Paystack vs Flutterwave: Settlement & Reconciliation for Lagos Merchants (2026)

The verdict in three sentences

Running two processors multiplies leak points: each gateway has its own settlement window, its own export format and its own fee schedule. Without automatic matching by order reference, a merchant leaves 3 to 8 % of revenue stranded in unreconciled flows. The fix: a normalised CSV pipeline pulls the target gap below 0.5 % and divides reconciliation time by eight.

Why multi-processor reconciliation is hard

The challenge is not collecting money, it is tying each payout back to the right order. Paystack settles T+1, Flutterwave settles on its own international calendar, and both deduct fees before the money lands. When a customer pays 25,000 NGN for order #1042, the net amount arrives reduced by fees and sometimes bundled with other payments into a single bank credit.

The three most common sources of divergence:

• Fees deducted at source: the credited amount never equals the invoiced amount.
• Aggregated settlement: several payments merged into one payout.
• Missing references: the customer pays without an order number.

Comparing the two gateways (2026 order of magnitude)

These are 2026 estimates to confirm against your merchant agreement. On a 8,000,000 NGN monthly turnover, a 0.1-point fee difference is roughly 8,000 NGN a month, small; the 0.5 % rolling reserve is the bigger cash-flow item.

The reconciliation pipeline that works

The goal: one table where each order line carries a verified payment status. Operational target:

The technical key: enforce a normalised order reference (e.g. KOL-1042) injected into the payment metadata, then auto-match the gateway export against the orders table. Anything that fails to match drops into an exceptions queue handled manually.

Mini case study

Emeka runs a skincare store in Lagos. He takes 220 orders/month, average basket 18,000 NGN, so 3,960,000 NGN in revenue. Split: 55 % Paystack, 45 % Flutterwave.

Before: he reconciled by hand, let ~5 % of revenue go unidentified, i.e. 198,000 NGN/month of murky flow, including about 90,000 NGN of real unpaid orders never chased for lack of traceability. It cost him 6 h/week.

After the pipeline: gap down to 0.4 %, i.e. ~16,000 NGN residual noise. He recovers ~74,000 NGN/month of previously lost payments and 5 h/week. The system pays for itself in month one.

FAQ

Can matching really be automated if customers forget the reference?

Yes, by combining reference + amount + timestamp + phone number. Even without a reference, 85 to 90 % of payments reconcile automatically; the rest go to an exceptions queue.

Do I need the Paystack API or is the CSV export enough?

The CSV export is enough to start and covers 95 % of the need. The API adds real time and removes the daily manual download, worth it beyond ~300 orders/month.

What does a 5 % reconciliation gap cost?

On 8,000,000 NGN/month, 5 % is 400,000 NGN of uncontrolled flow, part of it real unpaid orders never recovered. Pushing the gap below 0.5 % directly protects cash flow.

Does reconciliation handle refunds and chargebacks?

Yes, a good pipeline treats negative flows (refunds, chargebacks) as signed lines so the net balance ties to the books without double counting.

How long does setup take?

Allow 2 to 4 weeks for a single-store merchant, including reference normalisation and testing on one month of real history.

Let's talk about your project. We build your Paystack and Flutterwave reconciliation pipeline with a target gap below 0.5 %. WhatsApp +221 77 596 93 33.