The independent auditor was Daniel Kori, a man with the unsettling habit of reading a report without expression and asking his question only after everyone else had decided he had finished.
Amina gave him everything: spreadsheets, reconciliation notes, scanned receipts, bank correspondence, and a list of her assumptions.
“Do you want me to explain the methodology?” she asked.
“Later,” Daniel said. “I want to begin without your conclusion in my head.”
Amina closed her laptop and moved it aside.
Salim, who had come for the opening session, smiled. “Good. Finally someone neutral.”
Daniel did not answer him.
The first two days were harder than court. There were no speeches, no dramatic confrontations, no obvious villain across a room. There were receipts, bank dates, missing invoice references, and small variances that had to be tested.
Daniel confirmed early that, in several periods, the rent shown in the family ledger was lower than the totals supported by tenant receipts.
But he also rejected two of Amina's assumptions.
“Here,” he said, pointing to a schedule, “you calculated expected rent at the current contractual rate. These two months had a written concession.”
Amina checked the leases again. “You're right.”
Salim made a satisfied sound.
Daniel looked at him. “Her error does not validate your figures. It corrects hers.”
Amina changed the spreadsheet in front of them and logged the correction.
Later Daniel challenged her treatment of some tenant repair contributions.
“You assumed all contributions using the same description were for the same project.”
“The receipts use the same wording.”
“The dates alone do not prove allocation.”
Amina stared at the line. She had already used it as part of her working narrative.
“All right,” she said. “Move it from confirmed variance to unresolved.”
Daniel glanced at her. “No argument?”
“If I win the argument and the report becomes worse, we all lose.”
By evening, Daniel required every correction to be signed into a change log. Amina did it even when the mistake was small. The list grew several lines long, but the central problem did not disappear. Instead, weak assumptions were being stripped away from stronger findings.
Mariam looked at the correction log. “I thought a good audit was one without mistakes.”
“A good audit is one that finds and corrects its mistakes before other people rely on them.”
Daniel also chose his own samples instead of relying only on the months Amina had selected. He requested tenant receipts from different periods, invoices not mentioned in the complaint, and bank statements from months outside Amina's original review.
His corrections went beyond sampling. One payment Amina had grouped with unsupported company charges had a delivery note for genuine roofing materials. Another apparent rental gap came from the old ledger using the date money was received while the bank statement used the later deposit date, causing Amina to count one month twice. She removed both points from the confirmed irregularities. The remaining unexplained differences became smaller, but more defensible.
Daniel also refused to let anyone cherry-pick Hamza's email export. He wanted metadata, attachments, and the full conversation chain. One message supported Salim's instruction to relabel a management fee as an emergency repair; another showed a legitimate roofing job. Amina placed both in the chronology. ‘Why keep something that helps him?’ Hamza asked. ‘Because if I hide what helps him, this stops being an audit and becomes propaganda,’ she said. At the end of the day Daniel had everyone sign the document-receipt log. When Salim's lawyer objected to Hamza's USB, Amina answered by tightening the provenance record—source account, export date, device identifier, and checksum—not by arguing louder.
In one sample the family ledger was completely correct. Zulekha had paid exactly the amount recorded, and her repair contribution had been properly posted.
Daniel placed the entry on the table.
“This matters too.”
Amina nodded. “It shows the system was not falsifying every transaction.”
“And that makes the other irregularities specific rather than a slogan.”
Hamza heard the exchange. His father had repeatedly told him that Amina had decided everything was theft before she began. Now he watched the independent review preserve even the transactions that helped Salim.
On the second evening Hamza arrived alone carrying a small USB drive.
“I have things,” he said.
Amina did not reach for it. “Company material?”
“Emails. Invoices. An export from the folder I had access to.”
“How did you get it?”
“My company account. My password. I am a director, remember?”
Yusuf said, “Do not give us anything you have no right to possess.”
“This is my assigned account.”
Amina looked at Hamza. “Have you read the files?”
“Some.”
“Do you want to give your own statement?”
He swallowed. “Yes.”
They logged the source, device identifier, date, and scope before making a copy. Daniel calculated a hash for the copied data. The original USB stayed with Hamza.
Hamza exhaled. “Even when I'm helping, you're still auditing me.”
“That's how we protect you,” Amina said.
Daniel opened the email export only after the intake record was complete.
A pattern emerged slowly. One draft invoice said *management fee*. An email from Salim instructed someone to change it to *emergency roof repair* because, in his words, the family would question a management charge. Another set of documents showed *management fee* becoming *emergency repair*, then later *reimbursement*.
Hamza's face lost color.
“Dad said it was only accounting language.”
Amina did not say *I told you so*.
“Did you write those emails?”
“No.”
“Did you approve the payment?”
“I didn't know.”
“Then your statement should say what you know and what you do not know. Nothing more.”
Hamza put both hands over his face. “If I give this statement, I betray him.”
Amina sat beside him. “Truth does not belong to a father or a son.”
“Easy for you to say.”
“No. It isn't.”
She did not explain the argument that had driven her away seven years earlier. She did not need to.
Daniel asked for draft and final versions of the invoices before drawing any conclusion about altered descriptions. For two transactions they had both. For others they did not. He restricted his finding to the two supported examples.
He also requested an explanation from Kisiwa Heritage about its markup. Salim sent a memo saying the company had assumed procurement risk and cash-flow risk for contractors. Daniel asked why the related-party status had not been disclosed and why service descriptions changed. The follow-up memo did not answer those questions directly.
On the third day Daniel asked for two original invoices. Salim provided copies only. Daniel did not argue. He simply recorded the limitation in his report.
Amina saw the value of that. A good report did not hide what it could not prove. It stated what had been seen, what had not, and how that limited confidence.
Daniel also asked Hamza about benefits he had received through the company. Hamza admitted Kisiwa Heritage had paid for a laptop and one business trip.
“I didn't think it was a problem,” he said.
“Write that,” Amina told him. “Don't pretend you received nothing just because you did not understand the whole structure.”
His statement became harder on him—and more believable.
By the end of the third day, Daniel gathered them.
“I have a preliminary position.”
Salim had sent a lawyer rather than attend.
Daniel wrote two points on a sheet.
“First: there is substantial support for accounting irregularities and related-party payments. Second: not all of the debt is false.”
Mariam lifted her head.
Daniel put down another figure.
“There is real debt. A roof loan. Tax arrears. Legitimate repairs. But it is not the debt figure the family was given.”
Amina felt relief and weight at the same time.
Removing false charges did not make the house financially safe.
She opened a new page in her notebook.
The lie had been stripped away.
Now the real numbers still had to be paid.