Getting B2B Inbound Ready for the AI Era

A form can get you a way to reach someone. This is a study of what else B2B inbound forms are being asked to do. We rendered the conversion path of 15,237 companies and read every question on it: 11,257 forms and 71,965 labelled fields. This is what the corpus shows, with the method and the limits stated once.

How we did this

The sample was drawn from lists of companies identified as running a CRM or a marketing-automation product, from venture-capital portfolios, and from a purchased visitor-intent feed: 17,286 domains in all, of which 15,237 resolved, served a page inside 35 seconds and exposed a conversion path a headless browser could reach. For each domain we followed the public path an enquirer would take — the homepage, then the most prominent demo, contact, booking or free-trial entry point. Pages were rendered in a browser environment, waiting for client-side form embeds and late-loading fields, and every field label, dropdown option, required marker and submit button was extracted.

Final corpus: 15,237 company domains, 11,257 detected inbound forms and 71,965 labelled fields, crawled on 28, 29 and 31 August 2026.

How the questions were sorted. The charts on the page above sort every one of the 71,965 questions into five kinds by an ordered rule list applied to the question's label, its input type and whether it carries a list of options; the first rule that matches wins. In order: what the buyer needs (budget or revenue; timing; who decides; the problem, goal, requirement or use case; an open "how can we help" or "tell us about your project" question with no options; what they use today); for the company's own use (how the buyer found the site; marketing consent; a product, service, topic or interest picklist; interest checkbox lists); facts about their company (job title, company size, industry, location, website); basic contact details (name, email, phone, company name, in any language the corpus contains); and a blank box (a message box with no question attached). Captchas, passwords and the 8.1% of labels that match no rule are left out of every chart. Two further tags mark what a data provider of the Apollo or People Data Labs kind returns from a work email alone: strict (the company's name, size, industry, location and website, and the person's job title) and broad (strict plus the person's name and phone); the email itself is the key and is never counted..

Scope and limits: this study measures what public inbound forms ask, not whether those forms convert. No form was submitted. Because the sample is limited to companies already using CRM or marketing-automation software, the findings should not be generalised to all B2B companies — if anything, these are firms better equipped for lead capture than the market at large. The companies were drawn from lists built to find businesses that sell to other businesses; that selection was by list, not verified company by company, and about 1 in 12 describe themselves as also selling to consumers.

Each outside reference was checked against three published sources before it was used. Each is named in the sentence that uses it.

Three published buyer studies are cited in the report above. 6sense's 2025 Buyer Experience Report surveyed 3,744 business buyers, with a companion survey of 766, and was published in November 2025; 6sense sells sales software. Gartner's survey of 645 business buyers was fielded in August and September 2025 and published on 20 May 2026. Pew Research Center tracked the searches of 900 US adults through March 2025 and published on 22 July 2025.

Suggested citation. Formidable, The State of the B2B Contact Form, 2026. Crawl of 15,237 B2B inbound conversion paths, 28–31 August 2026. Second edition published 7 September 2026, corrected 10 September 2026.

The shape of inbound

Across the 15,237 companies, 74.5% place a form on the main path between a visitor and a conversation. A quarter offer a self-serve signup, though it is rarely the only route: 83.1% of companies with a signup also run a form, a booking link or a contact page. Only 4.7% offer all three.

A further 1,925 companies — 12.6% of the sample — present no form, no booking link and no signup anywhere the crawl could reach. Two thirds of those advertise a demo on their own homepage.

What the fields ask

Across the 11,257 forms, the median shows six fields and requires four. Nearly half, 46.3%, require three or fewer. The distinction matters when comparing a form against a published industry average, which is usually a count of fields shown rather than fields required.

Almost all of those fields are contact details. Fourteen of the fifteen most common labels in the corpus are a version of name, email, phone, company or country. The first label in the ranking that is not contact information is "message", at number eight. The first that is a question rather than a field is "How did you hear about us?", at number sixteen.

Qualification questions — size, budget, timing, industry, stated need — account for 12.1% of all 71,965 fields. This share was published as 6.0% and corrected on 10 September 2026, after an audit of 620 hand-labelled fields found the classifier was ignoring dropdown option lists and discarding checkbox arrays. Classifier-dependent secondary figures remain from the original derivation until the next full refresh.

At the level of an individual form the pattern is sharper. The median form asks no question that would establish whether an enquiry is worth a sales call. Only 13.3% ask even one strict qualifier. On a strict match of the field label, 1.7% ask about budget or timing, 4.7% about company size and 4.9% about industry.

Nearly one form in five consists of contact details alone. Another one in five is contact details plus a comment box. Together those account for 37.2% of the corpus.

Two further habits, both measured directly. Of the forms requesting a name, 61.2% split it into separate First and Last fields. And of those requesting an email, 32.0% specify a "work", "business", "corporate" or "professional" address in the label — a check on whether the enquirer represents a business, written into an ordinary contact field. It is the most common act of qualification in the corpus. The checking continues after submission: Chili Piper, a tool that routes form submissions, reports that 14.1% of roughly four million submissions were disqualified after the form was sent.

What companies write for themselves

Where a company writes an original question rather than using a standard field, the question is usually about the company.

"How did you hear about us?" appears on 917 forms, roughly twelve times more often than any other bespoke question, and 44% of the companies asking it will not accept a submission without an answer. More forms ask it than ask company size (722), industry (656), budget (237) or timing (108).

The question has a history. The case study that founded the short-form argument, published by the agency Imagescape in June 2008, cut a form from eleven fields to four; the required field it deleted was "How did you hear about us?". Eighteen years later it is the most common bespoke question in this corpus, and nearly half the companies asking it make it mandatory.

The same pattern appears inside dropdowns. There is no shared taxonomy for company size: across every size picklist in the sample, no single band accounts for more than 2.0% of options. And in industry, job-title and attribution picklists alike, the most frequent option is "Other".

The submit button

The most common label on a conversion button in this corpus is "Submit" — 3,919 of the 10,539 forms whose button could be read, or 37.2%. Widen the class to include Send, Send message, Continue, Next and OK and the share describing a database operation rather than the outcome the enquirer expects runs between 51% and 59%. The median button label is one word.

How forms get long

Length does not come from qualification. The accretion order across the corpus is legible: email, then name, then a free-text box, then company, then phone, then job title, then location — 26.1% of all forms ask for a country, state, city or postcode — then consent. Company size, budget and timing never take hold. Even on forms of 15 to 40 fields — at least two and a half times the median — only 11% ask company size, 10% budget and 7% timing.

Of the 278 forms asking sixteen or more fields, 70.5% ask nothing that would establish whether an enquiry is worth a sales call. The share of forms asking at least one qualifying question rises with length up to the eleven-to-fifteen band, then falls back beyond it.

Long forms are instead composed of multi-select interest checkboxes and postal address blocks. Of the fields on long forms the classifier could not name, 55.2% are checkbox arrays — interest[], services[], area-of-interest[] — and 78.6% of the fields on the longest forms are optional.

Teams building long forms appear aware of the length. Of forms with one or two fields, 52% mark every field mandatory; of those with 15 to 40, only 3% do.

Sorting forms by how much internal machinery they carry — a picklist, an attribution question, a location field: the three things that exist to route, score and report on a lead — shows what the length is for. 4,749 forms carry none of the three and average 4.53 fields; 3,455 carry one and average 6.47; 2,743 carry two and average 9.05; 310 carry all three and average 10.52. Across those four bands the share of forms with a free-text box is 40.5%, 40.0%, 43.3% and 41.3%. As a form takes on machinery it more than doubles in length, and the one field where the enquirer can say what they want does not grow with it.

The absence of qualifying fields is not new, and it is not an accident. The test that most published form-length advice rests on is Marketo's, written up by MarketingExperiments on 27 June 2011: it compared forms of five, seven and nine fields, and the fields it added were "# Employees", "Industry", "Work Phone" and "CRM System" — the asks this study finds almost nowhere. Interviewed for that write-up, Marketo's co-founder Jon Miller, then its VP Marketing, gave the reason: "Why should I ask for data, hurt my conversion rate and get bad data when I can buy it cheaper?" Fifteen years of argument about form length has been an argument about fields the market had already declined to ask.

Among the 1,007 companies running Marketo or Pardot — platforms rarely deployed without a demand-generation team — 54.5% ask eight or more fields, against 27.4% elsewhere in the sample. The best-resourced marketing teams in this corpus run longer forms, not shorter ones.

What almost nobody asks

Four questions would establish whether a new enquiry merits a sales conversation. Across 11,257 forms they appear as follows, counted by the semantic classifier rather than by literal label match, which is why they run slightly above the strict figures quoted earlier:

• Budget or revenue — 2.1% of forms
• Timing — 1.0%
• Team size or headcount — 6.4%
• Industry — 5.8%

A fifth field is a different case and is counted separately here, because an open box is not a question: the free-text field where an enquirer could describe the problem in their own words appears on 48% of forms, and nearly two thirds of those do not require an answer.

Figures behind the report

Added 15 September 2026, and superseded on 17 September when the charts above were rebuilt: they now sort every question into five kinds over 11,257 forms and 71,965 fields, and nothing in this block is the basis for any chart on the page. It is kept as the record of the 15 September derivation, which grouped every question into three kinds: contact (name, email, phone), lookup (job title, industry, company size, location, website: facts most teams can find from the email domain), and the buyer's needs (what they need, budget, when they plan to buy: answers only the buyer can give). Company size and industry count as lookup here; the "What almost nobody asks" section above, written earlier, counts them as qualifiers, which is why its figures run higher than the report's. The figures below were recomputed on 15 September from the same crawl with a rebuilt classifier; the recount finds 11,282 forms with at least one readable field and 73,254 fields, against the 11,257 and 71,965 published here, a difference of classifier edge cases. Shares are of 11,282 forms unless stated.

• The path to a sales meeting, 15,237 companies. A form (alone or beside a booking calendar): 11,347, 74.5%. A booking calendar and nothing else: 1,317, 8.6%. Only an email address or phone number: 1,121, 7.4%. Nothing the crawl could find: 804, 5.3%. Self-serve signup only: 648, 4.3%. A chat widget anywhere on the site: 2,345, 15.4%; of those, a conversational qualification tool: 642, 4.2%.
• Share of all 73,254 fields. Contact 50.9%. Lookup 13.0%. The buyer's needs 10.4%. Everything else (company name, "how did you hear about us", topic dropdowns, consent) 25.7%.
• Forms that ask at least one question of each kind. Contact 95.7%. Lookup 45.8%. The buyer's needs 57.9% (an open message box counts; 55.4% have one). Budget or timing 3.3%. Timing alone 1.2% (137 forms).
• Forms that require at least one question of each kind. Contact 77.8%; an email address specifically 76.3%. Lookup 30.2%. The buyer's needs 23.2%. So 76.8% of forms can be submitted with no required answer about the buyer's needs.
• "How did you hear about us?" 1,065 forms (9.4%), required on 44.0% of those; 7.8 forms ask it for every one that asks timing.
• Behind the button. Where the path's button says "demo" (3,576 forms): 14.5% require an answer about the buyer's needs and 54.3% ask a lookup question. Where it says "contact" (3,872 forms): 31.4% and 46.4%.
• Length. Forms with 9 or more fields: 20.5%. Forms with 16 or more: 292 (2.6%); of those, 82.2% ask nothing about budget or timing and 72.9% require no answer about the buyer's needs.
• By form software. Forms on HubSpot, Marketo or Pardot (3,035): 66.8% ask a lookup question, 20.6% require an answer about the buyer's needs, 7.2 fields on average. Other forms (8,247): 38.1%, 24.2%, 6.3 fields.
• Outside figures used in the report are listed under Sources with their sample sizes: 6sense (about 4,000 buyers; first vendor contacted wins 77 in 100), Gartner (645 buyers; 7 in 10 confirm with a salesperson), Pew Research Center (900 people; 8 in 100 click through with an AI answer, 15 without), WebAIM (top million websites; half have unlabelled form fields).

What we did not measure

This study describes what forms ask before a conversation begins. It does not describe what happens afterwards.

• No conversion data. No form was submitted, so the corpus contains no completion rate, no response time and no lead-quality measure. The study should not be cited as evidence that any form design increases or decreases conversion.
• No time series. One crawl over one week. The data supports no claim about change, trend or direction.
• Not generalisable to all B2B. Every company was drawn from a list of businesses already using CRM or marketing-automation software. The 1.7% that ask about budget or timing on a strict label match is therefore a plausible upper bound for the wider market rather than a typical figure.
• Predominantly US companies. 75.1% of domains are .com, and 0.31% of fields carry a label in a non-Latin script. German and Japanese B2B forms follow different conventions and are not represented here.
• Correction status. The qualification-field share was corrected on 10 September 2026; classifier-dependent secondary figures remain from the original derivation until the next full refresh.