How to Automate Employment Certificate Requests Step by Step

Why this costs more than it looks

A certificate looks like a one-page job. In practice it is a chain of small jobs. The request arrives by email, a chat message or a phone call. Someone finds the employee file, checks the dates and job title, types the letter, gets a sign-off and sends it. For former staff, one more step comes first: confirming that the person asking is the person named on the file.

The cost is staff time. Multiply the number of requests each month by the minutes each one takes, then by the loaded hourly cost of the person doing the work. That result is your own figure, and it is the number to take into a budget discussion. For context, a 2015 survey of 1,030 US business leaders run by BambooHR found that 60% of HR professionals spent more time on administrative and maintenance work than on strategic work. That survey is old and covers a wider job than certificates, so use it as background rather than as a benchmark for your team.

Delay is the second cost. An employee who needs a letter for a rental, a visa or a loan will ask again if the first request goes quiet, and a former employee who chases a letter often escalates. Count how many requests are chased in a typical month. That count is usually the clearest sign that the process is broken.

Manual way and automated way

The request-to-letter flow, step by step

Build the flow in the order below. Test each step with a set of past requests before you switch it on for real ones.

1. Decide which letters you issue and what each one contains. Most businesses need two: an experience certificate, covering dates, title and duties, and a salary certificate, which adds pay figures. Write the fields for each on one page. The UK regulator's guidance on data minimisation says to hold only what is adequate, relevant and needed for the stated purpose, so leave out any field the reader does not need.
2. Set up one intake point. A form built in Google Forms or Microsoft Forms works, and so does a WhatsApp Business number connected through the Cloud API. Meta has ended support for the older On-Premises API, which was sunset on 23 October 2025, so a new build should use the Cloud API. Each submission should create one row with a request number, date, letter type and staff ID.
3. Verify the requester. Current staff can be matched through their login. Former staff should be matched against details already on file, such as a personal email address or exit date. If the match is unclear, HR confirms by writing to the address or number on file before any document goes out. Some public employers accept automated requests only from employees themselves, which is a useful control to copy for former staff.
4. Pull the fields from the record. Use a read-only export or API from your HR or payroll system, or a spreadsheet that HR keeps current. Do not let the automation write back into payroll. Pull only the fields you defined in step one.
5. Fill the template. Keep one master letter in Google Docs with placeholder tags such as double-brace names. The Google Drive API can make a copy of the template for each request, and the Google Docs API can replace each tag in a single batch update. Microsoft 365 users can get the same result with Word mail merge. A language model can draft a short cover note from the same fixed fields, but it should not write dates, job titles or pay figures on its own.
6. Send the draft to HR for approval. The draft lands in a queue, which can be a status column in a spreadsheet or a workflow in your HR tool. HR checks the draft against the record, approves or rejects it, and records the reason for any rejection. Salary lines get a separate tick, and the employee's written consent should be on file before pay figures leave the company.
7. Deliver and log. After approval, the letter goes by email from the HR mailbox as a PDF. If WhatsApp is the channel, a business-initiated message outside the 24-hour customer service window must use an approved template. Meta assigns each template a category and reviews it, so submit templates early. Then write the log entry: time, requester, approver, template version and file link.
8. Handle older paper files. If past certificates or records exist only on paper, OCR software can turn scanned pages into searchable text so that the fields can be found. Check every value that OCR extracts against the scan before it reaches a letter, because OCR can misread digits and dates.

Tools that can carry each step

• Intake: Google Forms, Microsoft Forms, or a WhatsApp Business number on the Cloud API
• Records: your HRIS or payroll export, with a maintained spreadsheet as the fallback
• Drafting: a Google Docs or Word template, with placeholder tags and a batch replace
• Approval: a status column in a shared sheet, or a workflow in your HR system
• Sending: the HR mailbox for email, and approved WhatsApp templates for chat-based requests
• Language models: only for cover-note wording built from fixed fields, never for dates or pay

What breaks, and how to prevent it

Most failures start in the data, not in the software. Plan for the failures below before you go live.

Stale records are the most common. If the job title in the HR system is out of date, the letter will be wrong, and automation will produce that wrong letter quickly. Match the export to the most recent payroll run, and keep the HR approval step so that a person reads every draft against the record.

Blank or mismatched placeholders come next. A tag that does not match its field name will print as text or as an empty gap. Test with a handful of past records, and add a rule that no draft can be sent while a tag is still unfilled.

Wrong recipients are a real risk for former staff, because an old email address may now belong to someone else. Send any letter that carries pay only to an address that has been verified, and have HR read the address out loud before the send step, if your process allows it.

Salary released without consent is the biggest risk, and it is not technical. Keep a consent record for every salary certificate and a separate approval for pay lines. If you hold personal data of people in the European Union, the GDPR's Article 5 sets out the principles of purpose limitation and data minimisation, and local rules on salary disclosure differ by country, so check them with a lawyer before you set the template.

Automation prepares and delivers the letter. It does not make the judgment calls. Disputes, unusual wording, and requests from courts or banks that ask for a specific form still need a person to read the request and decide.

When not to automate

If you issue only a handful of letters a year, the setup can take longer than the time it saves. A shared template and a short checklist are enough at that volume.

If the records are messy and nobody owns them, automation will copy the mess faster. Fix the HR record first, then build the flow.

If the letter needs judgment, such as a reference after a dismissal or a note that explains an employment gap, keep a person writing it. A template should cover routine certificates only.

If a third-party verification service already handles requests for your company, check whether it covers your needs before you build anything. Some large employers send verification requests to outside providers, and that may be all the process you need.

If a current or former employee is in a dispute with the company, stop the automated flow and route the request to HR or a lawyer.

What it takes

The effort is mostly decisions and testing, not coding. Expect one session to agree on the letter types and fields, one session with HR to map the record, a template build, and a test run against past requests. A business with a clean HR record and one template can finish in a few working sessions. A business with data spread across spreadsheets and paper files needs longer, because the data has to be cleaned before it can be read reliably.

Name one owner. That person updates the template when the company name, signatory or policy changes, and reviews the log each month. Without an owner, the flow tends to slide back to email.

This explains the rule in general terms; it is not legal or tax advice for your case.