Key Takeaways
- Write the rubric before the first question. A field that feeds no score line and no filter has no reason to exist.
- Field count hurts more than page count. Baymard's 2024 checkout benchmark found field count matters more to completion than step count.
- Anything you sort or filter on must be a closed option set. Revenue as free text cannot be banded or ranked later.
- Cap prose questions and name their criterion. Two capped paragraphs beat six open boxes nobody can compare.
- A shared rubric raises reviewer agreement, it doesn't create it. Budget for a moderation step anyway.
Most application forms can't be scored. They collect prose where they needed a band and free text where they needed a filter, then hand a reviewer 140 submissions and a spreadsheet. This one gets built backwards, from the decision it supports. The worked example: a business support program, 12 places, 140 applications, three reviewers.
Skip this if: you need a statutory form. Passport, benefits, tenancy and immigration applications have mandated fields and legal retention rules that override every design choice here, and the question set is not yours to shorten.
Why most application forms can't be scored
The failure is not length. It's that the form and the decision were designed by different people at different times, so the fields don't line up with anything a reviewer has to judge.
Three symptoms give it away. Reviewers retype answers into a shape they can compare. Two reviewers score the same application several points apart and neither can say why. And the shortlist conversation keeps returning to information nobody asked for.
Length still costs you something at the front end. Baymard Institute's checkout research, published June 2024, put the average e-commerce checkout at 5.1 steps and 11.3 form fields, and found 17% of users have abandoned an order because the process was too long or complicated. The useful part for form builders is the conclusion: step count matters less to performance than the total number of fields a person has to manage. That's checkout, not applications, and applicants are more motivated than shoppers, so treat the direction as transferable and the figures as not. Cut fields, not pages.
Step 1: write the scoring rubric first
Open a blank table before a form builder. Name the criteria, weight them, and write what a strong answer looks like. For the 12-place program that lands at five criteria and 100 points:
| Criterion | Weight | What a top answer shows |
|---|---|---|
Eligibility | Pass / fail | Registered in the region, under 250 staff, accounts filed for the last financial year |
CriterionEligibility WeightPass / fail What a top answer showsRegistered in the region, under 250 staff, accounts filed for the last financial year | ||
Problem clarity | 25 | One named problem with a timeframe, not a list of ambitions |
CriterionProblem clarity Weight25 What a top answer showsOne named problem with a timeframe, not a list of ambitions | ||
Evidence of traction | 25 | A revenue band and a direction of travel, plus something already attempted |
CriterionEvidence of traction Weight25 What a top answer showsA revenue band and a direction of travel, plus something already attempted | ||
Capacity to take part | 20 | A named attendee with authority and committed hours |
CriterionCapacity to take part Weight20 What a top answer showsA named attendee with authority and committed hours | ||
Fit with the program focus | 20 | One of the four focus areas chosen, with a concrete twelve-month outcome |
CriterionFit with the program focus Weight20 What a top answer showsOne of the four focus areas chosen, with a concrete twelve-month outcome | ||
Reference | 10 | A contactable referee outside the applicant's own business |
CriterionReference Weight10 What a top answer showsA contactable referee outside the applicant's own business | ||
Now the rule that does the work: every field in the form has to trace to a row in that table. Anything that can't is either a hard filter, an administrative necessity like a contact address, or deleted.
That test is unsentimental in a way committee discussion never is. The question about company history survives only if "history" appears in a criterion. It usually doesn't.
Step 2: turn each rubric line into one question
One criterion, one question, is the default. Splitting a criterion across three questions means scoring it three times and arguing about the weighting afterwards. Where a criterion has two halves, as "evidence of traction" does with a level and a direction, use two questions and say in the rubric how they combine, before the round opens rather than during scoring.
Decide the answer shape at the same time. For each question, pick one:
- A closed option set when you will filter, band or sort the answer. Our notes on multiple choice question patterns cover option lists that don't overlap.
- A capped paragraph when you need reasoning, with the cap in the label. Open-ended question design is a separate skill from option-list design.
- A file only when a document is the evidence and a typed answer cannot stand in.
Example questions for a business program application
Fifteen questions in three blocks. Blocks matter because the first one rejects applications without a human reading anything, and reading 140 applications at eight minutes each is 19 hours.
Block 1: eligibility, answered before anyone reads prose
Every question here is a closed set, because every one is a filter.
| Question | Field type | Options |
|---|---|---|
Where is the business registered? | Dropdown | The program's regions, plus "none of these" |
QuestionWhere is the business registered? Field typeDropdown OptionsThe program's regions, plus "none of these" | ||
How many people does it employ? | Dropdown | 1 / 2 to 9 / 10 to 49 / 50 to 249 / 250 or more |
QuestionHow many people does it employ? Field typeDropdown Options1 / 2 to 9 / 10 to 49 / 50 to 249 / 250 or more | ||
Which financial year did you last file accounts for? | Dropdown | The last four years, plus "not yet filed" |
QuestionWhich financial year did you last file accounts for? Field typeDropdown OptionsThe last four years, plus "not yet filed" | ||
Have you taken part in this program before? | Yes / no | |
QuestionHave you taken part in this program before? Field typeYes / no Options | ||
Four answers decide whether application 87 joins the reading pile. Nobody opens it.
Block 2: the scored questions
| Question | Field type | Criterion it feeds |
|---|---|---|
What problem are you trying to solve in the next six months? (150 words) | Paragraph, capped | Problem clarity |
QuestionWhat problem are you trying to solve in the next six months? (150 words) Field typeParagraph, capped Criterion it feedsProblem clarity | ||
What have you already tried, and what happened? (150 words) | Paragraph, capped | Problem clarity and traction |
QuestionWhat have you already tried, and what happened? (150 words) Field typeParagraph, capped Criterion it feedsProblem clarity and traction | ||
Last full-year revenue | Dropdown of bands | Traction |
QuestionLast full-year revenue Field typeDropdown of bands Criterion it feedsTraction | ||
How did that compare with the year before? | Dropdown | Traction |
QuestionHow did that compare with the year before? Field typeDropdown Criterion it feedsTraction | ||
Who will attend the sessions? | Short answer plus role dropdown | Capacity |
QuestionWho will attend the sessions? Field typeShort answer plus role dropdown Criterion it feedsCapacity | ||
How many hours a month can that person commit? | Dropdown | Capacity |
QuestionHow many hours a month can that person commit? Field typeDropdown Criterion it feedsCapacity | ||
Which focus area matters most to you? | Single select, four options | Fit |
QuestionWhich focus area matters most to you? Field typeSingle select, four options Criterion it feedsFit | ||
What would have to be true in twelve months for you to call this worthwhile? (100 words) | Paragraph, capped | Fit |
QuestionWhat would have to be true in twelve months for you to call this worthwhile? (100 words) Field typeParagraph, capped Criterion it feedsFit | ||
Three deliberate choices in there. Revenue is a band because bands compare across 140 applications and exact figures invite both rounding and refusal. The attendee question asks who, not whether, which is harder to answer aspirationally. And the twelve-month question separates applicants who want the program from those who want something it provides.
Block 3: attachments, reference and consent
| Question | Field type | Note |
|---|---|---|
Upload your most recent filed accounts | File upload, PDF | Last, because an upload is where sessions die |
QuestionUpload your most recent filed accounts Field typeFile upload, PDF NoteLast, because an upload is where sessions die | ||
Name and email of one reference we may contact | Contact fields | Outside the applicant's own business |
QuestionName and email of one reference we may contact Field typeContact fields NoteOutside the applicant's own business | ||
Consent to the stated processing of this application | Checkbox | Separate from any marketing consent |
QuestionConsent to the stated processing of this application Field typeCheckbox NoteSeparate from any marketing consent | ||
Keep the two consents apart. Bundling them makes the data consent unfree, and GDPR Article 5 sets out the data minimisation principle that also gives you the cleanest argument for deleting a field: you can only collect what is adequate, relevant and limited to what the purpose needs. A field that feeds no rubric row fails that test as well as the design one.
Step 3: set the field types so answers stay comparable
The question list is the hard part. These mechanics stop good questions producing unusable data.
Put labels above fields, not inside them. Nielsen Norman Group's guidance on placeholder text in form fields, last reviewed September 2018, is that placeholders make it harder to remember what belongs in a field and harder to check and fix errors, with an extra cost for users with visual and cognitive impairments. Persistent hints below the label do the same job without vanishing on focus.
One fact per field. "Name, role and email" in one box produces three formats across 140 applications and none of them sort.
Cap every paragraph and show the cap. A visible 150-word limit doubles as instruction: it asks for an answer, not a pitch.
Band numbers you plan to compare. Revenue, headcount, hours and budget all belong in bands. The exception is when a figure is itself the evidence, and then it belongs in an attached document.
Put the upload last. A file field is the heaviest ask in the form, and anything behind it is at risk.
The review checklist
A scoreable form is half the job. This is the part most form guides skip, and where the rubric from step 1 earns its time back.
- Run the filters as filters. Sort by the Block 1 answers and reject the ineligible before reading starts. On 140 applications that is usually 20 to 40.
- Score one criterion at a time, across every application. Going application by application means recalibrating five criteria on every record. Criterion by criterion holds one standard across 100 answers.
- Hide the names for the prose questions. Block 1 already decided eligibility, so the scored pass does not need to know who wrote it.
- Write one sentence of evidence beside every score. Not a justification: a quote or fact from the answer. This is what keeps moderation short.
- Compare scores before discussing them. Collect all three reviewers' sheets, then flag any criterion where they differ by more than two points. Discuss only those.
- Fix the shortlist size before scoring. Twelve places means a shortlist decided in advance, usually 20 to 25, or the threshold drifts to fit whoever scored well.
- Log a reason code for every rejection, and read the codes next round. If 30 rejections cite the same missing information, that is a Block 1 question you haven't written yet.
A shared rubric helps less than people expect, and it does help. A 2023 study in BMC Medical Education had four evaluators blind-assess 42 student portfolios using both an analytic rubric and a numeric rating scale, and found higher agreement among evaluators with the rubric, though the rubric also lowered the scores awarded. Structure is not a solvent either: a 2026 study in Practical Assessment, Research, and Evaluation found substantial disagreement persisting between raters on structured tasks, particularly on reasoning and communication, and that raters' confidence in their scoring was often out of line with how much they agreed. Budget for a moderation step rather than assuming the rubric removes the need for one.
Tools that help
Any form builder can hold the fifteen questions above. What differs is whether the submission triggers anything, where the answers land, and whether the data stays under EU law. Google Forms, Typeform and Jotform all cover collection competently and leave the review in a spreadsheet.
meetergo is a scheduling and revenue-pipeline platform whose form and funnel builder sits in the same product as its booking, CRM and signature tools. For an application round the relevant part is that a submission can route and book rather than just file, on EU servers with no US corporate parent in the chain.
20+ field types, covering the three an application needs
Rating and NPS, file upload taking multiple files up to 10 MB each, signature, and payment fields (checked 2026-11-04). The upload and signature types let Block 3 live in the same form as the rest rather than in a follow-up email.
Conditional logic with point scoring
Branch on an answer to show or hide later questions, and assign points to options as people choose them. Honest scope: this scores closed option sets, so it covers Block 1 and part of Block 2's traction lines. The two capped paragraphs still need the human rubric.
Multi-step funnels and routing to a booking
Blocks become steps under one shared theme, and the answers can route the respondent to the right person or calendar at the end. Lead routing patterns carry over to routing a shortlisted applicant into an interview slot.
Where the submissions land
Every submission flows into the built-in CRM, and submissions can be exported to PDF (checked 2026-11-04). CSV export of appointment data is on every plan including the free one. PDF templates and e-signatures cover the acceptance paperwork for the 12 applicants who get in.
Honest limitations. There is no application-review workspace: no reviewer assignment, no blind-review mode, no side-by-side scoring screen. The seven-point checklist still runs in a spreadsheet or your CRM, and meetergo's part ends at collection, scoring of closed answers and routing. Access control is password protection and submission limits per form, with user and rights management from the Suite tier (checked 2026-11-04). Multi-step funnels and smart routing also sit on Suite at €29.90/mo rather than the free plan, so a single-page application form on the free tier is the honest starting point (see current pricing). For a one-off round of 40 applications, a free form and a spreadsheet is the right answer.
If the applications feed a sales or partner pipeline rather than a one-time cohort, the collection and routing step is worth testing in one tool. Start free with meetergo to build the Block 1 filters and see where they route; no credit card required.
Common mistakes
Asking for a document you won't open. A business plan upload that no criterion scores costs every applicant an hour, and costs you storage plus a retention obligation.
Making the long questions required and the filters optional. It happens more often than it sounds, because the filters get added last. Reverse it: Block 1 required, prose capped, Block 3 required only where a criterion depends on it.
Collecting a number you'll later band. Exact revenue typed as free text gives you 140 unique strings. If the rubric says band, ask for the band.
Using the same form for first-time and returning applicants. Returning applicants abandon questions they answered last round, and that abandonment reads as a weak application. One routing question in Block 1 and a shorter second path fixes it. The patterns in our piece on standardised forms apply directly.
Scoring before the shortlist size exists. Deciding how many places are on the shortlist after seeing the scores is how a 12-place cohort becomes 17 people and a budget conversation.
Booking + video conferencing in one tool.
Booking + video conferencing in one tool.
FAQs
How many questions should an application form have?
For a competitive program, twelve to eighteen, filters front-loaded, no more than three open text boxes. The real ceiling isn't a number, it's the point where a reviewer stops reading every answer. With five criteria, more than two questions each means you're collecting material you won't score.
What is the difference between an application form and a registration form?
A registration form records someone who already qualifies, so its job is accuracy. An application form supports a decision between people, so its job is comparability. That's why registration forms can take free text where application forms need option sets. Signup and registration form patterns solve a different problem.
Should an application form be one page or several?
Several, split along the blocks, so eligibility is settled before anyone invests in prose. The field count is what you should be cutting, not the page count, per the checkout research above.
Can I score an application form automatically?
Partly. Closed option sets can carry points and produce a running score as people answer, which handles eligibility and banded evidence. Open text cannot be scored this way without a human or a model reading it, and either way you need the rubric first.
How long should I keep rejected applications?
Only as long as the stated purpose needs, and say the period on the form. Data minimisation and storage limitation point the same way, and a named retention period is easier to defend than a deletion you have to justify later. Our notes on GDPR compliance strategies cover the posture.
Write the rubric, delete every field that doesn't trace to it, make anything you'll sort a closed option set, and put the upload last. Then decide what happens when someone hits submit, because the form will not decide that for you, and the gap between a submission landing and a reviewer scoring it is where application rounds fall apart.
