Dropdowns vs free text

A dropdown gives you data you can count and answers that may not be true. Free text gives you the truth and no way to count it.

That is the whole subject, and every practical decision follows from deciding which of those two costs you can afford for a given question.

The mistake most forms make is not choosing wrongly on any individual field. It is applying one preference to every field on the form, usually because someone decided that structured data is better.

It is better, when the question genuinely has a small set of correct answers. When it does not, structure manufactures data rather than capturing it.

The distinction is worth being strict about, because manufactured data is not merely useless. It is confidently wrong, it appears in reports as though it were observed, and decisions get made on it. A blank column would have been safer than a column of guesses.

A dropdown does not ask what is your situation. It asks which of these is closest to your situation, and it requires an answer.

When the visitor's real answer is on the list, this is excellent. It takes one click, produces clean data, and eliminates spelling variation entirely.

When it is not on the list, the visitor picks something adjacent. The value you receive looks identical to a true one and is wrong, and you have no way to distinguish them.

That is the central failure of forced choice: the errors are invisible. A blank tells you something is missing; a wrong selection tells you nothing at all.

The obvious fix is an Other option, and it does help, but it needs a text field beside it or it collects nothing.

An Other selected by fifteen percent of respondents with nowhere to elaborate is a signal that your list is wrong, and it is one of the more useful signals a form produces. Watch that proportion: when it rises, the categories have drifted away from what people actually want.

The related mistake is a list that is exhaustive on paper and unreadable in practice. Twenty options in a dropdown is a scrolling exercise, and people pick from what they can see rather than what fits.

Somewhere between four and seven options plus Other is the range where a dropdown is doing its job.

People say things you would not have thought to offer. That is not a defect to be engineered away, it is the entire value of an open field.

A message box regularly produces the sentence that qualifies the lead, the constraint that decides the price, and the phrasing your marketing copy has been missing. None of that survives a dropdown, and none of it would have occurred to you as an option to offer.

It is also the field that makes a screening decision possible at all, since an empty or nonsensical message is one of the clearest signals that a submission is not a person.

It is also the honest option. Nobody has to squeeze their situation into your taxonomy, which means nobody is submitting an answer they know to be inaccurate before you have even spoken.

The cost is real: free text cannot be counted, sorted or filtered reliably, and it takes longer to read than to write.

Use a short dropdown for routing, and free text for substance.

A single well-designed category field with a handful of options answers who should handle this and which queue it belongs in, which is the only structured question most enquiry forms genuinely need. The message box carries everything else.

This gives you a countable column for reporting and routing, alongside the detail that makes the enquiry answerable. Neither is doing the other's job.

Where a second structured field is proposed, apply the test for a required field before adding it.

The routing dropdown earns its place because it changes what happens next, which is the strongest justification a form field can have. It sends the enquiry to the right person and sets the right expectation, and both of those are visible in the row that lands rather than being reporting convenience.

A dropdown is a snapshot of how you thought about your business on the day you built the form, and it will be out of date before you notice.

Service lists lag what you actually offer. Industry lists lag the industries you have started selling to. Budget bands lag your own prices, which is particularly awkward because a stale band suggests a price you no longer charge.

Reviewing option lists is nobody's job, so it does not happen. The observable symptom is a rising share of Other and a falling share on options nobody selects any more, both of which are visible in the sheet at a glance.

Once a year is enough, and it takes ten minutes if you look at the distribution first.

Radio buttons show all options at once, which suits a short list where the choice is important and you want people to see the range. Dropdowns hide the options, which suits longer lists and questions of lower significance.

Checkboxes allow multiple selections, and they are where analysis gets harder than people expect: a column containing several values needs splitting before it can be counted, and the naive count of rows stops matching the count of selections.

If you use them, decide in advance how the values will be stored, because a comma-separated string in one cell is easy to write and irritating to analyse.

And avoid a dropdown with two options. That is a pair of radio buttons that has been made harder to use for no reason.

Some questions look structured and are not. Industry, company size band and role are the usual suspects.

People genuinely do not know which of your industry categories they belong to, and their answer varies by mood. Company size gets rounded and guessed. Job titles do not map onto anyone else's list of roles.

Where you need these, they are usually better inferred later from context, or asked in conversation, than collected at the door as a compulsory guess.

The exception is when the category changes what the form does next, in which case it is a routing question and belongs on the form. Everything else is reporting convenience being paid for by the visitor.

You do not have to choose between reading everything and counting nothing.

A prompt in the placeholder shapes what people write without constraining it: tell us what you need and roughly when produces messages that answer both without a single extra field. That costs nothing and improves the average message noticeably.

Afterwards, a category column that a person fills in while reading the enquiry gives you countable data without ever having asked. It takes a second per lead and the categorisation is far more accurate than the visitor's would have been.

That combination, honest input plus human classification, is what most small teams should be doing, and it is available in any spreadsheet.

Name, email, one short category dropdown for routing with an Other option, and a message box with a shaping placeholder. Optional phone.

Countable where it matters, open where it counts, and nothing on the form that asks someone to translate their own situation into your vocabulary.

The categorisation you actually report on gets added on your side, by whoever reads the enquiry, in a column beside it.

That arrangement also survives change better than a form does. When your services shift, you change how you categorise in the sheet rather than editing a live form, and the historical rows can be reclassified rather than being stuck with a taxonomy you have abandoned.

Structure a question when its answers are genuinely finite and you will act on the difference between them. Leave it open when the answer is the point.

And when a dropdown is right, keep the list short, include Other with a text field, and look at the distribution once a year.

How many options should a dropdown have?

Should Other always have a text field?

Are dropdowns better for data quality?

What is wrong with asking for industry?

How do I report on free text?

Radio buttons or a dropdown?

How do I handle multi-select answers in a sheet?

How often should I revisit my option lists?