MarqIQ
Browse all help topics
Explainer6 min read

What counts as a junk form fill?

Last updated 25 September 2026Available onFree TrialProBusinessEnterprisePartnerThe MarqIQ tracking script has to be live on the pages your forms sit on, because the Form fills analysis reads the fills MarqIQ captured rather than the records in your tool. You also need one CRM or marketing tool connected, because the analysis sits on that tool's page. Starter does not include the Analysis screens.

Calling a fill junk is a claim about a person, so MarqIQ only makes it for a reason it can show you on the record. There are seven reasons, two guarantees about what is never junk, and a way to check a sample yourself before you act on the number. If you want to stop the junk arriving rather than count it, work down the reason breakdown below and start with the largest reason.

What happens before anything is called junk

MarqIQ removes repeats first. The same person filling in the same form twice within two minutes is one fill, not two — a form that fires two events, or a visitor who double-clicks, would otherwise inflate your volume before any judgement is made.

The Form fills analysis prints the four numbers so you can see the whole population at once: Captured, Recorded twice, Junk and Real fills. Every rate on that page is measured on real fills.

Note

The analysis reads up to 60,000 captured fills, the most recent first. Above that the Captured tile reads "the most recent; older fills are not shown" instead of "every fill MarqIQ recorded", so you always know which of the two you are looking at.

The seven reasons

A fill is tested against these in order, and is labelled with the first that matches. The reason is shown on screen with up to three masked examples, so you can recognise the addresses without being handed a list to copy.

ReasonWhat it means
Test entriesThe email is plainly a test — test@, example.com and the like. Usually your own team checking a form.
Throwaway mailboxesA disposable mailbox service, used to get past a form without giving a real address.
Email addresses that cannot existThe address is malformed, or it uses a domain ending that does not exist.
Made-up namesThe whole name is filler or keyboard mash — asdf, test test — not an unusual real name.
One visitor, many email addressesThe same browser submitted four or more different email addresses. One person does not have that many.
Automated floodsEight or more fills on one form inside a minute, from different addresses, arriving from the same place. No group of people does that.
Filled in by a program, not a personHow the form was filled in gives it away. Read only for fills captured since you switched on fill-quality signals.

How a flood is separated from a busy minute

A flood has to be eight or more fills on one form inside a sliding minute, and they have to be eight or more different addresses. Eight fills from one address is a repeat problem, not a flood.

A flood also comes through one door. When a single source accounts for three-quarters or more of that minute, only the fills from that source are flagged — somebody who happened to submit at the same moment from somewhere else is not swept up with it.

How a fill is judged automated

This reason reads how the form was filled in, and only when you have switched the setting on. Any one of four things is enough.

  • The browser declared itself to be driven by automation.
  • It came from a headless browser, the kind test and scraping tools use.
  • It came from a program that identifies itself as a bot or a script rather than a browser.
  • It was submitted within a second and a half of the page loading, with no key press, click or touch at all.

Note

A fast fill by a person using autofill still involves a click, so it is never counted here. A fill with no signals recorded is not judged at all.

To switch this on, open Script & Deploy and turn on Spot form fills made by programs. It notes how long after the page loaded the form was submitted, how many times the visitor pressed a key, clicked or touched — a count, nothing more — and whether the browser says it is automated. It never records what was typed, reads no IP address and stores no browser fingerprint. It labels fills. It never stops anybody submitting a form. Only a workspace admin can change it.

The two guarantees

These are the answers to "could you be wrong".

  • An unusual real name is never treated as junk. Each word of a name is judged whole rather than searched for a pattern inside it, so Robert Young, Huy Tran and 林 修 are people.
  • A personal mailbox is never junk. Gmail, Outlook and Yahoo addresses are real people with real intent.

If you sell to companies, MarqIQ reports personal mailboxes separately, as a fact rather than a verdict, and tells you the share. That figure appears once at least 30 real fills carry an email address.

Checking a sample yourself

Do this before you take the number to anybody. It takes about five minutes.

  • Read the reason breakdown under What the junk is. One reason usually accounts for most of the count, and that reason tells you what kind of problem you have.
  • Take the three masked examples shown against the largest reason and find those people in your own tool.
  • Open How we worked this out at the foot of the page. It states how many fills were read, how repeats were handled, and both guarantees, in the numbers of your own workspace.
  • Compare the junk share by source. A source is called out only when it has at least 30 fills and its share is far from your overall share, so a small noisy source is never named on thin evidence.

Take care

The junk count describes the fills MarqIQ captured, not the records in your tool. If a form is not connected to your tool, or its email field is not mapped, junk was captured that never became a record — and cleaning your tool will not change this number.

If it did not work

A real colleague appears under Test entries

Why: Their address contains a word that is only ever test data. A mailbox at a domain like example.com or an address beginning test@ is judged on the address, not on the person behind it.

Fix: Agree one test address with your team and use it for every form check. It keeps your own tests together and out of everybody else's count.

The junk count is zero and you know you have spam

Why: Either the spam is arriving through a form the script is not on, or it is well-made spam — a believable name at a real domain, which none of the other six reasons can catch.

Fix: Switch on Spot form fills made by programs on Script & Deploy. It judges how the form was filled in rather than what was typed, which is what well-made spam cannot fake. It applies to fills from the next time a page loads your script, so it will not label the ones you already have.

Junk jumped in one month and you cannot see why

Why: A campaign started, or a form changed. The month-by-month view exists for exactly this.

Fix: Read Month by month, then Where the junk comes from. Between them they name the month and the source. If it is paid traffic, you are paying for it.

Next steps

Related articles

Was this article helpful?