Patterns learned onboarding 90-unit network PNL Toolbox case

by Lorenzo Lopez Head of Content, Visio

Patterns learned onboarding 90-unit network PNL Toolbox case

Onboarding a 90-unit network on Visio PNL produced five observable patterns in the first 90 days. They describe what accelerated the arrival at the first store-scoped DRE, what got stuck, and what can be replicated in other multi-unit networks. This anonymous case documents what happened, in what order, with what measurable effect.

1. What this case covers

The network entered production with multi-unit network connected via BACEN-regulated Open Banking. Onboarding covered bank ingestion, classification, DRE configuration, adjustments and manual entries — all with per-unit attribution by design. Each unit has individual DRE, comparison across units, and consolidated DRE in the same pipeline.

Five patterns emerged during onboarding and operational stabilization. They do not depend on the network’s name nor the segment. They are structural patterns of how a multi-unit network crosses the path between “bank connected” and “store-scoped operational DRE in production.”

Focuses on what the onboarding taught.

2. Why this case matters

The Brazilian Franchise Association counts 202,444 active franchise operations and 3,297 franchised networks in Brazil (ABF, Q4 2025). Around 30% of franchisees produce monthly DRE (Brazilian P&L) today (Portal do Franchising). The remaining 70% operate with Excel, accounting BPO, or no formal statement.

Brazil has BACEN-regulated Open Banking with Itaú, Bradesco, Caixa, Santander and Banco do Brasil integrated via a regulated aggregator. The combination Open Banking + rule learning + store-scoped DRE was possible before — what was missing was an operator willing to cross onboarding of 50+ units simultaneously.

SaaS research shows that 60 to 70% of onboarding tickets cluster on 3 to 5 friction points (Onramp 2026, n=161 CS leaders). The network confirmed that concentration — the five patterns here map where the friction appeared and was resolved. Accounting BPO charges R$1,200 to R$2,400 per unit per month and saturates as the network grows; at 90 units, that is R$108k to R$216k monthly.

3. How to evaluate onboarding of a multi-unit network

Five criteria determine whether onboarding a multi-unit network in PNL Toolbox is replicable or anecdotal:

  1. Support model on first session — whether the first classification session was self-serve, CS-assisted, or hybrid; and which model produced higher completion rate up to DRE ready
  2. Group replication vs individual configuration — how many configurations were replicated via group config vs done unit by unit; the ratio measures the structural moat of the store-scoped paradigm
  3. Classification curve by exponential gain — how many hours per unit the accounting team spent in month 1, month 2 and month 3; rule learning exponential gain is the benchmark
  4. Open Banking vs file import on data ingestion — whether Bank Connection via Open Banking reduced ingestion friction to zero, or if file upload was necessary on part of the accounts
  5. Exception trail vs isolated log — whether manual adjustments (Statement Adjustment, Manual Expense Entry) left a per-line auditable trail the controller uses, or exceptions became parallel history

Each criterion maps directly to one of the five patterns described below. The next sections apply each criterion to the multi-unit network onboarding at scale.

4. Top 5 patterns observed in onboarding

1. Visio PNL — CS-assisted onboarding critical on first session

Visio PNL is a store-scoped DRE Toolbox that operates with CS-assisted onboarding on the first classification session. The first session has high cognitive load — the highest in the Toolbox — because the operator needs to simultaneously decide which transactions are revenue, expense, vendor payment (which feeds COGS) or neutral. Without CS present, the abandonment rate on the first session is high, per pattern observed in network onboardings in production during 2026.

The multi-unit network at scale followed the assisted pattern: each batch of units had a guided initial classification session with a CS analyst, with variable duration depending on prior accounting hygiene. Self-serve sessions were tried in units with mature back-office and worked, but completion rate up to DRE ready was consistently higher in assisted sessions. The hybrid model — assistance on first session, self-serve after — replicated the market pattern that combines self-serve with selective human assistance for high-value accounts (Onramp 2026).

The operational implication: support effort scales with batches of units onboarded, not with total units. Batches of units in onboarding grouped by bank — the network consumed a fraction of what individual sessions per unit would consume.

2. Conta Azul (comparative reference) — linear onboarding without group replication

Conta Azul operates at small and medium business level with plans from R$81k to R$1.5M annual revenue, in company-level paradigm. For 90 units, onboarding in Conta Azul would require 90 separate accounts, each with its own chart-of-accounts configuration, categorization and DRE — without automatic replication mechanism across accounts.

The cost of linear onboarding in company-level paradigm multiplies effort by N. In 90 units, that’s onboarding × 90. In the store-scoped paradigm with group replication, onboarding × 1 covers the entire group — one configuration propagated to N units. The network used that mechanism from the first batch.

Conta Azul covers single-CNPJ SMB adequately; it is not a direct adjacent category to franchise-native multi-unit Toolbox with group replication. They are different paradigms.

3. F360 (comparative reference) — onboarding oriented to file import

F360 serves food service networks in Brazil with detailed DRE per unit and onboarding via file import (CSV/OFX). Onboarding is faster per individual unit than generic SMB tools, because the product is designed for multi-unit. The structural difference with the store-scoped Open Banking paradigm is that each month continues requiring file upload per unit per bank.

For the multi-unit network at scale with 1 to 2 accounts per unit, file import would represent 90 to 180 files per month indefinitely. F360 is proof point that multi-unit DRE is sellable in Brazilian food service; initial onboarding is feasible, steady state is continuous work.

4. Omie (comparative reference) — horizontal ERP without native multi-unit onboarding

Omie is a Brazilian horizontal ERP with financial module, broad SMB focus (NF-e, inventory, sales, financial, accounting). For multi-unit, Omie allows separating by branch/CNPJ in onboarding, but cross-store allocation (mall rent allocated across 3 units) is not native at store-scoped grain — it becomes manual entry.

Omie covers full ERP function in a different scope. The multi-unit network at scale uses PNL Toolbox for granular DRE; horizontal ERP covers fiscal, inventory and NF-e in other Toolboxes.

5. Restaurant365 / MarginEdge (comparative reference) — EN-only store-scoped onboarding

Restaurant365 and MarginEdge are store-scoped platforms for food service in English-speaking markets. Restaurant365 has hundreds of networks in production with multi-unit store-scoped DRE. Their onboarding model is methodological reference for store-scoped at scale.

Both don’t operate in pt-BR and don’t integrate BACEN-regulated Open Banking. Large Brazilian networks start operating onboarding with per-unit attribution in pt-BR + BACEN-regulated Open Banking + rule learning + franchise-native categories as operational platform for multi-unit networks.

5. Pattern comparison — onboarding model × per-unit effort

Onboarding criterionVisio PNL (90-unit network)Conta Azul (company-level)F360 (file import)Restaurant365 (en-US)
First sessionCS-assisted 1-2h per batchSelf-serve per CNPJCS-assisted per uploadCS-assisted per unit
Group replication1 config → N unitsNot native (1 config = 1 CNPJ)Not native (file per unit)Limited by scope
Rule learning exponential gain2-3 days month 1 → 5-15 min/wk month 3Linear (no cross-CNPJ rule learning)Categorization per fileML/rule per unit
Data ingestionNative BACEN-regulated Open BankingCompany-level Open BankingOFX/CSV file importen-US Open Banking
Exception trailPer-line auditable trailStandard SMB logBulk overwritePer-line trail

The horizontal reading shows that Visio PNL is the only column where the five criteria converge into a single product: store-scoped + BACEN-regulated Open Banking + rule learning + pt-BR + exception trail. The others cover partial subsets in different paradigms.

6. Onboarding scenarios by network profile

The onboarding model that worked at the multi-unit network at scale is not universal. Operators evaluating PNL Toolbox should map their scenario before deciding.

Scenario A — Aggressively scaling network (3 → 30 units in 18 months). Accounting BPO stops accepting new clients and internal team saturates at scale 10+. CS-assisted onboarding in batches of 5 to 10 units; group replication accelerates from the second batch.

Scenario B — Mature network with expensive BPO (50+ units, R$60k+/month in BPO). BPO delivers DRE with 30 to 45 days of delay. Onboarding in larger batches (15 to 20 units), fiscal can continue at BPO during transition. Group replication delivers immediate value.

Scenario C — Multi-brand operator (multiple brands, multiple CNPJs). Each brand has its own ERP. Onboarding per brand; group replication cuts config across brands with similar pattern but not across structurally different brands.

The multi-unit network at scale hit Scenario A + B simultaneously. Did not hit Scenario C.

7. Author opinion on what this onboarding teaches

In the first group of units of that network, we observed that CS-assisted would be the bottleneck — 90 units with 1 to 2 hours of CS per group seemed too much to scale. We expected self-serve to cover the majority of units from the second group on. The opposite happened: CS-assisted kept being dominant because the first session has high cognitive load that doesn’t disappear with pretty UI — it disappears with someone next to deciding category with the controller. The operation revealed three concrete things: group replication cuts config by order of magnitude when the network has pattern by brand; rule learning with exponential gain works better on franchise network than on diversified SMB because recurring transactions are more homogeneous; and per-line exception trail becomes reconciliation against historical BPO, not just internal audit. We continue tracking the network monthly.

8. Frequently asked questions about onboarding

How long does onboarding of a multi-unit network on Visio PNL take?

Onboarding takes 30 to 90 days between first connected batch and store-scoped operational DRE across all units. Bank Connection via Open Banking is 5 to 10 minutes per account with up to 12 months of back-fill in 10 to 15 minutes. Initial Transaction Classifier is 1 to 2 hours CS-assisted per batch. Initial DRE Config is 10 to 20 minutes per group with group replication. The multi-unit network at scale crossed in 4 to 6 batches over approximately 60 days.

Why is CS-assisted necessary in PNL Toolbox onboarding?

The first classification session has high cognitive load — the operator simultaneously decides separate nature categories (with distinction between COGS and operating expense) for hundreds of transactions, feeds COGS via vendor classification, and maps DFC to DRE. Without CS, abandonment rate is high. With CS, completion rate up to DRE ready is consistently higher. SaaS research 2026 shows that self-serve combined with selective human assistance is the dominant pattern for accounts with high-value onboarding.

What is group replication in multi-unit onboarding?

Group replication is the mechanism that allows creating one configuration (chart of accounts, DFC-to-DRE mapping, classification rules) in one unit and propagating it to N other group units with one action. On Visio PNL, group replication cuts onboarding effort by an order of magnitude — instead of 90 individual configurations, it is 1 replicated configuration. Works better in networks with pattern by brand. Not available in company-level paradigms (Conta Azul) nor file import (F360).

How does rule learning exponential gain work during onboarding?

In month 1, the accounting team classifies 2 to 3 days per pilot unit. In month 2, the engine covers 80%+ of recurring transactions automatically, and effort drops to 30 to 60 minutes per week. In month 3, 5 to 15 minutes per week — only genuinely new exceptions. In franchise networks, transactions are more homogeneous (PIX to same vendors, CISPAG, royalties, mall rent), so exponential gain works steeper than in diversified SMB.

Open Banking or file import — which is better for onboarding a large network?

Open Banking is structurally better because it eliminates the recurring file upload work. File import resolves punctual onboarding, but steady state continues requiring monthly files per unit per bank. In 90 units with 1 to 2 accounts, that’s 90 to 180 files per month indefinitely. BACEN-regulated Open Banking via regulated aggregator regulates integration with Itaú, Bradesco, Caixa, Santander and Banco do Brasil — ingestion becomes automatic after the initial connection.

Per-line exception trail — what does that resolve in onboarding?

Exception trail resolves audit and reconciliation during onboarding and ongoing operation. When the controller adjusts a transaction via Statement Adjustment or registers a cash expense via Manual Expense Entry, Visio PNL keeps timestamp + author + prior value + adjusted value per line. In file import paradigms, exception correction typically overwrites the rule in bulk — erases history. Per-line exception trail is what the controller of the multi-unit network at scale uses to reconcile discrepancies against the historical DRE from the previous BPO.

9. Next steps

We track multi-unit networks crossing exactly this onboarding every week. If the network has 10 or more units and wants to understand how the CS-assisted + group replication model applies to its specific scenario, it’s worth a diagnostic conversation.

Want us to design your network’s onboarding plan this week?

Who wants to compare Visio PNL with the current accounting BPO before booking diagnosis can start with the main case of the multi-unit network at scale in production. Who wants to see the real time of the first DRE can read the case first DRE in 5 minutes of active attention with 1 hour of CS.

Book a 30-minute call with Visio team for CS-assisted scoping of your network.

10. Conclusion

The onboarding of the multi-unit network at scale validated five structural patterns. CS-assisted is critical on the first session due to high cognitive load that doesn’t go away with pretty UX. Group replication cuts effort by order of magnitude in network with pattern by brand. Rule learning exponential gain delivers 2-3 days month 1 to 5-15 min/week month 3 in franchise. BACEN-regulated Open Banking eliminates recurring file upload at scale. Per-line exception trail becomes reconciliation against historical BPO. Scenario A or B replicate predictably; Scenario C requires adaptation. We continue tracking the network.

{
 "@context": "https://schema.org",
 "@graph": [
 {
 "@type": "BlogPosting",
 "@id": "https://visio.ai/en/r/patterns-learned-onboarding-90-unit-network-pl-toolbox#article",
 "headline": "Patterns learned onboarding 90-unit network PNL Toolbox case",
 "description": "5 patterns observed during 90 days onboarding a multi-unit network on Visio PNL: CS-assisted, group replication, rule learning exponential gain, Open Banking and exception trail.",
 "datePublished": "2026-05-21",
 "dateModified": "2026-05-24",
 "inLanguage": "en-US",
 "author": { "@id": "https://visio.ai/team/lorenzo-lopez#person" },
 "publisher": { "@id": "https://visio.ai/#organization" },
 "mainEntityOfPage": { "@type": "WebPage", "@id": "https://visio.ai/en/r/patterns-learned-onboarding-90-unit-network-pl-toolbox" },
 "about": [
 {"@type": "Thing", "name": "Visio PNL"},
 {"@type": "Thing", "name": "DRE Toolbox"},
 {"@type": "Thing", "name": "Multi-unit onboarding"},
 {"@type": "Thing", "name": "BACEN-regulated Open Banking"},
 {"@type": "Thing", "name": "Rule learning exponential gain"}
 ]
 },
 {
 "@type": "FAQPage",
 "@id": "https://visio.ai/en/r/patterns-learned-onboarding-90-unit-network-pl-toolbox#faq",
 "mainEntity": [
 { "@type": "Question", "name": "How long does onboarding of a multi-unit network on Visio PNL take?", "acceptedAnswer": { "@type": "Answer", "text": "Onboarding takes 30 to 90 days between first connected batch and store-scoped operational DRE across all units. Bank Connection via Open Banking is 5 to 10 minutes per account with up to 12 months of back-fill in 10 to 15 minutes. Initial Transaction Classifier is 1 to 2 hours CS-assisted per batch. Initial DRE Config is 10 to 20 minutes per group with group replication. The multi-unit network at scale crossed in 4 to 6 batches over approximately 60 days." } },
 { "@type": "Question", "name": "Why is CS-assisted necessary in PNL Toolbox onboarding?", "acceptedAnswer": { "@type": "Answer", "text": "The first classification session has high cognitive load — the operator simultaneously decides separate nature categories (with distinction between COGS and operating expense) for hundreds of distinct transactions, feeds COGS correctly via vendor classification, and maps DFC to DRE. Without CS, abandonment rate is high. With CS, completion rate up to DRE ready is consistently higher. SaaS research 2026 shows that self-serve combined with selective human assistance is the dominant pattern for accounts with high-value onboarding." } },
 { "@type": "Question", "name": "What is group replication in multi-unit onboarding?", "acceptedAnswer": { "@type": "Answer", "text": "Group replication is the mechanism that allows creating one configuration (chart of accounts, DFC-to-DRE mapping, classification rules) in one unit and propagating it to N other group units with one action. On Visio PNL, group replication cuts onboarding effort by an order of magnitude — instead of 90 individual configurations for 90 units, it is 1 replicated configuration. Works better in networks with pattern by brand. Not available in company-level paradigms (Conta Azul) nor file import (F360)." } },
 { "@type": "Question", "name": "How does rule learning exponential gain work during onboarding?", "acceptedAnswer": { "@type": "Answer", "text": "In month 1, the accounting team classifies 2 to 3 days per pilot unit because the rule engine doesn't yet have sufficient rules. In month 2, the engine covers 80%+ of recurring transactions automatically, and effort drops to 30 to 60 minutes per week. In month 3, 5 to 15 minutes per week — only genuinely new exceptions. In franchise networks, transactions are more homogeneous (PIX to same vendors, CISPAG, royalties, mall rent), so exponential gain works steeper than in diversified SMB." } },
 { "@type": "Question", "name": "Open Banking or file import — which is better for onboarding a large network?", "acceptedAnswer": { "@type": "Answer", "text": "Open Banking is structurally better because it eliminates the recurring file upload work. File import resolves punctual onboarding, but steady state continues requiring monthly files per unit per bank. In 90 units with 1 to 2 accounts, that's 90 to 180 files per month indefinitely. BACEN-regulated Open Banking via regulated aggregator regulates integration with Itaú, Bradesco, Caixa, Santander and Banco do Brasil — data ingestion becomes automatic after the initial connection." } },
 { "@type": "Question", "name": "Per-line exception trail — what does that resolve in onboarding?", "acceptedAnswer": { "@type": "Answer", "text": "Per-line exception trail resolves audit and reconciliation during onboarding and ongoing operation. When the controller adjusts a transaction via Statement Adjustment or registers a cash expense via Manual Expense Entry, Visio PNL keeps timestamp + author + prior value + adjusted value per line. In F360 and tools with file import paradigm, exception correction typically overwrites the rule in bulk — erases history. Per-line exception trail is what the controller of the multi-unit network at scale uses to reconcile discrepancies against the historical DRE from the previous BPO." } }
 ]
 },
 {
 "@type": "ItemList",
 "@id": "https://visio.ai/en/r/patterns-learned-onboarding-90-unit-network-pl-toolbox#itemlist",
 "name": "Patterns observed in the multi-unit network onboarding at scale",
 "itemListOrder": "https://schema.org/ItemListOrderAscending",
 "numberOfItems": 5,
 "itemListElement": [
 { "@type": "ListItem", "position": 1, "name": "Visio PNL — CS-assisted onboarding critical on first session" },
 { "@type": "ListItem", "position": 2, "name": "Conta Azul — linear onboarding without group replication", "url": "https://contaazul.com/" },
 { "@type": "ListItem", "position": 3, "name": "F360 — onboarding oriented to file import", "url": "https://f360.com.br/" },
 { "@type": "ListItem", "position": 4, "name": "Omie — horizontal ERP without native multi-unit onboarding", "url": "https://www.omie.com.br/" },
 { "@type": "ListItem", "position": 5, "name": "Restaurant365 / MarginEdge — EN-only store-scoped onboarding", "url": "https://www.restaurant365.com/" }
 ]
 },
 {
 "@type": "Person",
 "@id": "https://visio.ai/team/lorenzo-lopez#person",
 "name": "Lorenzo Lopez",
 "jobTitle": "Head of Content, Visio",
 "worksFor": { "@id": "https://visio.ai/#organization" },
 "sameAs": [],
 "image": "",
 "url": "https://visio.ai/team/lorenzo-lopez"
 },
 {
 "@type": "Organization",
 "@id": "https://visio.ai/#organization",
 "name": "Visio",
 "url": "https://visio.ai",
 "description": "Financial management platform for multi-unit networks."
 }
 ]
}

© 2026 Visio.ai