Case study: how a Sfax manufacturing SME cut payroll processing time by 5× with Mawered

This case study documents the migration to Mawered of a Tunisian manufacturing SME (technical textiles, 68 employees, single site in Sfax). The figures presented are real; the company name has been anonymized at their request. The deployment was carried out between January and March 2026.

Company Profile

Sector	Manufacturing — technical textiles (personal protective equipment)
Headcount	68 employees: 42 production operators, 18 technicians and supervisors, 8 support functions
Work organization	2 alternating shifts (6am–2pm / 2pm–10pm), fixed scheduling, rotating leave
Location	Sfax industrial zone — 1 production site
Existing biometric equipment	3 ZKTeco F18 terminals (factory entrance, changing rooms, main workshop)
HR management before Mawered	Excel + CSV files manually exported from each terminal, external accountant for payroll

The Situation Before Mawered: 4 Concrete Problems

1. Monthly Payroll Took 4 Working Days

Each month, the payroll process started with a physical round of the 3 ZKTeco terminals to export CSV attendance files. These files were then manually consolidated into a master Excel file — with all the copy-paste error risks that entails. The external accountant then received this file, calculated the payslips, and sent them back for verification before issuance. The entire process occupied the administrative manager for 3 to 4 full days every month — roughly 40 working days per year spent solely on payroll.

2. CNSS Declarations Carried Regulatory Risk

The CNSS calculation base was manually reconstructed from payroll files. In March 2025, a transcription error led to a declaration that was 8% undervalued for that month, detected by the CNSS in a review six months later. The penalty applied (10% of the underdeclared amount) came to 1,200 TND. The incident also consumed two working days from the accountant and the owner to prepare the regularization file.

3. The 3 ZKTeco Terminals Operated in Isolation

Without a central connection, the 3 terminals did not synchronize with each other. An employee who badged at the factory entrance was not recognized by the changing room terminal if registered separately. Onboarding a new employee required manual entry on each of the 3 terminals — a task taking 20 to 30 minutes per hire and often forgotten for one terminal.

4. No Real-Time Visibility on Absences

The production manager had no way to know in real time if shifts were fully staffed. Unannounced absences were only reported at the end of the shift, when the team leader manually passed on the information. Reorganizing teams in case of absence remained chaotic and untracked.

The Mawered Deployment: 21 Working Days, Weeks 1 to 4

Weeks 1–2: Configuration and Data Import

The Mawered team started by importing the 68 employee records from the existing Excel file, verifying and cleaning the data (registration numbers, hire dates, base salaries). The Tunisian payroll engine was configured together with the company's accountant: CNSS employee/employer rates, the IRPP scale with applicable allowances, fixed bonuses (meal allowance for operators, seniority bonus). The alternating shift calendar was set up accounting for 2026 Tunisian public holidays.

Week 3: Connecting the 3 ZKTeco Terminals

ADMS configuration was completed in half a day across the 3 F18 terminals. Each terminal was assigned to the Sfax site in Mawered with its name and location. The registration number mapping check revealed 4 inconsistencies (registration numbers with leading zeros in one terminal but not in the Excel file) — corrected in 30 minutes. The first badge test confirmed real-time punch reception from all 3 terminals simultaneously.

Employee synchronization from Mawered to all 3 terminals was completed in 2 clicks from the interface: all 68 profiles were pushed to the 3 devices simultaneously, eliminating future manual entry on each terminal.

Week 4: First Parallel Payroll Run

The first payroll calculation was run in parallel with the existing method (Excel file + external accountant) for cross-validation. Out of 68 payslips, 7 differences were identified — including 5 in favor of employees (unfavorable rounding in the old method) and 2 data entry errors in the old file. All differences were verified and the Mawered calculation validated by the accountant. From the second month, payroll was issued directly from Mawered without a parallel run.

Measured Results at 90 Days

Indicator	Before Mawered	After Mawered (90 days)	Change
Monthly payroll processing time	3–4 days (24–32 hours)	5–6 hours	−80%
Payroll errors detected after issuance	2–3 per month	0 in 3 months	−100%
CNSS penalties (rolling 12 months)	1,200 TND (March 2025 incident)	0 TND	Risk eliminated
New employee onboarding time (terminals)	20–30 min (entry on 3 terminals)	2 min (central sync)	−93%
Unplanned absence alert time	End of shift (4–8 hours)	Real-time (<5 min)	Alert time −98%
Total cost (SaaS + deployment)	—	2,400 TND/year + 1,800 TND deployment	Positive ROI in 6 weeks

Challenges Encountered and How They Were Overcome

No migration is friction-free. Here are the three concrete obstacles encountered and the solutions applied — useful context before starting a similar deployment.

Obstacle 1: Inconsistent Registration Numbers Between Excel and Terminals

During employee mapping, 4 registration numbers stored with leading zeros on one terminal (e.g., "042") did not match entries without zeros in the Excel file ("42"). Mawered detected these inconsistencies automatically during the mapping verification. The correction took 30 minutes — but without this verification step, these 4 employees would have had unattributed punches for weeks. Lesson: systematically check registration number formatting before any ADMS migration.

Obstacle 2: One Team Leader's Resistance to Process Change

One of the three team leaders, accustomed to reporting attendance manually at end-of-shift via a shared Excel spreadsheet, initially resisted the change. He perceived the new system as increased surveillance rather than an efficiency tool. The solution: giving him read access to his team's dashboard in Mawered (real-time attendance, leave balances), which provided visibility he hadn't had before. Within two weeks, he had become the system's most enthusiastic advocate among operators.

Obstacle 3: First Month of e-CNSS Declaration From the HRIS

The first e-CNSS declaration automatically generated from Mawered required validation with the external accountant — who had their own data presentation habits. An intermediate export in their preferred format was configured to facilitate verification. From the second month, the declaration was issued directly from Mawered without external involvement, the accountant only intervening for special cases (exceptional bonuses, contract modifications).

What the Owner Took Away

"What surprised me most wasn't payroll — it was absences. Before, if an operator didn't show up in the morning, I'd find out an hour after the shift started, when the team leader called me. Now I get a notification on my phone within 5 minutes if someone hasn't badged. It completely changes how we organize replacements."

— CEO, manufacturing SME, Sfax (anonymized at request)

Beyond the numbers, the most notable change was freeing the administrative manager from 3 to 4 days of monthly work spent on manual payroll data consolidation. This time was reallocated to tracking HR indicators (turnover rate, overtime cost by shift, payroll trend) — analyses that had never been done before due to time constraints.

6 Months Later: What Changed Durably

At a follow-up review 6 months after deployment (July 2026), two additional developments were observed beyond the 90-day results:

Faster recruitment decisions: access to attendance and performance history by position enabled better justification of seasonal hiring needs to management. The decision process for 2026 summer reinforcement hires was reduced from 3 weeks to 10 days.
Detection of a recurring absenteeism pattern: the absence tracking module identified a significantly higher absenteeism rate on Monday mornings for the morning shift (6am–2pm). This data, invisible in the old system, led to a discussion with team leaders about causes and an adjustment to the allocation of heavy tasks at the start of the week.

These developments illustrate an often-underestimated reality of HRIS migrations: the most important benefits are not always those anticipated. Faster, error-free payroll was the primary use case — real-time visibility into attendance and behavioral patterns proved equally strategic.

What This Migration Teaches Other Manufacturing SMEs

Three transferable lessons for Tunisian industrial SMEs in a similar situation:

Don't start from scratch on existing ZKTeco terminals. This company already had 3 F18 devices in good condition. The ADMS migration preserved the existing hardware investment — no new terminals needed. If your terminals support the ADMS protocol (verifiable in the Network Settings menu), they are likely recoverable.
Run the first payroll in parallel. Week 4 allowed Mawered calculations to be validated with the accountant before cutting the old method. This step reassures all stakeholders and often reveals errors accumulated in the old system — as with the 7 calculation differences found in this case.
ROI is measured in weeks, not months. For this 68-employee SME, the annual software cost (2,400 TND) was recovered by the second payroll closing — in HR time savings alone. The elimination of CNSS penalty risk materialized from the first automated declaration.