Writing product descriptions for an online store: do not convey the name and do not overstate the price of the supplier

You can, but then you should not expect that the card will win the search: the same text is available in twenty other stores, and the system will choose one of them. A compromise works on a large catalog: we take the technical specifications from the documentation as is, and we write the introductory block and "who suits" ourselves. It is cheaper than a full transfer and gives the card its own face. We separately check what came from the supplier: parameters from a neighboring modification regularly occur in the prices, and together with the text, you inherit someone else's error.

As much as it is necessary to remove the question before the purchase. This is usually 600-1500 characters plus a table of characteristics. Long texts on the card are rarely read, but the lack of an answer about compatibility, size or kit is immediately visible - from questions in the chat and from returns. And one detail about the form: we write the first paragraph so that it is read separately from the rest, because it is this piece that goes to the product feed and to the snippet, where no one will see the continuation.

From priority categories and from basic fields. First, the title, page title, text characteristics and markup — this is covered by the rules for the entire catalog in a few days. Then the descriptions for the items that are for sale. Writing a description for each card of the catalog with tens of thousands of items is a budget that almost never pays off, and we refuse to set the task in this way during the analysis.

Yes, and usually significantly. We measured the store for 48 products, where there was no block of structured data and no micro-markup: the price and availability in the issue were not announced at all. That's a few days of work versus a few weeks of writing. Fair margin: the markup does not increase the position or force the search engine to show the extended snippet - the decision is always up to you. It only gives such a chance and removes contradictions between what is written on the page and what the robot will read.

Your specialist, always. We write based on the documentation and data you have provided and do not make claims about properties that are not there. In goods where a mistake costs money - tools, medical products, auto parts - this is not a formality, but a condition of work. In practice, it looks like this: a lot of 20-30 cards goes to you for reconciliation, you mark the disputed, we rule, and only after that the lot goes to CMS.

They affect as part of the overall picture, but there is no separate guarantee here, and we do not give it. What can be seen more accurately and quickly: a card with answers to questions better retains a person, and this is visible in your analytics. Our advice is to measure the behavior on the cards with the new descriptions, rather than trying to assign a position to a specific paragraph. Another honest nuance: an indexed card does not mean a visitor. A product page without demand can be in the index and not get a single visit, and the description does not change that.

Yes, and it's often the cheapest option. We make a structure, a pilot batch and a list of rules, then your content manager works on them, and we selectively check. For directories that are constantly updated, this is the only scheme that does not turn into an infinite subsequence. We give the rules in text with examples: what is required on each card, in what order the blocks go, what statements cannot be written.

It often makes the most sense there. In the furniture store we measured, there are 76 products and exactly one category in the site map: it is useless to divide the selection into subcategories, so the buyer makes all the decisions on the product page. It is realistic to describe such a volume completely by hand, in two or three weeks, and after that the catalog is closed completely - without an endless tail that is never added to.