How we scout, how we rank

What we scout before we filter

We look at Marrakech first as an observation field before we sort it. The city gathers a density of offers that has nothing to do with its size on paper: over 400 riads in the medina, several thousand restaurants from the street-corner snack to the starred table, a day-trip industry that has exploded over the last ten years, a cultural calendar tight around a few big dates (marathon, ATP, film festival, religious holidays).

What we watch as a priority:

• The density and volatility of the addresses: where to concentrate our watch (medina, Gueliz, Hivernage, palm grove, rising outskirts like Sidi Ghanem), and where the churn of openings and closings forces a tight update rhythm.
• Weak signals before they surface in public reviews: a change of owner, a chef leaving, a new project in a quiet quarter. Our contacts on the ground send them back to us before they show up in aggregate scores.
• Real seasonality: the city is not lived the same way in February and in August. Some out-of-season excellent addresses become unliveable in July (heat, failing air conditioning, closed terraces). Others lose their point out of tourist season.
• Trends that change how we read a whole cluster: the rise of pool riads, the shift of big desert trips to shorter formats, the rise of "experience" wellness against the traditional neighbourhood hammam.

This scouting feeds the statistical selection that follows. Without it, we would filter addresses out of context: two hotels at the same threshold can have opposite trajectories, and only the ground tells you which.

How we filter before we rank

Every candidate address to appear in our hubs has to cross a hard statistical threshold first. This filter is the pivot of the method: it is what makes the rest possible and honest.

The primary sources change according to the type of address: Booking and Hotels.com for lodging, GetYourGuide, Viator and Tiqets for activities, day trips and ticketing, Google Reviews and TripAdvisor for restaurants. We systematically choose the platforms where the review volumes are high enough for the law of large numbers to play and individual bias to cancel out. We cross several sources to check that no anomaly skews the average (dubious reviews, coordinated campaign, exceptional period).

An address that slips under the threshold across the seasons drops out of the corpus at the next refresh of the site. We do not freeze fiches: the ranking follows the reviews, it does not wait for a manager to "defend" their address.

How we rank after filtering

Once the threshold is cleared, the desk does its work. The ranking mixes objective criteria and an owned share of editorial choices. The point: guarantee the reader that they will not waste their time on an address that disappoints half of travellers, and that they find a variety that covers the main stay profiles.

Lodging (hotels & riads)

Of the 400-plus medina riads that sell a room online, about 70 clear our threshold. We keep 30 to 40 of them in our hubs, spread by quarter, standing and price bracket, to cover every budget without piling up neighbouring doubles. Editorial ranking criteria:

Restaurants

Same method as for lodging, sources crossed between Google Reviews and TripAdvisor (the two platforms that concentrate the usable review volume for restaurants in Marrakech). The threshold stays the same: average ≥ 8/10 with more than half of the reviews at 9 or higher. We rule out restaurants rated below, even if they are popular on social media or recommended by the press: aggregate customer experience wins over the buzz of the moment. We also check the coherence between the menu shown and the menu served, the gap between listed prices and prices charged, and the presence of French or English service when it is advertised.

Activities, day trips & wellness

Sources GetYourGuide, Viator and Tiqets by segment. Same threshold (≥ 8/10, more than half at 9 or higher). We do not list a provider below this threshold, even if they show up high in a platform's search engine. We then check, by activity type: the official permit for guided day trips, animal treatment where there are any (calèche, camel, quad), the listed price held on arrival, and the alignment of spa or hammam practices with the fiche content. Watch out for misleading offers: 90 minutes advertised, but transfers included in the count.

Places to see (monuments, gardens, museums, souks, quarters)

A review threshold does not make sense here: a UNESCO-listed monument does not have to prove itself to travellers, and a quarter is not rated. The selection draws on official sources (institutions, UNESCO, reference scholarly editions), cross-checking with the specialist press, and seasonal update of the practical info (opening hours, prices, access). The desk adds the narrative and the editorial scouting (what time of day to go, what traps to avoid, how to string it together with the other visits).

Events

We cross the official sources of the organisers (festival site, federation, ministry) with the specialist national and international press, the ATP calendar and the international marathon calendar. Every fiche is updated at each edition (edition number, confirmed or announced dates, status). No event fiche is published without a verifiable official source.

What we rule out systematically

An address can tick every box and still be ruled out of the corpus. Here is what triggers an automatic rejection, regardless of the score:

• Score below the threshold: no exceptions, even for a "nice story".
• Local-only addresses: useful to neighbourhood residents, no point for a passing traveller (a local gym, a pharmacy).
• Tourist boards and ad networks: sources biased by construction, we do not cite them as references.
• Paid placement: a manager who offers to pay to appear or to rise in a ranking is ruled out of the corpus for good, score or no score.
• Refusal of price transparency: an address that lists a price online then charges another in practice (WhatsApp, unannounced "high season price") drops out of the corpus at the next flag.
• Non-attributable photos: we do not reuse an image whose author cannot be traced (MCG desk rule).

How we update

Aggregate scores are refreshed several times a year, on a cycle that follows the tourist seasons (end of winter, end of summer). An address that slips under the threshold drops out at the next automatic site refresh: no notice, no last chance. An address that comes back can be brought back in. Practical info (opening hours, prices, access) is checked at each update cycle.

You have spotted a stale piece of info, a manager who has changed, a price that no longer matches? Write to us, we look at it within the week.

How we are paid

Booking, Hotels.com/Expedia, GetYourGuide, Viator, Tiqets and Aviasales pay us a commission when a reader books via an MCG link. It is post-click: no one pays to appear on the site, no one pays to rise in a ranking, no one has a say over what we write. The hard threshold stays the hard threshold, even on a partner address.

The detail of the partnerships, the rates and what it actually funds: see the affiliate disclosure.