds nothing the reader cannot see. Interpret instead: which number matters, and why. |
On length. There is no word count that ranks. A 1,200-word roundup that settles the decision beats a 3,500-word one padded to look authoritative, and padded articles read exactly like what they are. Write until the question is answered, then stop — and if you cut four products, cut them properly rather than demoting them to a list at the bottom.
Roundups decay faster than anything else
They name specific products, so every discontinuation ages the page. A roundup where two of five picks cannot be bought is worse than useless — it actively misleads, and the reader who discovers it does not come back.
- Check the products quarterly. Discontinued and out-of-stock items return a healthy 200, so nothing flags them. → how to find them
- Replace, do not rewrite. Swap the dead product for its successor and update that section. Keep the URL and what the page is about. → refreshing without losing rankings
- Fewer products means less maintenance. Five entries is half the decay surface of ten, which is another argument for the shorter version.
- Show when you last checked. “Products verified August 2026” is a genuine trust signal on a page whose whole value is currency.
The test before publishing
One question, and it is unforgiving: if a reader saw your article and the three ranking above it, would they be able to tell them apart?
If the honest answer is no, the article is not finished — and no amount of additional length will fix it, because length is the thing everyone else already has. What separates it is a photograph, a rejection, a stated method, or a group of readers you told not to buy.
Part of The Amazon Affiliate Site Playbook — the whole job in six stages, from choosing a niche to keeping a mature site earning.
Draft the structure, keep the judgement
Ama Affiliate Pro builds the roundup scaffolding, the comparison table and the product boxes in minutes — so your hours go into the rejections, the photographs and the verdict, which are the only parts a competitor cannot reproduce.
Start a 14-day trialCommon questions
How many products should a roundup include?
Three to five. Past five, entries exist for length rather than for the reader, and every extra product is another thing to maintain when it gets discontinued. Cutting four weak options makes the remaining ones more persuasive.
Should I rank products or group them by use case?
Group them. “Best for small kitchens” lets a reader find themselves; “#3” asks them to accept your ordering. Segmenting also removes the dishonest implication that one product is objectively third-best for everybody.
How long should a roundup post be?
As long as it takes to answer the question — there is no word count that ranks. A 1,200-word roundup that settles the decision beats a padded 3,500-word one, and padding is obvious to readers. Cut the generic buying guide and the category history first.
Where should the verdict go?
In the first hundred words. The convention of a long introduction before the first product is a leftover from when length was a ranking signal. Give the answer immediately, then justify it — readers who trust you leave satisfied and readers who want reasoning keep going.
Should I include products I do not recommend?
Yes, in a short “what we rejected” section with the reason for each. It demonstrates that you evaluated rather than assembled, gives readers information the winners-only pages lack, and is reliably the section people quote when they link to you.
Do I need a “what to look for” buying guide section?
Almost never. It is on every competing page, readers skip it, and it usually exists to hit a word count. If a specification genuinely decides the purchase, explain it where the products are rather than in an abstract section beforehand.
How often should I update a roundup?
Check the products quarterly. Roundups decay faster than any other format because they name specific items, and discontinued products return a healthy 200 so nothing flags them. Replace dead picks rather than rewriting the page, and keep the URL.
My roundup is not ranking. What is wrong?
Usually one of two things. Either the keyword was never winnable for your site, which is a targeting problem rather than a writing one. Or the article is indistinguishable from the three above it — in which case ask whether a reader could tell them apart, and fix that rather than adding words.