How the Recommended Sort Ranks Your Listings on HstockPlus

Most advice for sellers is unfalsifiable. Write better titles, price competitively, stay active. None of it is wrong and none of it tells you what the platform is actually doing with your listing.

Here the platform is doing something specific. Recommended is the sort a buyer gets by default on every category, subcategory and shop page, and it is a scoring formula held in configuration. This is what it reads.

The four weights

• Recent sales, thirty five per cent. Units sold in the last thirty days, scaled logarithmically against the best-selling listing in the same view. The log matters: the gap between zero and ten sales is worth far more than the gap between a hundred and a thousand.
• Quality, thirty five per cent. Half rating, half complaint rate. The two halves are not remotely equal in practice, and the next two sections are about why.
• Freshness, twenty per cent. An exponential decay with a thirty day half-life, measured from when the listing was created.
• Stock, ten per cent. Available units divided by one hundred, capped at one.

Those four are added into a base score, and the base score is then multiplied by a complaint penalty. Multiplied, not added, which is the part worth understanding properly.

Chasing reviews is chasing a constant

A five star review is created automatically for every confirmed order that the buyer did not review themselves. The result is that of more than eighty two thousand reviews on this site, all but a few dozen are exactly five stars, and only one hundred and sixty eight carry any written text at all.

The scoring accounts for this. Ratings are smoothed toward a prior of four stars, and the smoothing only starts to relax after ten reviews on a listing. Around eighty three per cent of active listings have no reviews at all and only about three per cent have ten or more, so for the overwhelming majority the rating term sits at the prior and does not move.

Work through what that leaves. A listing with no reviews scores four out of five. A listing with a thousand perfect reviews scores five out of five. The entire distance between the worst and best possible rating position is worth around three per cent of your base score. It is not nothing, and it is nowhere near where your attention should go.

The complaint multiplier, which is the only thing that can gut you

Your complaint rate is the higher of two ratios over the last ninety days: tickets divided by orders, and disputes divided by paid orders. It hits the score twice.

First inside quality, where the complaint half falls from full marks to zero across a complaint rate of nought to ten per cent. Then again as a multiplier on the whole score:

Complaint rate over ninety days	Everything you scored is multiplied by
up to 2%	1.0
2% to 5%	0.8
5% to 10%	0.5
above 10%	0.2

A listing that crosses ten per cent keeps a fifth of whatever it earned everywhere else. No amount of sales volume, freshness or stock recovers that, because the penalty is applied last and applies to all of it.

Compare the two halves of quality honestly. Ratings can move you by about three per cent. Complaints can move you by a factor of five. If you change one thing after reading this, make it the rate at which buyers have to contact you.

One protection is built in for small listings. Below five orders in ninety days the complaint rate is scaled down in proportion, so a single ticket on a listing with two sales does not read as a fifty per cent failure rate. It also means the penalty arrives properly only once a listing has real volume, which is exactly when it hurts.

Freshness decays faster than sellers expect

The half-life is thirty days. A listing one month old carries a bit over a third of its original freshness score, and at three months it is down to about five per cent. Since freshness is a fifth of the base score, an unchanged catalogue sinks steadily without anything going wrong with it.

Note carefully what resets it, which is nothing you can do by editing. The clock runs from when the listing was created. Editing the title does not restart it. The only way to hold freshness across a shelf is to keep publishing, which is a real cost and worth deciding on deliberately rather than by accident.

For scale: of roughly sixteen thousand seven hundred active listings on the site, around thirteen hundred are under a week old and around three thousand are under a month. Recent listings are a small enough slice that being in it is a genuine advantage.

Stock is the cheapest ten per cent available

Stock score is available units over one hundred, capped at one. A listing with a hundred or more units in stock earns the entire ten per cent. A listing with ten units earns one per cent of the total, and a listing at zero earns nothing and is also excluded from the new-listing slots described below.

There is no reward above a hundred, so uploading a thousand rows to one listing buys exactly the same ten per cent that a hundred buys. If you have depth, spreading it across listings scores better than concentrating it in one.

The page is interleaved, not a straight ranking

Scores decide the order within pools, but the page is built by cycling through three kinds of slot: seven from the highest scores, then two reserved for new listings, then one reserved for exploration.

• The new slots take listings under seven days old, with stock above zero, a complaint penalty of at least 0.8, and a score of at least half the median. This is a real opening: a fresh listing with clean history can sit high on page one without out-scoring anything.
• The explore slot takes listings sitting between roughly the thirty fifth and eighty fifth percentile by score, with some freshness left. It is how mid-table listings get impressions they did not earn on score alone.

Two consequences. Your listing appearing above another does not mean it outscored it. And a brand new listing that never sells will still lose its two-in-ten placement after seven days, so the window to convert it is short.

What the sort does not read

It does not read your seller level. It does not read how long you have been on the site, how many listings you hold, or your shop rating separately from your listings. And the Top Picks shelf is not this formula at all: that is an editorial flag, set by hand, and no amount of scoring puts you on it.

So the actionable list is genuinely four items long: sell recently, keep tickets rare, publish new listings, and keep a hundred units available. Everything else in the usual advice is either not measured or measured so weakly it will not show.