Reviews and Comparisons That Actually Sell

Readership and revenue are two different things

The first time I put traffic statistics next to commission payouts on one of my sites, the result surprised me. The most-read article of the year, a long guide on how to choose equipment as a beginner, earned less over twelve months than a short review of one specific model that a fraction of those people read. The guide was read by people who were still looking around. The review was found by someone who had already picked the product and needed one last confirmation that they were not making a mistake.

The difference lies in the intent people bring with them from the search engine. Queries like "what is affiliate marketing" or "how does a robot vacuum work" bring curious visitors who enjoy a good read and then leave. Queries like "review", "experience", "comparison", "best robot vacuum under 5,000 CZK" or "where to buy" are typed by someone with their wallet open. These are the queries worth writing affiliate content for, because here the link is not an obstacle but a help.

That does not make informational content useless. It builds trust, attracts backlinks and gives the site the topical breadth that search engines reward. But if your time is limited, and every publisher's time is, it pays to consciously distinguish between the article you write for readership and the one you write for revenue. The second kind follows different rules, and those rules are what this text is about.

Buying intent in search

The simplest test of buying intent is to imagine what the person would do after reading the perfect answer. If they would close the browser and go make dinner, it is an informational query. If they would open an online shop, it is a buying query. There are plenty of shades between the two poles, but a few types keep coming back:

• Product name plus "review" or "experience": the person has chosen and is looking for confirmation or a warning. Conversion tends to be the highest, and so does competition.
• "Comparison" and "X vs Y": they are deciding between two or three candidates. They need a clear answer about which one suits whom.
• "Best ... under 5,000 CZK" or "for beginners": they have a budget and a need, but no specific model. Here a reasoned selection sells.
• "Where to buy", "discount", "deal": the product is chosen, only price and availability remain. A short article with current prices and codes is enough.
• "Is it worth it": they are hesitating about whether the purchase makes sense. An honest answer, including when it is not worth it, converts better than enthusiasm.

Notice that with all of these queries the reader expects an opinion. A neutral list of specifications copied from the shop will not help, because they have already seen that. They came for an experience they do not have themselves, and for a decision they do not dare to make alone.

Structure of a review that converts

Reviews that make money have a surprisingly similar skeleton. It is not a template to be followed to the letter, but the order of information matches the way people make decisions.

Who the product is for and who it is not for

The first paragraphs should tell the reader whether they are reading the right article. "This vacuum suits a flat of up to 70 square metres with smooth floors and no deep-pile carpets. If you have a two-storey house and two dogs, skip to the end, where I recommend a different model." An opening like this turns away part of the visitors, but they would not have bought anyway. The rest get the feeling they are reading someone who understands their situation. A verdict at the start is not a spoiler, it is a service to the reader who has no time.

Experience from actual use

This is where a review parts ways with a rewritten brochure. Concrete situations you lived through with the product: what happened after three weeks, what surprised you, what you had to look up in the manual, how a warranty claim went. Your own photos, not the manufacturer's press kit. If you do not own the product, admit it and write a comparison based on specifications and other people's experience instead of pretending you tested it. Readers notice faster than you think, and trust once lost on a website comes back very slowly.

Specific parameters

After the experience come the numbers: dimensions, battery life, noise, consumption, compatibility. Ideally in a short table that can be skimmed. Choose the parameters that influence the decision, not every one the manufacturer lists. With headphones, people care about battery life and weight, not the Bluetooth protocol version to the third decimal place.

Price and where to buy

Give an approximate price with the date it applied and a link to the shop where it can be verified. The phrase "roughly 4,500 CZK (October 2026)" ages gracefully, whereas "only 4,499 CZK" with no date is untrue within six months. If the shop has a discount code or bonus for your readers, put it here, not hidden in the conclusion.

A link early in the text and again at the end

Some readers decide within the first two paragraphs and look for somewhere to click. Give them the first link right at the verdict in the introduction. The second link belongs with the final summary, where the reader who wanted to read everything will arrive. In between, link at specific claims where it makes sense, but do not push: three to five links in a 1,500-word article is usually enough.

Comparison tables without cheating

A comparison of two to five products is the most valuable format of affiliate content, because one article covers more queries and more programs. It is also the format that is easiest to spoil. A table in which the product with the highest commission always wins is seen through quickly, and the reader remembers it.

A few principles that work for me:

• Set the criteria before you look at the commissions. Write down what the product should do and at what price, and only then look for candidates.
• Limit the table to five to seven rows with the parameters that decide. A twenty-row table looks thorough, but nobody reads it.
• Give each product a role. The reader finds themselves in it and clicks on "their" row.
• If a product has no affiliate program, include it anyway when it belongs in the selection. The reader will buy elsewhere this time, but will come back next time because you did not advise them according to commission.
• With the verdict, write why this one, and for whom a different one would win.

Below the table there should be a short paragraph on each product with its own link. The table is for orientation, the paragraphs are for the decision.

Link placement and weak banners

A banner in the sidebar is the most convenient way to deploy an affiliate link, and at the same time the least effective. People learned to ignore banners long before ad blockers arrived; the eye skips over them like noise. Banner click-through rates tend to sit well below one percent, whereas a text link inside a paragraph that responds to a specific claim can be clicked by anywhere from a few percent to tens of percent of the readers who get that far.

A link works when it stands exactly where the reader thinks "and where do I get this?". You write that the battery lasted ten hours of use, and right after it there is a link carrying the product name. Or you sum up the verdict and below it there is a button reading "Current price at the retailer". A button works better than plain text because it visually separates action from reading, but there should not be more than two or three on a page, otherwise they turn into another banner.

The link text should say what happens after the click. "Here" or "click" promises nothing. "See the price at the shop" or the model name itself tells the reader where they are going and helps search engines understand the context too.

Updating old articles

The cheapest revenue growth I know is dusting off articles that already rank in search. An article that has held a first-page position for three years brings visitors for free. But if it quotes a 2023 price, two of the models are discontinued and one link leads to a 404 page, it converts a fraction of what it could.

What to check during an update:

1. Availability of every linked product. Replace sold-out or withdrawn models with their successors.
2. Prices and the date next to them. Rewriting the number and the month is enough.
3. Whether the affiliate links still work. The program may have ended, the shop may have moved to another network or changed its URL structure.
4. New models released in the meantime that the reader will know about. If you do not mention them, the article feels dated even when it is otherwise correct.
5. The introduction and the date of the last update, so it is visible that someone is looking after the text.

A sensible rhythm is to go through the ten best-earning articles once a quarter and the rest once a year. One afternoon of work on an article that already has traffic usually earns more than a week of writing a new text that will spend a year climbing to its positions.

Labelling affiliate links

The reader has a right to know that you earn a commission from their purchase. In most European countries it is also an obligation: an unlabelled affiliate link is hidden advertising. A short sentence above the article or below the introduction is enough: "This article contains affiliate links. If you buy through them, we receive a commission from the shop; you pay the same price." In my experience such a notice does not reduce conversion. People who read you regularly click on purpose, because they want to support you.

The technical side matters just as much. Affiliate links should carry the attribute rel="sponsored", or the combination rel="sponsored nofollow". This tells search engines that it is a paid relationship and that the link should not pass authority. A site with hundreds of unmarked affiliate links risks being treated by a search engine as a sales catalogue without content of its own. Most content management systems let you add the attribute globally for a given domain, or through a plugin that manages links centrally, which is also useful when swapping links, a topic we will come to below.

Keywords without over-optimisation

You need to know the keyword in order to hit the reader's intent, not in order to repeat it in every paragraph. An article for the query "XY robot vacuum review" should have that phrase in the title, in one heading and naturally in the introduction. The rest of the text you write for the person who is already on the page. Search engines today understand synonyms, inflection and context well enough that unnatural repetition does more harm than good.

More useful than keyword density is covering the sub-questions a reader with buying intent is dealing with: battery life, noise, service, comparison with the predecessor, whether the more expensive variant is worth it. When you answer them in subheadings, the article ranks for dozens of related queries you would never have guessed. Look at the autocomplete suggestions and the related questions block in the search engine; it is free and it is enough.

One review per product. If you also want to cover a comparison with a competitor, write a separate article and link the two together. Two articles for two intents rank better than one that tries to serve everybody.

Measuring: which articles earn

Without measurement you write blind. The minimum you need is to know how much each article earned, how many people clicked from it and how many of those clicks ended in an order.

EPC per article

Most networks let you add your own identifier to a link (called SubID, channel or something similar). If you put the article slug in it, the network statistics will show commissions broken down by article. From that you calculate EPC, the earnings per click, and that is the number decisions are made on. An article with an EPC of 20 CZK and a hundred clicks a month deserves to be expanded and given a better spot in the menu. An article with an EPC of 1 CZK and a thousand clicks is either badly targeted or links to a program that does not convert.

Clicks without orders

The most instructive articles are those where people click but do not order. There are usually several causes, and each has a different fix:

• The product is sold out at the retailer or more expensive than you state. The fix is an update.
• The reader came with informational intent and clicked out of curiosity. The fix is to rewrite the introduction so that only those who want to buy get the link, or not to monetise the article at all.
• The program has a short cookie or the shop works poorly on mobile. The fix is to try another retailer selling the same product.
• The link leads to the homepage instead of the product. A deep link straight to the product page lifts conversion considerably.

Keep a record of revenue by program and month in one place, even if you promote programs from several networks. Only when you look across networks do you see that one program carries half your income and that you are dangerously dependent on it. If you record your monthly revenue and commissions in AffilPlanet, you see them in a chart across programs and do not have to piece them together from five different exports.

When a program changes its terms

It happens to everyone: the shop cuts the commission from 8 to 3 percent, shortens the cookie from 30 to 7 days, or ends the program altogether. If you have twenty articles on that program, you usually find out from a shrinking payout two months later. There are two ways to prevent that.

The first is watching programs. In the catalogue of affiliate programs on AffilPlanet you can mark the programs you promote as watched, and when the commission, terms or codes change you receive a notification by e-mail and in your account. You learn about it the moment the shop records the change, not from the payout.

The second is central link management. If you hard-code every affiliate link into the article, swapping a program means opening twenty articles and rewriting by hand. If you manage links through redirects (for example /go/product-name) or a plugin, you change the destination in one place. When a program ends, you redirect the links to another retailer selling the same product and the articles keep earning without any rewriting.

When a program changes its terms for the worse, do not panic. Check in the catalogue who else sells the same product and on what terms. You will often find that a competing shop offers a higher commission or a longer cookie, and the change ends up meaning more income for you.

Choosing programs for your content

A good review falls flat if it links to a program that does not convert. When choosing a program for a specific article I look at several things: whether the shop sells exactly that product and has it in stock, how long the cookie is (for pricier products where people hesitate for weeks, 30 days is the minimum), what the commission is relative to the product price, how long approval takes and what customer ratings the shop has. A high commission at a shop people do not trust earns less than a lower commission at a retailer where they complete the order.

In the AffilPlanet catalogue you can filter by category, network, commission level and cookie length, and for every program you see the terms, bonuses for publishers and current codes. If the shop manages its own program, you can message it directly and arrange, say, a higher commission for a series of reviews or a product for testing. For a publisher who writes reviews, a product loan is often worth more than a couple of extra percent.

The opposite direction works too. When you list your website on AffilPlanet with its content type and topics, it becomes visible to the affiliate managers of shops looking for reviewers in their field. From my own experience as a shop owner I know that partnerships where the manager approaches the publisher first tend to be the most stable, because both sides know from the start what they want from each other.

A procedure for your next article

If you take one procedure away from this text, let it be this: pick a product you have at home and genuinely use, check in the search engine that people search for its name together with "review" or "experience", find a program in the catalogue with a reasonable cookie and stock availability, and write a text whose introduction says who the product is for and who it is not for. Put the link at the verdict and again at the end, mark it as sponsored, and put the article slug in the SubID. After three months look at the EPC and decide accordingly whether to expand the article, update it or leave it alone.

It is not a quick path to income. A review takes months to reach its positions and the first commission arrives only after the shop approves the orders. But an article written this way earns for years, and more with every update, while an article written for readership mostly earns attention.