To evaluate an auto apply bot, run it on a small, controlled batch of real jobs and measure three things: how many useful applications it sends per hour (speed), whether it keeps your account and reputation intact (safety), and whether the submitted answers are ones you would sign your name to (quality). A sales page cannot tell you any of this. A one-week test with ten to twenty jobs can. This post gives you that test, step by step, so you can judge any bot with your own evidence instead of someone else’s marketing.
Step 0: The paper check before you install anything
Spend fifteen minutes reading before you give a tool access to your browser or login. You are looking for answers to four questions:
- Where does it run? Some bots run inside your own browser as an extension. Others run on the vendor’s servers and act on your behalf. Browser-based tools see what you see and stop when you close the tab; server-based tools can run while you sleep but usually need more access.
- Which sites does it actually support? “Works on all job boards” is rarely true. Look for a specific list. A tool that clearly says what it does not support is usually more honest about what it does.
- What does it do when it does not know an answer? The right answer is “it stops and asks you.” The wrong answer is “it guesses” or silence.
- Can you export or see a log of every application? If you cannot see what was sent, you cannot evaluate it.
If a tool fails two of these four, skip it. You will save yourself a week.
Step 1: Set up a fair test
Pick one job title you genuinely want, one location rule, and one resume. Do not test with your entire search at once. Then fill the tool’s answer settings with truthful values: notice period, expected salary range, years of experience with your main skills, work authorization, and willingness to relocate. Write these down separately too, because you will compare them against what the bot actually submitted.
Set the daily limit low. Ten applications is plenty for day one. A tool that will not let you set a limit has already told you something about its priorities.
Step 2: Measure speed the right way
Raw speed (“applies to a job every 20 seconds”) is the least useful number in this category. What you want is useful applications per hour of your attention. Track these during the test:
- Time you spent on setup and fixing things.
- Number of applications actually submitted.
- Number of those that matched your title, seniority and location. Count honestly.
- Number of runs that stalled, looped or needed a restart.
A bot that sends 40 applications in an hour, of which 12 fit, is slower in practice than one that sends 15, of which 14 fit, because you will eventually spend time cleaning up the mismatches: withdrawing, explaining, or just being ignored by recruiters who saw a poor fit.
Step 3: Run the safety tests
Safety has two parts: your platform account and your professional reputation. Test both on purpose.
Account safety
- Does the tool pace itself, or does it fire applications back to back with no pause?
- Does it respect the limit you set, even across browser restarts?
- Does it ask for your password directly, or does it work through your already-logged-in browser session? Handing over a password to a third party is a bigger step than many people realise.
Reputation safety
- The duplicate test: manually apply to one job, then let the bot search the same title. Does it skip the job you already applied to, or apply again?
- The unknown-question test: find a job with an unusual screening question (a specific certification, a security clearance, a niche tool). Does the bot pause for you, or invent an answer?
- The external-site test: include a job that redirects to a company careers page. Does the bot record it as skipped so you can apply by hand, or quietly drop it?
These three tests reveal more about a tool’s design than any feature list.
Step 4: Audit the quality of what was sent
At the end of day two, open five submitted applications and read them like a recruiter would. On LinkedIn, the applied jobs list lets you see what you applied to. Check:
- Was the correct resume attached?
- Do the numeric answers (years of experience, salary, notice period) match the truthful values you wrote down?
- Were yes/no questions answered consistently across jobs? A bot that says you have five years of React on one form and two on another creates a problem you cannot fix later.
- Did any free-text answer contain filler, placeholder text, or claims you would not make in an interview?
If even one answer is something you would be embarrassed to defend, lower your trust in the tool and tighten its settings, or stop using it.
Step 5: Score it in one line
Give each area a score from 1 to 5 and write a one-sentence reason. For example: “Speed 4: steady pace, stalled once. Safety 5: skipped duplicate, paused on clearance question. Quality 3: used wrong resume on two jobs.” Any tool scoring below 3 on safety or quality should not get a bigger daily limit, no matter how fast it is. Speed is the easiest thing to buy and the least valuable once the other two slip.
Red flags that end the test early
- It submits an answer you never configured.
- It applies to the same job twice.
- There is no log, or the log does not match what you see on the job site.
- It promises to be “undetectable.” A tool built around hiding from the platform is telling you how it expects the platform to react.
- Support cannot explain what happens when a question has no saved answer.
How AutoApply CV handles these tests
Since you are reading this on our site, here is how our own extension behaves against the same tests, so you can check it rather than take our word for it. The AutoApply CV Chrome extension works on LinkedIn Easy Apply, Naukri and Indeed Easily apply jobs and runs inside your logged-in browser. It fills forms from screening answers you save once, skips jobs you have already applied to, pauses when a question needs a human answer, and records company-website jobs as skipped so you can apply manually. The free plan gives you 3 auto-apply actions a day, which is exactly the size of batch this test needs. You can read more about the flow on how it works.
If you want to compare evaluation results across several tools, our 12-feature checklist turns these tests into a pass/fail list, and what to measure weekly covers the numbers to keep watching after the trial.
Your next step
Pick the one bot you are considering, write your truthful answers in a note, and run the duplicate, unknown-question and external-site tests on five jobs today. By tomorrow you will know more about that tool than any review can tell you.
