How to judge a web app in ten minutes

Most evaluation advice tells you to list requirements and score candidates against them. That works for procurement and is useless for the actual decision, which is usually made in an afternoon between three tools that all look fine.

These are the checks that separate them, ordered so the fastest eliminations come first.

1. Can you get your data out? (two minutes)

Find the export function before you create anything. Not the marketing page's claim that export exists — the actual button.

Then look at what it produces:

• Open formats — CSV, JSON, Markdown, ICS, standard image and video containers. Your data is portable.
• A proprietary archive that only this tool can read. Your data is hostage in a way that will not become obvious until you try to leave.
• PDF only. That is a printout, not an export.
• No export, or export behind a paid tier. This is the strongest single signal about how the company sees the relationship.

Export quality predicts more than any feature comparison, because it determines whether choosing wrong is a reversible mistake.

2. What does it cost after the trial? (one minute)

Find the pricing page and answer three specific questions:

1. What is the price per user per month, billed monthly? Annual pricing shown as a monthly figure is the standard presentation and rarely the number you will pay at first.
2. Which feature you are about to rely on is on a higher tier? Export, permissions, integrations and version history are the four that most often sit one tier above where you expect.
3. What happens when you exceed the free tier? There is a large difference between "read-only until you upgrade" and "we delete the oldest items".

3. Does it work without an account? (thirty seconds)

For a single-purpose tool — converting a file, checking a value, generating something — an account requirement is a business decision, not a technical one. Tools that do the work in the browser without one exist for most of these categories.

For a collaborative tool, an account is obviously necessary. The question there is whether it requires a work account, an SSO provider, or a phone number, all of which have implications you should notice before your data is inside.

4. Who is behind it, and does that hold up? (two minutes)

Find the about page, the changelog and the status page.

• A changelog with recent entries is the best available evidence that a product is alive. A blog whose last post is two years old is evidence of the opposite.
• A status page with history tells you both that they take reliability seriously and how they behaved during incidents.
• A company address and a named team matter more than they seem, because they are what makes a data request or a refund enforceable.

Absence of all three does not make a tool bad. It makes it a tool you should not build a workflow on.

5. What does it do offline, and on a bad connection? (two minutes)

Open the app, then disable your network and keep working for a minute.

Three behaviours, in descending order of preference: it keeps working and syncs later; it goes read-only and says so; it silently discards what you typed. The third is common and only ever discovered at the worst moment.

6. Read one negative review properly (three minutes)

Not the star rating. Find a detailed one-star or two-star review and read what actually happened.

You are looking for whether the complaint is about a limitation you can live with or about behaviour you cannot — billing that continues after cancellation, support that does not respond, data lost during a migration. The first is a trade-off; the second is a pattern.

A tool with an honest limitation is a better bet than a tool with no visible complaints, because the second usually means nobody has used it seriously enough to have any.

The order matters

Run these in sequence and most candidates fall at step one or two, before you have invested any time in learning an interface. The mistake is doing it the other way round — spending a week getting comfortable with something, and only then discovering that leaving it means retyping everything.