What is kept off this site, and why

A comparison is only worth reading if the two sides are the same thing. Most of what this project collects never reaches these pages. This is the whole list of reasons, so that an absence can be read as a decision rather than a gap.

We track 89 products across four shops: 18 in Refrigeration, 17 in Washing machines, 12 in Dishwashers, 14 in Televisions, 9 in Vacuum cleaners and 19 in Cooking appliances.

1. Only products we can prove are the same product

Two shops describe the same washing machine in two different ways. We match on the manufacturer’s model code and keep only matches graded exact. A probable match is not shown, and not shown with a badge either: a badge would ask the reader to price our uncertainty, which they cannot do.

The rule applies twice. The product must be an exact match, and each individual shop listing attached to it must be an exact match too. The second half of that rule removes more products than anything else this project does.

2. Prices that disagree too much to be the same item

Where the same model code carries prices across shops that differ by more than 40%, or where one shop lists two prices for it more than 15% apart, the product is removed rather than shown. A wide gap is usually a sign that the code has been attached to something that is not the plain product.

Two real examples from what this screen removed. In one, a shop had listed a small-appliance bundle under the model code of the item given away with it, so the code identified the free gift rather than the thing being sold. In another, a shop sold a set of three of an item under the single item’s code. Both would have produced a row telling a reader that one shop charges several times what another does for the identical product. Neither figure appears anywhere on this site, including in this explanation.

3. Sources that contradict themselves

Where a shop’s own stated rate and its own stated monthly payment do not agree, every plan on this site generated from that rate card at that length is removed until it can be re-checked — the defect belongs to the card, not to one product. Where a shop states two different things about the same listing, the listing goes. We record the contradiction internally and correct nothing silently.

4. Anything we cannot show is current

Each plan is checked against the shop’s own page as we recorded it. If the card it came from cannot be shown to be the one the shop is currently publishing, the plan is held back rather than shown with a caveat.

5. Products too few shops price

A comparison needs two sides, and two is the rule: a product is shown only where at least two shops publish an installment plan for it. Where a product is compared at exactly two shops, its page says so above the table, its entry on the index says so, and the category heading says how many of that category’s products are in that position. 64 of the 89 products on this site are compared at two shops; the rest are compared at three or more.

A shop that has a price for a product and publishes no installment plan we can print for it is not one of the two sides. Where that happens the shop is named on the product’s own page, and no price of its is shown there.

This rule governs the cross-shop comparison. Phones are not compared across shops and are screened differently — section 8.

6. Categories too thin to be worth a page

A category has its own page once at least ten products in it survive everything above. Below that, a category page would read as a market summary while resting on a handful of products, so the category has no page and is named, with its count, in the table of categories, which also gives the one exception. This site currently publishes eight categories.

Two things follow from that, and both are worth saying plainly. A category being absent is not evidence that installment plans there are cheaper or fairer — it means we could not identify enough products to the standard above. And the 89 products here are not a sample of the Egyptian market; they are what survived.

7. Lenders a shop offers but does not price

Several shops name lenders they work with and publish no terms for them before checkout. Those do not appear as empty rows. Where it applies to a shop on a product page, it is said in a sentence on that page.

8. Phones are screened differently

Rules 1 to 6 exist to make a cross-shop comparison safe. On phones we do not make one, so those rules are not what governs the phone pages.

The reason is identity. Three of the six shops we read for phones — five of which supply the listings shown here — publish no manufacturer barcode or model number on a phone page at all, so there is nothing to match them on. We do not match phones on their names alone: the same name can be two different products, and in this collection it is. Rather than drop the data or compare on a guess, the phone pages ask a narrower question first — at this shop, which lender and which length? — and every figure in that part stays inside one shop’s own listing. That is why 122 phone listings are shown across four shops while a two-shop rule sits above.

Two shops are compared on a phone in two places, and they are not the same claim. A phone proven identical by barcode is held to a stricter identity rule than anything above: the manufacturer’s barcode, printed by each shop on its own page and compared digit for digit. Three phones on this site meet that test. The two-shop rule in section 5 and the category gate in section 6 apply to the cross-shop comparison; where a phone is proven identical at only two shops, its own page says so in its header rather than claiming more than two prices can support.

The second is the ‘same model, storage, memory and colour’ section below it, and it proves nothing about identity: it puts side by side the listings on which each of the shops in a row prints the same model, storage, memory and colour, and its header says in its first clause that this is not a barcode and that the listings may be different regional versions with different warranties. 22 phones are compared that way. Where a shop prints a barcode or model code we check it, and a pair whose codes disagree leaves the section — for the one phone in this collection where they do, the whole group left on the day the section shipped.

Everything else on this page still holds on phones: a shop that contradicts itself is dropped, a plan that cannot be shown to be current is held back, and a lender a shop names but does not price is a sentence rather than an empty row. The phone pages.

How much of the lending market this is

Against the financial regulator’s register, re-captured 9 September 2026, 33 licensed lenders were in scope for this work. The project holds at least one collected row for 14 of them and a computed effective rate for 13 — 39.4%; the fourteenth appears on a shop page as a named provider whose installment table is empty, so it is identified but never priced. The denominator has not moved since it was first captured on 24–26 August 2026, and both counts remain ceilings: a register can lag a licence it has already granted. Of the 20 it does not price, 18 leave no public trace at all: no shop listing, no rate card, nothing.

Read as relationships rather than lenders, the fraction is smaller: the project holds a priced offer for 12.5% of the 1,094 lender-and-shop pairs it can see. In most of the rest, a lender names a shop in its own network and the shop publishes no lender list of its own.

Both are counts against a dated denominator, not a claim about which lenders matter. And 39.4% is a ceiling, not a current figure — though for a narrower reason than we used to give. The register the regulator lets you download still stops at licence No. 48 from May 2024; re-fetched on 9 September 2026 it was byte-for-byte the same file, and the regulator’s own index page still links it as the current one. But the regulator’s searchable register goes further. Opened one company at a time, it publishes a licence number and a licensed activity for each of the 388 companies it lists, and it carries consumer-finance licences up to No. 55 — including the No. 54 we previously had only from a lender’s own website footer. Every licence beyond No. 48 belongs to a company already on our list, so the denominator is 33 on the regulator’s current record rather than an estimate from a stale one. One gap remains and we cannot close it: three licensees — Mnzl, ADVA and Telda — have a register entry that prints no licence number at all, so 55 is the highest number the regulator publishes, not provably the highest it has issued.

What this site does not do

It does not score, rank or badge lenders. It does not publish averages by category: the arithmetic that would produce them is driven by shops’ rate cards rather than by anything about the category, so a category average would imply a difference that does not exist. It does not add up costs across plans of different lengths and present the sum as a finding.