The cheapest price is rarely the cheapest

Every comparison site ranks on the sticker price, and then the coupon turns out not to apply, or delivery adds a third again. This one computes the total first and ranks on that. Where it cannot establish a delivery cost, it says so rather than guessing.

Built and owned by BDGDublin, Ireland

Status
In development
Platform
Web
Built with
Laravel, SQLite
Market
Ireland

Who it is for

  • You are about to buy something and three tabs disagree about what it costs.
  • You have typed a coupon into a checkout and watched it be refused.
  • You would rather see one total than four numbers to add up yourself.

How it works

01

Coupons combined the way a checkout would

The engine tries only the combinations the coupons themselves allow — one on the product and one on delivery at most, and together only when both say they stack — and keeps whichever leaves the least to pay. Each one's validity window, minimum spend, ceiling and payment method are honoured, and a discount the shop has already applied is kept apart from the coupon so nothing comes off twice. Adding up every code that exists produces a total nobody can pay.

02

Whole cents, not floating point

Every figure is computed in integer cents, in one pure service with no network and no database behind it. A comparator that rounds differently from the shop is wrong in the only place it matters.

03

Delivery counted as part of the price

The total includes delivery to the postcode entered, and the payment method is part of the sum — card or PayPal, in Ireland. A first-purchase discount is counted only at the shops the buyer says they have never ordered from, not at every shop that offers one.

04

An unknown delivery cost is not ranked

Where delivery cannot be established the offer is still shown, and kept out of the ranking rather than estimated. A guess produces a cheapest offer that is not the cheapest.

05

Nothing about the search is kept

Searches and postcodes are not persisted, responses are marked no-store, and the endpoint takes thirty queries a minute. What somebody is shopping for is their business.

What it does not do

None of these are real prices yet

The current build queries no shop at all. Prices, coupons, sellers, delivery times and delivery costs are demonstration data, the DEMO codes cannot be redeemed, and there are no purchase links and no commission. Delivery changes with the start of the postcode for one reason: to show the ranking changing. Ireland is the only country open; twelve more are configured, and none of them is offered to anybody until we are actually pursuing it. Before any of it becomes real, each partner programme has to be read for what it actually permits — display rules, retention, channels — and the data has to arrive through an authorised API. Scraping without permission, and redistributing somebody else's named coupons, are the usual shortcuts and they are not ones we will take. DealSum is a working name; neither the brand nor a domain of its own has been validated.

Not built yet

In the order it matters. Published because a roadmap somebody can hold us to is worth more than a page that looks finished.

  1. Connectors with authorised access to partner APIs, one platform at a time.
  2. A persistent catalogue, matching on exact model, variant and barcode rather than on a name.
  3. Real delivery costs, with their origin and their expiry.
  4. Coupon and delivery benefits read from each shop's own rules, rather than from ours.
  5. An approved affiliate programme — declared where a reader can see it. The click log, the signed way out and commission reconciliation are built; what is missing is a programme to put a tag in.
  6. A review panel, queues and observability, so a wrong price is found before a customer finds it.
  7. Opt-in price alerts, with retention set for a live service rather than for a demo.

The same work we do for clients

Something like this, but yours

Tell us what you are thinking of building. We come back within one business day with an honest read on whether it is worth building — including when the answer is no.