The same four inputs, weighted differently
Every directory published under our directories runs on the same scoring engine, and that engine reads the same four things: review volume, review recency, the shape of the rating distribution, and what reviewers actually say.
What changes from one directory to the next is how heavily each of those counts, and particularly what the fourth one is read for. That is deliberate across all of our independent Malaysian service directories, and this page explains why.
A worked contrast: the locksmith and the homestay
Take two of our directories and set them side by side.
A locksmith is scored heavily on response. Someone locked out of a flat at midnight has one requirement that outranks every other: somebody answers, and somebody comes. A locksmith with beautiful workmanship who does not pick up after 6pm has failed the job as it is actually experienced. So the method reads reviews for whether out-of-hours calls were answered and whether the van arrived when it was said it would.
A homestay is scored on none of that. Nobody has ever booked a homestay urgently. It is chosen weeks ahead, for a specific occasion, by a group with a specific shape. So the weighting goes to suitability: did the pool exist and was it usable, did the property actually accommodate the number of people the listing claimed, did it work for the family gathering or the event it was booked for.
Running the locksmith criteria over homestays would produce a ranking that rewarded fast replies to booking enquiries, which tells a family almost nothing about whether they will have a good weekend.
| Directory | Weighted heavily | Not scored |
|---|---|---|
| Klang Valley Locksmith Directory | Out-of-hours availability, arrival time, quote accuracy | Suitability for an occasion |
| Melaka Homestay Guide | Fit for the stated purpose, accuracy of the listing | Response time |
| Klang Valley Cleaning Services Directory | Consistency across repeat visits, reliable attendance | Speed of arrival |
| Malaysia Confinement Centre Guide | Quality of care, staffing continuity | Package price |
Why one universal score would be wrong
The alternative is a single model applied to everything, and it is worth spelling out why that fails rather than simply asserting that it does.
A universal model has to pick criteria that apply everywhere, which in practice means star average and review count and nothing else. That model cannot distinguish between a plumber whose reviews say “came within the hour” and one whose reviews say “took three days but did tidy work”, because both were given five stars. The information that would actually help you is in the text, and a universal model has to discard it to stay universal.
Worse, a universal model quietly imports the assumptions of whichever category it was designed around. Most were designed around emergency trades, because that is where the lead-generation money is. Apply that to restaurants and you end up rewarding fast replies to reservation messages.
The criteria follow the trade, not the other way round
If a category’s criteria cannot be stated plainly for that category, the ranking is measuring the wrong thing.
How the criteria for a category get set
The per-category weighting is not arrived at by intuition. It comes from reading what people actually write about that trade.
Reviews in any category cluster around a small number of recurring themes, and those themes differ sharply between trades. Plumbing reviews return again and again to arrival time and whether the repair held. Homestay reviews return to whether the property matched its listing and whether it suited the group. Confinement centre reviews return to attentiveness at night and staff continuity.
Those recurring themes are what a category is judged on, because they are what the people who used the service considered worth mentioning. It is a deliberately conservative approach: the criteria are drawn from the evidence rather than imposed on it.
Where a category’s themes change over time, the criteria can change with them, and any change is written into the published method rather than applied quietly.
What stays constant across every directory
Three things never vary, whatever the category.
Scores are built from public customer reviews. Rank cannot be bought, and the ranking engine cannot see whether a business has paid for anything. Every listing is re-checked monthly, and closed businesses come off automatically at that re-check, a cycle described in full in how often listings and scores are updated.
So what shifts between directories is the emphasis, not the integrity. A locksmith and a homestay are judged on different things because they are different things, but neither can pay its way up.
Where the per-category criteria are written down
Each directory page on this site states its own weighting before it links out to the directory itself. If you want the whole scheme in one place, including the inputs and the explicit exclusions, read the scoring method.
And if you want to see the difference in practice rather than in principle, open two directory pages from each directory and compare what each one says it weights. They are not the same, and they are not meant to be.