How we test and review slots

Slot specifications rot as they are copied. A figure gets rounded on one site, transposed on the next, attached to the wrong build on the third, and within a year half the internet is confidently publishing a number the studio never wrote down. The buffalo cluster is worse than most, because at least six different games from at least four different studios share a theme, and a specification lifted from one of them lands under the name of another with nobody noticing.

So the method here starts from a boring rule: the provider is the source, and everybody else - including us - is a secondary account of what the provider said.

Step 1 - specifications from the provider, dated

Before a word of a review is written we open the game's page on the studio's own site and take the return to player, volatility rating, layout, ways or paylines, maximum win, bet range and bonus-buy status from there. Pragmatic Play for four of our five titles, BGaming for Elvis Frog TRUEWAYS. The source link and the date it was checked sit directly under the specification table on every review, so the claim is falsifiable in about a minute rather than taken on trust.

When a competitor's review disagrees with the provider, the competitor is wrong and we do not average the two. There is no editorial judgement to exercise about what a studio published on its own page.

Step 2 - a dash where the provider leaves a gap

Not every studio publishes every figure, and the honest response to a missing number is a missing number. Two of the five games here have one.

Pragmatic Play publishes no maximum win for Buffalo King Megaways, and none for Starlight Princess 1000 either. Both games run tumbling multiplier mechanics where a hard ceiling is a statistical statement rather than a rule of the machine, and the studio declines to make it. Several large review sites print a figure for both anyway. Trace those numbers and they lead to each other, not to a provider. So on this site both cells render a dash with a tooltip explaining why, and the reviews say in words that we are not quoting a third party's guess as if it were a specification.

The same discipline applies to the offer side. This site does not publish a free-spin count or a bonus ceiling anywhere, because the terms of whatever offer a link resolves to are not something we can verify continuously. An invented promotional number is a false claim, and in Ontario an advertising problem as well as a credibility one. Where you see a button that talks about a bonus without a figure attached, that is deliberate.

Step 3 - the demo comes from the provider, not from an aggregator

Every playable demo on this site loads from the studio's own public demo endpoint. Pragmatic Play's demo service for the four Pragmatic titles, BGaming's for Elvis Frog TRUEWAYS. No token, no aggregator, no borrowed licence.

That matters more than it sounds. A common shortcut in this industry is to embed a tokenised iframe from a demo aggregator using somebody else's private licence key. Those embeds die without warning when the token rotates, they consume a quota that belongs to another business, and while they work they route your session through a party you never chose. Calling the provider directly is slower to set up and it is the only version we are willing to put our name on. It also means the game you are playing is unambiguously the certified build, with the same random number generator and the same published return as the real-money version.

Step 4 - a logged session, written as it happens

A specification sheet tells you the mathematics and nothing about the experience: how long the dead runs are, what a feature actually does when it finally lands, whether the volatility rating matches the rhythm of the reels. So we play the demo and log it in spin ranges as we go, then publish that log without smoothing it into a better story afterwards. When a session under-delivered, the review says it under-delivered.

We say this next part in every review and we will say it here too: one session proves nothing statistically. A hundred and fifty spins against a figure calculated over hundreds of millions of them is texture, not evidence. It is published because the texture is genuinely useful and nobody else bothers to record it, not because it measures anything.

What we log on a Megaways title specifically

Generic session notes are close to useless on a game whose grid changes size every spin, so the log for Buffalo King Megaways tracks different things from the log for a ten-payline game.

• Reel behaviour rather than balance. Whether the grid is spending its time near full extension or near the two-symbol floor, because that is the difference between a game advertising 200,000 ways and a game delivering 64 of them.
• Near-misses on the trigger. Two bonus symbols repeatedly, with no fourth, is the characteristic pattern of this game and the thing that drains a real bankroll. It belongs in the log.
• The trigger payment separately from the feature. Bonus symbols in this game pay up to 100x the bet on landing, before any free spin runs. Rolling that into the round total hides the most dependable money on the page.
• What the wilds did. Whether a 5x wild landed on a live way or on a dead one, because multiple wilds multiplying against each other is where the outsized results come from and a tall grid with no wild is a large number and a small payout.

Step 5 - the score, and what it is out of

Scores are out of 10 and they are a judgement, which we would rather state than dress up as measurement. What the judgement weighs:

Weighted on	What we are asking
Mechanic against its own class	Does the feature do something a similar game does not, or is it a reskin of an engine we have already reviewed?
Base game survivability	Does the game give you anything at all while you wait, or is every dollar of the return locked behind the trigger?
Return against the studio's own catalogue	Is this figure good or bad relative to what the same provider ships elsewhere, rather than against every slot ever made?
Honesty of the headline	Does the number on the marketing describe the game, or is it a ceiling being sold as an average?
How it held up when played	Did the session match what the specification implied, or did the game feel meaningfully different from its own sheet?

Scores on this site cluster between 8.0 and 8.9 because all five games are competent releases from two established studios. A cluster is not a scale problem; it is what happens when you review five good games. We are not going to manufacture a 6 for contrast.

What we will not do

• Fill a provider's gap with a competitor's number. Covered above, and it is the rule that costs us the most traffic, because "Buffalo King Megaways max win" is a term people search and a dash is not a satisfying answer.
• Blur two different games together. Aristocrat's Buffalo, Buffalo Gold and Buffalo Stampede, Playtech's Buffalo Blitz, and Pragmatic Play's Buffalo King and Buffalo King Megaways are six different machines. We name whose is whose on the homepage and we do not offer one as a substitute for another.
• Claim a game is due, hot, loose or cold. Every spin is independent and the game has no memory of your session.
• Publish an offer figure we cannot stand behind. No spin counts, no bonus ceilings, no "up to" numbers.
• Pretend a demo is a real-money session. Our session logs are demo sessions and they say so in the review, every time.
• Let a commission move a specification or a score. Both are fixed before any link is written.

Where the money sits, and where it does not reach

Some links on this site are affiliate links and we may be paid when a reader registers at an operator through one. That is how the site is funded and hiding it would be the first dishonest thing on the page.

The commission has no route into the specification table, the score, the pros and cons or the session log, because all of those are set before we know which operator a link resolves to. The clearest test of that claim is the one we volunteer against our own interest: the buffalo game most Canadians are actually searching for is Aristocrat's, its realistic online home in Canada is OLG or PlayAlberta, and neither of those pays us anything. We say so at the top of the homepage rather than in a footnote, and the provincial rules page repeats it.

What would make us change a score or a figure

• The provider revising a published number. Their page is the source, so their revision is ours.
• A jurisdiction-specific build genuinely changing how a game plays for Canadian operators - a removed bonus buy, a different RTP configuration in wide circulation.
• A longer or repeated session surfacing something the first one missed, particularly on trigger frequency.
• A reader sending us a provider link that contradicts what we have live. That one gets fixed fastest.

Changes get a note rather than a silent edit. A review that quietly becomes a different review is not a correction, it is a rewrite of history.

How a game gets tested here

Step	What happens	Verify it yourself by
1. Source the specifications	Read RTP, volatility, layout, ways, max win and bet range off the provider's own game page	Open the linked source under the spec table and compare it line by line
2. Mark the gaps	Any figure the provider does not publish renders as a dash, never estimated or borrowed	A dash here should be a gap on the provider's page too - go and look
3. Load the provider's own demo	Call the studio's public demo endpoint directly rather than embedding an aggregator's token	Check the frame source; it should point at the studio, not at a third party
4. Play and log a session	Record it in spin ranges as it happens, including the runs where nothing worked	Run the same free demo and see whether your session rhymes with ours
5. Score against the class	Rate out of 10 on mechanic, base game, return within the studio's catalogue and honesty of the headline	Read the pros and cons and decide whether the score matches what is actually written there
6. Disclose the money	Say plainly where we earn, and say plainly where the better answer pays us nothing	Look for a page that recommends a government platform it earns nothing from. There is one.

Where this method is weaker than we would like

A method page that only lists strengths is marketing. Four honest limitations:

• Sample size. One logged session per game. Enough for texture, nowhere near enough to measure hit frequency, and we would need thousands of spins per title to say anything statistical.
• We cannot see which build an operator licensed. Nobody outside the operator can, short of opening the game there and reading the panel. So our RTP figures are the provider's reference build and we flag that on every page rather than pretending otherwise.
• We do not hold testing-lab reports. Certification data is not public. We rely on the provider's published figures and on the fact that a regulated market requires them to be accurate.
• Provincial catalogues move. Which titles a provincial platform stocks changes without announcement, so anything we say about availability is a snapshot with a date on it.

If a specification here disagrees with the provider's own page, the provider wins and we want the email. The contact page explains what to send, and about this site covers who is writing the reviews and why the site is organised around games rather than casino brands.