How to compare products without losing a Saturday

Everyone has done this. You need a robot vacuum. Four hours later you have nineteen tabs open, three of them are the same product on different shops, and you have somehow developed strong opinions about lidar. You still cannot say which one you should buy.

The problem is not that you lack information. It is that you gathered it in the order the internet handed it to you, which is the worst possible order.

The tabs are the symptom, not the problem

Open shopping research always fails the same way. Each product page is designed to be read alone, so each one presents whichever specs flatter it. One listing leads with suction in pascals. The next does not mention pascals at all and leads with brush width. A third gives you a runtime figure measured in a mode you would never use.

By the time you reach the fourth tab you have quietly adopted a different set of criteria than the one you started with — the union of whatever every shop chose to advertise. That set is large, inconsistent, and mostly irrelevant to you. It is also unfalsifiable: you cannot compare products on a criterion that only one of them publishes.

Pick your columns before you open a single tab

The fix is almost annoyingly simple. Decide what you are comparing on before you look at anything you might buy.

Write down four to six things, and be specific enough that a stranger could fill them in for you:

What must be true — the hard constraints. Fits under 12 cm of clearance. Works without a subscription. Ships to your country.
What you are trading off — the two or three axes where more costs more. Price, battery life, noise.
What you will never actually use — write these down too. They are the specs that will otherwise eat your afternoon.

Four to six is not arbitrary. Below four you are not really comparing; above six or so, every product wins on something and the table stops producing a decision. If you find yourself at eleven columns, at least half of them are things you would not change your mind over.

Fill the table for every candidate, including the blanks

Now go collect. The rule that matters: every product gets every column, even when the shop does not publish the number.

A blank cell is real information. If three of five listings do not state a noise level, that tells you noise is not something this category competes on, and you should probably drop the column. If exactly one product omits the figure that its four competitors all advertise, that is worth a second look.

This is also the step where comparison shopping actually goes wrong in a way that costs money. It is very easy, on tab fourteen, to write down the sale price for one product and the list price for another, or to record a capacity in litres for one and quarts for the next. A table makes those errors visible because they sit next to each other in a column. Nineteen tabs do not.

Sort by the thing you are actually optimising

With a filled table, the decision is usually mechanical. Drop everything that fails a hard constraint — that is normally half the list. Then sort the remainder by your primary trade-off and read the top three rows.

If two products are genuinely tied after that, they are tied. Buy the cheaper one, or the one with the better return policy, and stop researching. The hours you would spend separating two equivalent options are worth more than the difference between them.

Where the software comes in

None of the above needs a tool. You can do it in a notebook, and for a €30 purchase you probably should.

It becomes worth automating at the point where the collection step dominates — where you have eight candidates and six columns and are facing forty-eight manual lookups across sites that all format their specs differently. That is what Comparadise does: you paste the product links, name the columns you decided on, and the extraction fills the grid. The columns are yours, so the table answers your question rather than the union of what each shop wanted to advertise.

Some worked examples, if you want to see the shape of a finished table before building one:

Notice what those have in common: not many columns. The coffee machine table is useful because it is six columns wide, not despite it.

The short version

Write four to six columns before you open a tab.
Fill every column for every candidate, and leave the blanks blank.
Cut everything that fails a hard constraint.
Sort by your one real trade-off, read the top three, buy one.

The method is the point. The table is just where it becomes visible.