The 7-Point Candidate Experience Audit You Can Run on Your Own Site
A candidate experience audit is a timed walkthrough of your own site, run as a candidate would run it, on the device they actually use. It takes under an hour and needs no tooling beyond a phone and a stopwatch. The seven checks below each produce a number you write down, and each is measured against a published benchmark rather than a feeling.
This is the method. If you want the diagnosis of where and why applicants drop out, that is a separate piece: why candidates leave a recruitment website without applying walks the five-stage funnel. Read that one to understand the leaks. Read this one to measure your own.
Before you start
Two rules make the difference between an audit and a browse.
Use a device you have never signed into the site on, and use mobile data rather than office wi-fi. Your own laptop has your cache, your cookies and a fibre connection, none of which your candidates have. Appcast's Mobile Recruitment Trends 2024 puts 67% of job applications on mobile, so a desktop-only walkthrough tests the minority experience.
Record numbers. "The form felt long" is not something you can act on in three months' time. "Eleven fields, 4 minutes 20 seconds, two errors on submit" is.
Check 1 · Can a candidate find your vacancies at all
Open your homepage cold and count the taps to a live vacancy. Then search your agency name plus a role you are hiring for, and see whether a vacancy page comes back.
This sounds beneath auditing until you look at the base rate. In our June 2026 audit of 391 UK and US recruitment agencies, 66% of the sites that responded had a live vacancies page. A third of agencies whose whole business is filling roles had no browsable list of the roles.
Record: taps from homepage to a live vacancy. Pass: two or fewer, and the vacancy list is reachable without using site search.
Check 2 · Time the application, do not count the fields
Field counts are a proxy. Duration is the thing that correlates. Appcast's Recruitment Marketing Benchmark Report 2024 found applications completed in under five minutes convert at 12.5%, while those running past fifteen minutes fall to 3.6%. The same report attributes 60% of abandonment to length or complexity.
So start the stopwatch when you tap Apply and stop it when you see a confirmation. Do it honestly: type a real cover note if you ask for one, upload a real CV, answer the screening questions properly.
Record: elapsed time, field count, and how many times you had to leave the flow to find something (a file, a reference, a date).
Pass: under five minutes end to end, with nothing that forces the candidate out of the flow.
Check 3 · Do the whole thing on a phone, including the CV upload
The upload step is where mobile applications die, and it is the step nobody tests, because the person testing has the CV sitting on their desktop. On a phone the candidate has to find a document in cloud storage, grant permissions, and wait on a mobile connection.
Run it as a candidate would: no CV saved locally, no desktop fallback. Note whether the form survives a rotation, whether the keyboard covers the field you are typing into, and whether a validation error scrolls you back to the top and clears what you entered.
Record: whether the application completes on mobile at all, and where it first broke if it did not.
Pass: completed on a phone, first attempt, without switching devices.
Check 4 · Load a job page cold and watch the clock
Google's page-experience guidance puts roughly a 7% conversion reduction on each additional second of load time. On a job detail page that cost lands on the exact click you are trying to earn.
Load a job page from a fresh session on mobile data and time it to the point the job title is readable. Then reload with your browser's cache disabled and note the difference. The gap between the two is what a first-time visitor pays.
Record: seconds to a readable job title, cold.
Pass: under three seconds, and no layout jump that moves the Apply button after you have reached for it.
Check 5 · Ask what a candidate can verify about you
A candidate deciding whether to hand you their CV is making a judgement about a company they have never dealt with. What is on the site to help them?
Our audit found 44% of the 391 agencies run a blog or insights hub, and 3% have live chat. Most agencies give a candidate nothing to read and nobody to ask. Consultant profiles, sector pages with actual detail, and a named person against the vacancy all do the same job, which is turning an anonymous form into a company.
Record: whether the vacancy names a consultant, and whether a candidate can reach a human without submitting the form first.
Pass: both.
Check 6 · Read what you send after submit
Submit the application properly and then read the confirmation screen and the email as if you were waiting on them. Two questions: does the candidate know what happens next, and do they know when?
A confirmation that says "we will be in touch" sets no expectation, so the candidate applies elsewhere as insurance and takes whichever offer moves first. A confirmation that says "a consultant reviews applications for this role on Tuesdays and Thursdays, so you will hear from us by Friday" is the same effort and a different outcome.
Record: the exact words of the confirmation, plus whether an email arrived and how long it took.
Pass: a named next step with a stated timeframe, delivered on screen and by email.
Check 7 · Check you could see any of this in your own data
The first six checks tell you what happened to you once. Instrumentation tells you what happens to everyone.
Across the 391 agencies we audited, 53% run GA4, 11% have a heatmap or session-recording tool, and 1% use anything that identifies visiting companies. Most agencies can see that traffic arrived and nothing about what it did on the way out.
Record: whether you can answer, from your own analytics, how many people started an application last month and how many finished.
Pass: you can produce both numbers without asking anyone.
Scoring the audit
| # | Check | Benchmark |
|---|---|---|
| 1 | Vacancies findable in ≤2 taps | 66% of responding sites have a live vacancies page |
| 2 | Application under 5 minutes | 12.5% completion under 5 min vs 3.6% past 15 |
| 3 | Completes on mobile, first attempt | 67% of applications are submitted on mobile |
| 4 | Job page readable in under 3 seconds | ~7% conversion cost per additional second |
| 5 | Named consultant and a human route | 44% run a blog; 3% have live chat |
| 6 | Confirmation states next step and timing | — |
| 7 | Start and finish counts available | 53% run GA4; 11% have a CRO tool |
Write down the seven results and keep them. The score matters less than having a baseline you can re-run against. If you fail more than four, treat it as sequencing rather than a redesign: checks 2, 3 and 6 are the cheapest to move and the ones candidates feel first.
Run it with someone who does not work for you
You cannot audit an interface you built. You know where the menu is, you know the CV is called cv-final-v3.pdf, and you skim the copy you wrote. Give the seven checks to three people outside the agency, ask them to send you their numbers, and compare.
The marketing-side equivalent of this audit is the 34-signal marketing benchmark, which scores what your site does for search, capture and AI visibility against 391 of your peers. Between the two you have both halves: what a candidate goes through, and what the site is doing while they do it.
Frequently asked questions
How long does a candidate experience audit take?
Under an hour for one person running all seven checks on one device. Add a second run on a phone, on mobile data, which is the one that finds most problems.
What should a recruitment application take to complete?
Under five minutes. Appcast's 2024 benchmark reports 12.5% completion for applications under five minutes, falling to 3.6% once they pass fifteen.
Why test on mobile data rather than office wi-fi?
67% of job applications are submitted on mobile (Appcast, 2024). Office wi-fi and a cached browser hide the load times and upload failures your candidates actually hit.
What is the most common failure in these audits?
The CV upload on a phone, followed by the confirmation message. Both are invisible to anyone testing on a desktop with a CV already on it.
How often should the audit be repeated?
Every quarter, and after any change to the application form or the job template. Re-run all seven so the numbers stay comparable.
Sources: Redsun Research, The State of Recruitment Marketing 2026: first-party audit of 391 UK and US recruitment agencies, June 2026 (vacancies page, blog, live chat, GA4 and CRO shares; behavioural figures are share of the 367 sites that responded). Appcast, Recruitment Marketing Benchmark Report 2024 (completion by duration; abandonment). Appcast, Mobile Recruitment Trends 2024 (mobile share). Google, Core Web Vitals and Page Experience (load time and conversion).