Product and strategy · 5 min read
Ask what they are trying to accomplish
A contact form is usually the least designed thing on a website and the only one that carries the business. Treating it as a product changes the first question you ask.
The contact form is where most websites stop trying.
Everything upstream is considered — the typography, the case studies, the way the navigation behaves on scroll. Then the visitor arrives at name, email, message, send. The one interaction the entire site exists to produce is the one nobody designed.
It shows in what comes back. Enquiries that are one line long, or four paragraphs about a solution with no mention of a problem, or from people you cannot help at all. Then you spend the scarcest resource a small firm has — the founder's attention — working out which is which.
The first question decides everything
Most forms open with "How can we help you?", which invites a description of a solution the visitor has already chosen.
We open with: **What are you trying to accomplish?**
That is not a softer phrasing of the same question. It asks for the outcome rather than the implementation, and the answers are different in kind. "We need a mobile app" tells you what someone decided. "We are losing customers between quote and order and we do not know why" tells you what is actually true, and it may not need a mobile app at all.
"I'm not sure yet" is the most important option
Eight paths, and one of them is not knowing.
It gets identical visual weight to the other seven — same size, same treatment, no apologetic framing, no placement at the bottom as an afterthought. It has the same number of steps as every other path and asks the accomplishment question outright as a required field.
We test this rather than trusting it. There is an assertion that the not-sure path is never thinner than the leanest specified path, because that is exactly the kind of thing that erodes quietly during a redesign.
The reasoning is commercial as much as philosophical. A visitor who has already specified the solution has usually already decided, and often decided wrongly; the conversation is about price. A visitor who knows the goal but not the answer is the engagement where the work is worth most to them and most interesting to us. Designing an interface that treats that person as an incomplete version of a real lead is designing away your best clients.
Asking about budget without gatekeeping
Budget is the field everyone argues about. Demand it and you frighten off early-stage founders. Omit it and you spend hours on conversations that were never going to happen.
We ask, optionally, as a range, with this framing:
Not a quote. It helps us tell you quickly if we are the wrong fit, which saves you more time than it saves us.
The framing is the whole design. It makes the question a service to the visitor rather than an extraction from them, and it is true — a fast, clear "we are not right for this" is worth more to someone than three weeks of polite ambiguity.
Qualification by making fit visible
A form should let people work out they are in the wrong place, and let them do it early and without embarrassment.
That is a strange thing to optimise for until you count what an unqualified conversation costs. In a small firm it is not the meeting. It is the proposal, the follow-up, the internal discussion, and the attention that was not on delivery. Saying no well, quickly, is part of the strategy — and it produces referrals, which a slow non-answer never does.
The confirmation screen is a product moment
Almost everyone wastes it. "Thanks, we'll be in touch."
Ours reads the answers back — in the words the visitor chose, not the values we stored. Someone who selected "Within three months" sees "Within three months", not `quarter`. That was a bug we shipped first: the summary showed our internal values, which is not summarising back so they can see they were understood.
It also names who will reply and when, says the first conversation is a conversation rather than a pitch, and offers a way to add something they forgot.
**Only promise a response time you can hold during a delivery crunch.** One working day, held every time, beats four hours promised and missed once. That reply is the first evidence a prospect has of how you keep commitments, and it arrives before you have done any work for them.
The details that decide whether it works at all
Progressive disclosure — a visitor asking about growth never sees platform questions. Answers persist through a refresh or a back button. Validation on blur, never blocking forward movement. Errors that are specific and programmatically associated with their field, with focus moved to the first one.
**No CAPTCHA.** It is an accessibility barrier that punishes real people for a machine problem. Spam is handled with a honeypot field, a minimum fill time, and per-client rate limiting — none of which the visitor ever sees, and all of which we verified by trying to defeat them.
The one number worth watching
Not submissions. **Per-step abandonment.**
It is the only metric that tells you *which specific question is costing you business*, and it turns the form from a static asset into something you can actually improve. Total conversion tells you there is a problem. Per-step tells you it is the budget question, on mobile, on the improve-an-existing-system path.
The other number worth watching is how many people choose "I'm not sure yet". If that stays near zero, the option is not really being offered as an equal, whatever the design says.