Know what you actually pay
The shelf price is not the price. NiftySeeker tracks what retailers charge, works out the real landed cost including shipping, duty and fees, and tells you when something worth knowing changes.
No card required. No tracking unless you say yes.
The cheaper one is not the cheaper one
Two listings for the same robot mower, on the same day. The left is £60 cheaper on the shelf and £43 dearer on the invoice.
| Retailer A | Retailer B | |
|---|---|---|
| Shelf price | £899.00 | £959.00 |
| Delivery | £49.00 | Free |
| Import duty | £54.00 | £0.00 |
| Handling fee | £12.00 | £0.00 |
| Voucher applied | −£0.00 | −£12.00 |
| Landed cost | £1,014.00 | £947.00 |
Every line is stored as an integer number of pence against an explicit currency, with the date it was observed. Where a component cannot be established — a duty rate that depends on where you are, a delivery charge only shown at checkout — it is marked unknown and the total says so, rather than being quietly assumed to be zero.
Illustrative figures, not a live comparison.
What it does
One catalogue, many listings
The same product appears under a different name at every retailer. NiftySeeker matches them together — by identifier where one exists, by evidence where it does not — so you compare products rather than product pages.
The real cost, not the sticker
Shipping, duty, fees, and currency all land on the invoice. The landed cost calculation shows each component, and says plainly when a component is unknown rather than quietly guessing at it.
Prices over time
Every observation is kept, so you can see whether a discount is a discount or a return to the usual price. Charts are drawn on the server — no scripts, no third parties watching you read them.
Alerts that mean something
Rules you write: a price below a threshold, a drop by a percentage, stock returning. They fire on a change, not on every check, so a quiet inbox means nothing happened.
Purchase records
What you bought, from whom, for how much, with the paperwork attached and an approval trail. The part that matters at year end and is never there when you need it.
Exports and an API
CSV for a spreadsheet, JSON over a read-only API for anything else. Your data leaves as easily as it arrives.
How it is built
Decisions that are much easier to make at the start than to retrofit.
- Separated, not filtered
- Every workspace's data is isolated at the query level, on every query. Not a WHERE clause somebody has to remember.
- An audit trail nobody can edit
- Significant actions are recorded append-only, including from inside the application. Nobody can quietly rewrite what happened.
- Money as integers
- Amounts are stored in minor units with an explicit currency. No floating point anywhere near a price.
- Nothing loaded from elsewhere
- No CDN, no web font, no third-party script. Analytics runs only if you accept it, and is genuinely absent from the page if you do not.
- AI is optional and never automatic
- Connect your own key from any supported vendor, or none at all. Suggestions stay suggestions — a person always decides, and product data is all that is ever sent.
- Backups that are verified
- Nightly, checksummed, tested for readability, and pruned. A backup nobody has restored is a hope, not a backup.
What it does not do
- It does not sell anything, and takes no cut of what you buy.
- It does not promise a price is current. Retailers change prices constantly; their listing is always the authority, and the interface says when a figure was observed.
- It does not guess at a landed cost component it does not have. An unknown is shown as unknown.
- It does not scrape sites whose terms forbid it. What you monitor is your configuration and your responsibility.
Start with one product
Add something you buy regularly and a couple of the places you buy it. You will know within a week whether the price you are paying is the price you should be paying.
Create a workspace