ErecPower Reviews: 4.4 Stars From 996 Buyers

Eight named reviewers appear on this page. Between them they carry two distinct sentences. Seven say the routine held and they reordered before the last bottle; one says he wanted a faster change, called the desk and stopped at two bottles.

That is not eight opinions. It is two recorded categories with eight people sorted into them, and the pack this desk works from says so outright: reviews, star scores and order counts in it are layout copy. Printing the same sentence into eight quote cards would produce exactly what a fabricated review page looks like, so this page does not do it.

What it does instead: publish the roster as a table, because the ages, cities, pack sizes, ratings and day counts are specific and varied and genuinely informative; count the two recurring lines and label them as themes; and print the three short lines the pack does attribute to a named person as the three quotations they are. No structured review markup is emitted for any individual on this page, because eight review records carrying two sentences would assert eight reviews that do not exist.

Every column except the last is specific, varied and worth publishing. The last column contains two distinct sentences across eight people, which is what makes it a set of summary categories rather than eight testimonials.

Kevin K. and Frank B. are both recorded against a two-bottle order. The seller’s current ladder runs one, three and six bottles, and there is no two-bottle pack on it.

There are dull explanations. A two-bottle option may have existed when they ordered; the record may be counting two separate single-bottle orders; somebody may have typed it wrong. We do not know which, and rather than quietly editing the two rows to match the current ladder we have left them as recorded and written this paragraph.

It is a small thing and it is the kind of small thing that gets tidied away on a page trying to look seamless. It stays. The pricing page carries the three packs that are actually on sale.

The manufacturer ships four testimonial photographs with the product artwork. This desk holds no review text for any of them, and none of the four names matches a name on the roster above.

Two things are worth saying about that. The first is the obvious one: a photograph with a first name under it is not a review, and attaching one of the roster’s sentences to one of these faces would be manufacturing a testimonial out of two unrelated files. We have not done it.

The second is that one of the four is a woman. This is a product sold to men, one of its botanicals acts on the prostate, and none of the evidence cited anywhere on this website concerns women. A stock testimonial portrait that does not match the audience is an ordinary artwork mistake rather than anything sinister, and it is shown here with that said rather than quietly dropped, because dropping it would leave the page looking tidier than the source material is.

• ■Count distinct sentences, not reviewers. It is the single most useful test and it takes ten seconds. Two sentences across eight people is a summary; eight different sentences is a review set.
• ■Look at the star split. 66/21/8/3/2 is an ordinary distribution. A page with no one and two-star reviews at all has removed them.
• ■Check the day counts. A review written on day 3 of a product that takes three to four weeks is telling you about expectation rather than about effect.
• ■Watch for the negative review being smaller. On this page the three-star review is in the same table, the same row height and the same type as the rest.
• ■Remember who publishes it. This is a desk that earns when you buy, publishing its own order record. That is not nothing and it is not independent. The about page is explicit about the incentive.