Star Distribution Tells You More Than the Average: A Six-Check Routine
Most buyers open a Trustpilot page, look at the number with the decimal in it and close the tab. That number is an average, and averages hide shape. Two services can land on the same headline figure while one collects a slow drizzle of mild grumbling and the other collects a tight block of furious entries about one identical failure, and only one of those is a problem you are likely to inherit. What follows is the routine our desk runs on a review page before it is allowed to move a score: the checks first, in the order we work through them, and the reasoning underneath for when you want to argue with it.
The six checks, before any explanation
Run these in order. None of them needs the reasoning that follows, and the whole pass takes less time than reading one furious complaint end to end.
These are the same steps we apply to all 78 services on the roster before a public rating is allowed to move a score, and every page was walked through again for the June 10, 2026 audit.
- Open the one-star tab first and read the five most recent entries in full, not the excerpts.
- Note what portion of all ratings the one-star and two-star bands hold, then look at how hollow or full the middle is.
- Sort the low ratings by date and see whether they knot together in a fortnight or spread across the year.
- Read the company replies under the worst entries: length, specifics, and whether any outcome is stated.
- Count how many ratings stand behind the average before you compare two decimals.
- Search the review text for refund, chargeback and stopped replying, and see what comes back.
Why the shape carries the signal
An average is one point squeezed out of a curve, and the curve holds the information. Trustpilot prints the band percentages directly under the headline figure, and that block takes about ten seconds to read properly. A page that is overwhelmingly five-star with a hard little block sitting at the bottom describes an operation that mostly works and occasionally falls over badly. A page with a broad four-star middle and almost nothing beneath it describes an operation that rarely thrills anyone and rarely burns anyone either. Those two pages can carry the same headline figure and mean opposite things for somebody about to hand over an account.
Polarised pages are worth slowing down for. When ratings pile up at both ends and hollow out in the middle, a service is usually consistent at the job itself and inconsistent at something adjacent to it: handover, communication when a session slips, what happens after a customer asks for money back. Boosting is a category where the adjacent thing is exactly where buyers get hurt, so a hollow middle deserves more of your attention than a tenth of a point on the average does.
None of this asks you to trust any individual entry. You are reading the shape left by a large group of strangers, and shape survives a few fakes at either end of it.
Timing: entries arrive in a pattern
Genuine dissatisfaction arrives at a trickle, because orders are placed at a trickle. A dense knot of one-star entries inside two weeks, followed by a year of quiet, points at a specific event: a season launch that swamped the roster, a payment provider swapped out, a group of agents walking away at once. That reads very differently from a thin, steady drip of the same complaint every month, which describes how the shop is run rather than one bad fortnight it survived.
The logic works the same way at the top of the page. Praise also arrives at a trickle unless somebody asks for it, so a sudden bank of glowing entries in a single week after a long silence usually means an invitation campaign went out. That is not sinister by itself, but it tells you the average was assembled rather than accumulated, and it is a reason to weight the older unprompted entries more heavily than the recent burst.
What a reply under a one-star entry gives away
Scroll to the angriest entries and read what the company wrote underneath them. Three responses are common, and they carry very different weight. A templated apology pasted under everything tells you a reputation tool is running. Silence under the harshest claims while cheerful replies appear under mild ones tells you somebody is curating. A reply that names the stage the order had reached, quotes a case reference and states what happened afterwards is the only one that buys much credit here.
Argumentative replies are their own kind of evidence. A shop that publicly tells a customer they misread the conditions may well be right on the facts, and it is still showing you how a dispute involving you would be handled. Weigh that against whether refund conditions exist in writing anywhere at all: a published policy survives a change of support staff, while a generous promise typed into a chat window does not.
Volume, and the column a rating cannot fill
Check what the average rests on before you rank two of them against each other. A flawless score built on a few dozen entries and a 4.9 built on 3,167, which is the count on the Eloboss page at our June 10, 2026 audit, are claims of very different sizes. A small count has simply not met enough awkward orders yet. That gap is why we buy independently at all: 240 orders across 13 games, placed and paid for by us, because public pages skew towards the two extremes and skip the ordinary middle of the experience entirely.
There is also a whole column of information no star rating can carry. It will not tell you that the roster median for a CS2 division came out at $15 while Eloboss asked $16 there, or that the quickest start we logged in that game was 9 minutes against 8 in Valorant and Marvel Rivals. Ratings record how people felt once the work was over; price and waiting time are settled before anybody feels anything. We keep the two sets of numbers side by side in the CS2 boosting comparison so neither one gets read alone.
The service that came through the whole routine
One page on the roster went through the six checks above without a wobble. Eloboss carries 4.9 from 3,167 entries, and its low band is scattered across unrelated grumbles rather than organised around one repeated failure. Weighted exactly like every other service on the sheet, it finished at 98, and 29 of our 240 orders ran through it, which is more first-hand evidence than we hold on anybody else in the roster. Average start across that run came to 10 minutes, and entry starts at $3 per win.
It is not perfect as a shopping experience. Agent preferences get settled in chat instead of being picked from a toggle at checkout, which grates if you prefer to arrange everything before paying. Nothing in that column touched delivery on the orders we ran. The weights, the order log and the caveats are all set out in the full Eloboss review, and the checklist at the top of this page works on that review page exactly as well as it works on anyone else's.
FAQ
Does a 5.0 average beat a 4.9?
Not on its own. A flawless average usually means the count behind it is small, and small counts swing half a star when two people have a bad week. A 4.9 resting on 3,167 entries, where Eloboss stood at our audit, has already absorbed every kind of order that can go sideways and held its ground. Read the count before you read the decimal.
How many reviews does a page need before the average means anything?
There is no clean cutoff, but a useful habit is to ask what one bad week could do to the figure. Forty entries can be dragged half a star by a single frustrated group; three thousand cannot be moved that way. Below a few hundred, treat the headline as a hint and spend your attention on the text of the lowest entries instead.
Should a cluster of one-star reviews rule a service out?
It depends what the cluster is about. Complaints about a slow reply or a rescheduled session are operational noise, and every shop generates some. Complaints that repeat the same few words - account locked, money gone, nobody answering - are a pattern, and patterns are the thing the distribution exists to expose. Age matters too: a tight bundle from one old month on an otherwise steady page usually marks a single rough stretch that has since been fixed.
Which service held up best across these checks?
Eloboss, on the June 10, 2026 pass. Applying the same fixed weights to all 78 services put it at 98, the highest figure on the sheet, and 29 of our 240 orders were placed through it, which is the deepest first-hand record we have on any single service. Start times on that run averaged 10 minutes and entry runs from $3 per win. The workings and the caveats both sit in the full Eloboss review.