French staffing application form benchmark 2026: ten staffing brands scored

For a staffing agency, candidates are the supply side of the business. Every temp placed starts as a visitor who found a job ad and pressed “Postuler”. What happens next, a short form or a log-in wall, decides whether the agency gains a candidate or the visitor applies somewhere else. We audited the candidate application forms of ten of the largest staffing brands in France to see how they have built that step, and where they lose candidates.

Key findings of the French staffing application form benchmark

• The ten scored brands average 6.4 out of 10. Triangle Intérim leads with 8.0, just ahead of Randstad with 7.9. Manpower is last with 5.0.
• Six of the ten brands create a candidate account before or while the candidate applies, five of them as a requirement.
• Two of the ten largest groups could not be benchmarked at all. House of HR only accepts candidates in its mobile app, and Groupe Crit asks for a password before any personal detail.
• A candidate fills in a median of 8 fields before leaving the first page with personal details. Manpower asks for the most, 14.
• Validation and AI readiness are the gaps every brand shares. Eight of ten give no inline feedback while the candidate types, and none of the ten forms can be read and completed by an AI agent.

Source: Exatom form-audit engine, desktop, 4 November 2026.

How French staffing brands score

The audit gives each form a score from 0 to 10. In a multi-step flow the score is taken on the page where the candidate first fills in personal details. Where an account is required, that is the sign-up page.

Position	Brand	Overall score
1	Triangle Intérim	8.0
2	Randstad	7.9
3	Adecco	7.0
4	Proman	6.4
4	Actual	6.4
6	Start People (Recruit)	6.3
7	Synergie	6.1
8	Gi Group	5.9
9	Hays	5.3
10	Manpower	5.0

The ranking is tight: six brands sit between 5.9 and 7.0. The two leaders do the basics well across the board. Triangle Intérim has a short, clean page with linked labels, the right input types and a perfect style score; Randstad is the most consistent, with no scored category below 7.3 apart from AI readiness. At the bottom, Manpower combines the longest form with the weakest usability, and Hays, despite one of the shortest forms, loses its whole web-standards score on labels and input types. The scores per category are in the benchmark report.

These scores come from the current version of the audit engine, which added the Privacy and AI readiness categories. They cannot be compared with the German car insurance benchmark, which used the earlier version.

Group size does not predict form quality

The best form belongs to Triangle Intérim, a French independent outside the European top ten, and the second best to Randstad. Adecco, the largest group, is third. Manpower, third by market share, is last: it combines an account wall, 14 fields and switched-off autocomplete.

The account comes before the job

Eight of the ten brands collect the application on a single form page. Only Adecco and Triangle Intérim split it into steps. A one-page form does not mean a one-click start, though. Only Hays and Actual open the form straight from the job ad, and Actual goes furthest by placing it on the job page itself. The other eight put at least one screen in front of the form, and that screen is almost always about the account, not about the job.

• Choice screens. Randstad, Proman, Synergie and Start People first ask how the candidate wants to apply: with or without a CV, with a new or an existing account.
• Log-in walls. Manpower, Gi Group and Triangle Intérim show a log-in first, with account creation as the way through. At Gi Group the apply button opens a separate portal on another domain. Triangle Intérim makes the wall lighter by using the phone number instead of a password.
• E-mail first. Adecco asks only for the e-mail on the first screen, then uses one long page to create the account and send the application.

Five brands require an account (Adecco, Manpower, Gi Group, Start People and Triangle Intérim), and Synergie’s form states that applying also creates a candidate space. Manpower and Start People ask the candidate to choose a password and type it twice in the application form itself.

The scores do not reward or punish the account as such: the four guest forms average 6.5, the six account flows 6.4. The cost of the account shows up in the number of screens and fields, and in the two top-10 groups that dropped out of the benchmark because the account comes before everything else.

Eight fields from job ad to application

Counted from the job ad up to and including the scored page, totals run from 6 to 14 fields, with a median of 8: one or two application fields, six personal fields and one or two consent boxes. Unlike a car insurance quote, the candidate is mostly asked about themselves; the CV does most of the work on the application side.

• The shortest forms, Actual and Triangle Intérim (6) and Hays (7), ask for name, e-mail and sometimes phone, plus the CV.
• The longest, Manpower (14), Adecco (13) and Synergie (12), combine account creation with the application: password and e-mail confirmation, desired jobs, availability, a disability question, a sponsorship code.
• Start People asks for postcode, city and a password, but not for the candidate’s name.

What every staffing brand gets wrong

Finding	Count	What it means
Not ready for AI agents (no WebMCP)	10 of 10	An AI assistant cannot fill in or submit the application
No inline validation	8 of 10	Errors only appear after pressing the button
Field borders with low contrast	7 of 10	Inputs are hard to locate, above all for low-vision users
Account created when applying	6 of 10	Password, confirmation or log-in screens before the first job-related question
Labels set too close to the fields	5 of 7	Label and input blur together, so the form is harder to scan
No mobile keyboard for phone or postcode	4 of 10	Candidates on a phone get a full keyboard for digits

Validation is the weakest scored category, with an average of 3.4. Only Randstad and Gi Group give any inline feedback; the other eight wait until the candidate presses the button. AI readiness is 0 for all ten: no form exposes its fields to AI agents through WebMCP. As candidates start to use AI assistants to search and apply for jobs, a form an agent can read and complete becomes a new entry route.

How we researched the benchmark

Which brands are in the benchmark

The starting list is the ten largest staffing groups in Europe by market share (SIA). Each group was audited on its French candidate site. Where a brand could not be researched, the next largest staffing group with a French candidate site took its place: Actual and Triangle Intérim. SIA publishes ranks beyond ten only to members; the groups outside the top ten each hold less than 1% of the European market.

Design data

Design attributes come from walking each application journey by hand, from a job ad on the French site to the first page that asks for personal details, with a fictitious candidate living in Paris. For each brand we recorded whether the flow is one-step or multi-step, whether an account is needed, and how many fields are asked.

Score data

Scores come from the Exatom form-audit engine, which evaluates the HTML of a form: labels, input types, autocomplete, contrast, spacing, validation, call-to-action, page speed, privacy cues and readiness for AI agents. The audit ran on 4 November 2026 in desktop Chrome. No application was submitted and no account was created; all entries were fictitious test data.

This benchmark contains no behavioural data. Scores describe how a form is built. They do not show where candidates give up.

What staffing brands can do next

Four priorities follow for any staffing brand in this list.

1. Let candidates apply first and create the account afterwards. Actual and Randstad show that a CV plus four to six fields on or next to the job ad is enough. If an account is needed, Triangle Intérim’s phone-based log-in without a password is a lighter alternative than a password typed twice.
2. Fix the markup. Link every label to its input, use the e-mail and phone input types, keep autocomplete on and give fields a visible border. For Hays, Proman, Start People and Gi Group this is the largest single gain available, and it changes nothing in the design.
3. Validate while the candidate types. Confirm a valid e-mail or phone number field by field instead of after the submit button. Eight of ten brands do not do this today.
4. Prepare for AI agents. No form in the benchmark is ready for WebMCP. Exposing the application form to AI assistants is a new lever that nobody in the French market has pulled yet.

What the account walls and confirmation fields really cost in applications is something only live data shows. Form analytics measures drop-off per screen and per field on your own application form, and that is where the optimisation effort should start.