Contact form or feedback widget: what it costs a visitor to tell you something is broken

A visitor notices that a filter on your site does nothing, or that the opening hours are last year's. Leaving takes one click. Telling you takes more than that, and how much more depends on what you put on the page: an email address, a contact form, or a small box that opens where they already are.

I'm a UX designer and I make a feedback widget, Two Cents, so I have a side in this. Each claim below links to the person who made it, and I say where the evidence is thin.

An email address on the page

The cheapest thing to build is a line in the footer with an address, usually as a link that opens the visitor's mail program.

Jakob Nielsen objected to those links back in 2000, in an article on mailing lists. People expect a click to show them another page, and this one drops them into a different application. It may also pick a sender address they didn't mean to use. Twenty years later Adam Silver and Amy Hupe listed what still goes wrong. The address is awkward to copy, and the link opens whichever mail app happens to be the default. For someone with no mail app set up, it leads nowhere. They argue from their own experience and cite no study, and I found no reliable figure for how many visitors that last case covers. Chris Coyier's guide to these links adds the webmail case, where the page they were reading can be replaced by Gmail.

Say the link works. The visitor still has to write a subject, explain which page they were on, and send all of it from their own address, under their own name.

They could also phone. For Nielsen Norman Group, Anna Kaley watched 20 business professionals use 40 corporate websites and wrote up what they made of the contact pages. When they had a question for a company, most of them wanted to call. They also got lost among numbers grouped by departments an outsider has never heard of, and nobody enjoys waiting on hold. For a typo or a dead filter, I doubt many people pick up the phone.

A contact form

A form keeps the visitor on your site. It has costs of its own.

In the same study, with a form, people felt they had lost control. They couldn't tell whether anyone would answer, or when, and they suspected their email address was being collected for marketing. Her advice is three to five fields, asking for nothing you don't need in order to reply, and never an account. The sites were corporate ones, so read it as a study of people with business questions.

The rest of the literature on forms points the same way. Kathryn Whitenton's ten recommendations open with length: each field you cut raises the share of people who finish. The GOV.UK Service Manual tells government teams to "know why you're asking every question". The Baymard Institute lists being made to create an account among the reasons shoppers give for abandoning a checkout. A shopper at least gets a parcel at the end. Someone reporting a typo is doing you a favour.

A short form on the page itself

The third option skips the trip to the contact page. Kaley's later guidelines on asking for feedback recommend a small feedback tab at the side of the page, there whenever the visitor wants it, and put a feedback survey at about a minute.

GOV.UK works this way. Its feedback component sits at the foot of its pages. A visitor who chooses to report a problem gets two questions, what they were doing and what went wrong, and stays on the page. When the team behind it added the link to every page of one section in 2013, reports were anonymous, which also meant nobody could write back. The design guide notes that some users are wary of giving an email address at all.

There is something in it for you as well. Nielsen Norman Group's framework for shortening forms goes eliminate, automate, simplify, and the second step says not to ask for what the system already has. A box that opens on a page already knows which page it is.

What the message leaves out

Then the message reaches you. Say it's an email that reads "the filter doesn't work". You don't know which filter, which page or what device, so you write back to ask, and a day goes by.

Software teams have measured this. Thomas Zimmermann and five colleagues surveyed developers and users of Apache, Eclipse and Mozilla, 466 responses in all, for a paper published in IEEE Transactions on Software Engineering in 2010. The things developers found most useful in a bug report, such as the steps to reproduce the problem, were the things reporters found hardest to provide. A 2019 paper by Oscar Chaparro and colleagues starts from the same observation: reports written by users often describe those steps poorly, and sorting them out takes time.

Both papers are about people who file bugs in open-source trackers, a more technical crowd than the visitors of a shop or a directory. Neither studied website visitors, and nothing in them suggests a casual visitor would do better.

Why I built Two Cents this way

While building cyclingkits.cc, my directory of cycling clothing brands, I wanted a quick inbox. With email or the usual contact form a visitor can't really be anonymous, and some people don't bother opening a mail client, writing and sending. A small widget you click, type into and send is far less friction. Every idea I had for it went one of two ways: easy for the visitor to leave their two cents, or easy for me to read it in context.

For the visitor, that means one sentence is enough and nothing pops up uninvited. The email field is optional, and a site owner can switch it off. What they typed survives closing the panel. The widget speaks their browser's language (English, French, Dutch, German or Spanish), sets no cookies, and its text meets WCAG AA contrast.

On my side, the page, the browser, the screen size, the language and how long the page took to load arrive with every message. A bug report adds the console errors and the requests that failed, and a screenshot if the visitor attaches one. Knowing which page a message was sent from is what lets me understand it correctly.

It still works. Between 13 and 21 September 2026, 35 messages arrived on cyclingkits.cc, 31 of them answers to one question, "Suggest a brand". Three came from the empty search results: someone looked for a brand I hadn't listed, found nothing, and named it on the spot. Each arrived with the address of that search, so I could see what they had typed. Messages still come in, and some visitors leave their email address so I can reply.

What a widget doesn't replace

Keep your contact page, phone number included. Kaley treats a form as an addition to a phone number and an address, and a widget is no different. It suits the visitor who has one thing to tell you and no wish to start a conversation.

Mine has limits you should know about. You can't answer from inside Two Cents: a reply leaves from your own mailbox, and it needs the visitor to have left an address. There is one login per account. Visitors can't draw on a screenshot yet, and nothing syncs to an issue tracker yet. I'm looking into both. Tools that do those things today are in Best website feedback tools in 2026, and I compared mine with two of them, Feedback Fish and Ybug.

A box on the page also won't make everyone speak. Why most visitors stay silent, and what you still have to learn by watching people, is the subject of my article on UX feedback.

Two Cents is free for one site and 25 messages a month, with no card.