SaaS/ISV Rapid Integration of Embedded BI: Lessons from 200+ Partner Implementations

Introduction

When SaaS vendors’ customers increasingly ask “can your system directly show data analysis,” a key decision arises: build BI features in-house or integrate a third-party BI platform?

The temptation to build in-house is great—full control, deep customization, no external dependencies. But hands-on experience repeatedly proves: the typical cycle for in-house BI features is 12-18 months, with R&D investment at the hundreds of yuan level, and continuous maintenance and iteration investment required after launch. More critically, BI is a cross-domain product—involving data connection, modeling, calculation, visualization, interaction, permissions, multi-tenancy, and more than ten other professional domains; any weakness in any domain leads to significant product experience defects.

Hengshi Technology, based on 200+ ISV partner implementation experiences, summarizes the best practice path for embedded BI integration. This article provides a practical guide for SaaS/ISV vendors to “launch professional data analysis capabilities within 2 weeks” from four dimensions: integration architecture, implementation steps, common challenges, and success stories.

---

I. Core Decision for ISV BI Integration: In-House vs Embedded

1.1 Hidden Costs of In-House BI

The explicit costs of in-house BI are easy to estimate—R&D manpower, time cycles, infrastructure investment. But hidden costs are often underestimated:

Cost Dimension	Hidden Cost Description	Quantified Impact
Domain span	BI involves 10+ professional domains, each requiring specialized talent	High recruitment costs, long team formation cycles
Feature completeness	Core features (data connection, visualization) are easy to implement, but enterprise features (multi-tenancy, permissions, auditing) are easily overlooked	Low post-launch customer satisfaction, large demand gaps
Iteration pressure	BI products need continuous iteration to keep up with data engine, visualization technology, and AI capability evolution	R&D resources continuously occupied; core business innovation blocked
Talent dependency	BI talent is scarce; if key personnel leave, product iteration rhythm may be interrupted	High product stagnation risk

1.2 Core Advantages of Embedded BI

Choosing embedded BI is not “giving up autonomy” but “choosing expertise”:

• 2-week launch: Complete process from data connection to visual display can be finished in 2 weeks
• Complete features: All BI capabilities—data connection, modeling, calculation, visualization, interaction, permissions, multi-tenancy—obtained at once
• Continuous evolution: Hengshi releases 2-3 version iterations annually; ISVs obtain the latest BI capabilities without R&D investment
• OEM-friendly: BI capabilities delivered under the ISV’s own brand; customers cannot perceive the existence of a third-party platform

1.3 Decision Matrix

Evaluation Dimension	In-House BI	Embedded BI (Hengshi)
Launch cycle	12-18 months	2 weeks
R&D investment	Hundreds of yuan	Integration cost (1-2 people / 2 weeks)
Feature completeness	Gradually supplemented	Complete (obtained at once)
Multi-tenancy support	Need to build in-house	Natively supported
AI/ChatBI	Need to develop in-house	Natively supported
Continuous iteration	Bear by yourself	Platform auto-updates
Brand ownership	Own brand	OEM white-label

---

II. Three-Layer Embedding Architecture Technical Implementation

2.1 ISV Integration Typical Architecture

Hengshi’s embedded BI ISV integration architecture consists of three layers:

┌─────────────────────────────────────────┐
│  ISV Business System                     │
│  ┌───────────┐  ┌───────────┐           │
│  │ CRM Func  │  │ ERP Func  │           │
│  └───────────┘  └───────────┘           │
│  ┌─────────────────────────────────────┐│
│  │  Embedded BI Analysis Module         ││
│  │  (HENGSHI SENSE)                    ││
│  │  ┌─────────┐ ┌─────────┐            ││
│  │  │Dashboard│ │ChatBI   │            ││
│  │  └─────────┘ └─────────┘            ││
│  └─────────────────────────────────────┘│
│  ┌─────────────────────────────────────┐│
│  │  SSO/Permissions/Tenant Isolation   ││
│  └─────────────────────────────────────┘│
└─────────────────────────────────────────┘

2.2 L1-L3 Integration Mode Details

L1: URL Embedding (Lightest)

Applicable scenarios: ISV needs to display fixed data dashboards on specific pages of the business system.

Technical implementation:

• Configure iframe in business system page, embed HENGSHI SENSE dashboard URL
• Control dashboard display style (theme, layout, hide navigation bar, etc.) through URL parameters
• Implement single sign-on through SSO protocol

Implementation cost: 0.5-1 person-day

L2: API Deep Integration (Most Common)

Applicable scenarios: ISV needs to deliver BI capabilities as part of its own product, requiring full control over interface style and interaction logic.

Technical implementation:

• Call HENGSHI SENSE capability layer through RESTful API, obtain dataset, metric, and chart configurations
• Render data analysis functions in ISV’s own interface framework, using H5 components or custom rendering logic
• Implement automated integration of identity authentication, permission mapping, and tenant isolation through SDK

Implementation cost: 5-10 person-days

L3: Full-Process Embedded ChatBI (Deepest)

Applicable scenarios: ISV needs to embed data analysis capabilities into instant messaging and workflows, achieving a new paradigm of “business as analysis.”

Technical implementation:

• Integrate ChatBI Agent into work groups supported by ISV (WeChat Work, Feishu, DingTalk)
• Configure Agent analysis scenarios through API (metric range, dimensional relationships, permission boundaries)
• Configure automatic push of scheduled analysis tasks (such as daily business briefs, anomaly alerts)

Implementation cost: 8-15 person-days

2.3 ISV Integration Mode Selection Decision Matrix

ISV Requirement Characteristics	Recommended Integration Mode	Key Considerations
Only need to display fixed dashboards	L1	Fast launch, low cost
Need to deliver BI features under own brand	L2	OEM white-label, complete interface integration
Need to integrate analysis into workflow	L3	ChatBot integration, instant delivery
Multi-tenant SaaS scenario	L2+Multi-tenancy	Tenant isolation, permission mapping
Large enterprise group	L2+L3	Deep integration + ChatBI embedding

---

III. Key Steps for 2-Week Launch Detailed

3.1 Days 1-3: Data Connection and Dataset Creation

Key tasks:

1. Identify ISV business system’s core data sources (typically 1-3: main database, data warehouse, API data source)
2. Create data connection in HENGSHI SENSE, configure connection parameters and authentication methods
3. Create core datasets, define field mappings and data types
4. Verify data connection availability and dataset completeness

Common issues:

• Data source type mismatch: Hengshi supports 30+ data source types (MySQL, PostgreSQL, Oracle, SQL Server, Apache Doris, Greenplum, MongoDB, Excel files, custom APIs, etc.), covering the vast majority of ISV scenarios
• Large data volume causing slow queries: Hengshi has a built-in high-performance engine that can pre-compute aggregate results for commonly used dimensional combinations, significantly reducing query response time

3.2 Days 4-6: Metric Modeling and Dashboard Creation

Key tasks:

1. Sort out ISV core business metric list (typically 5-10 core metrics)
2. Define atomic metrics and business metrics in HENGSHI SENSE, configure HQL calculation expressions
3. Create core dashboards (typically 2-3: business overview, business details, KPI dashboard)
4. Configure dashboard filters, parameters, and interaction logic

Key recommendations:

• Metric modeling is the foundation of the entire integration. Invest sufficient time ensuring metric standard accuracy and dimensional relationship completeness—this is the only way subsequent dashboard creation and ChatBI configuration can proceed smoothly
• Dashboard design should be guided by ISV business scenarios, not by “displaying more data.” Each dashboard should serve a clear business decision scenario

3.3 Days 7-9: Embedding Integration and SSO Integration

Key tasks:

1. Complete embedding configuration according to selected integration mode (L1/L2/L3)
2. Configure SSO single sign-on, ensuring ISV users can use BI features without additional login
3. Complete permission mapping, ensuring ISV’s role permission system automatically synchronizes with HENGSHI SENSE’s permission system

Two SSO integration methods:

• Standard protocol integration: OAuth 2.0, SAML 2.0, and other mainstream SSO protocols, suitable for ISVs using standard authentication frameworks
• API integration: Implement custom authentication flow through HENGSHI SENSE’s authentication API, suitable for ISVs using non-standard authentication methods

3.4 Days 10-12: Multi-Tenant Configuration and Permission Testing

Key tasks (applicable only to SaaS scenarios):

1. Configure multi-tenant isolation strategy, ensuring each tenant’s data space, metric definitions, and dashboard configurations are independently managed
2. Configure tenant-level permission system, ensuring tenant administrators can independently manage their tenant’s user permissions
3. Execute comprehensive permission testing, verifying the effectiveness of data isolation, functional isolation, and permission isolation

Key parameters for multi-tenant configuration:

Configuration Item	Description	Recommended Value
Data isolation mode	Independent data space for each tenant	Logical isolation (shared engine, independent schema)
Computing resource quota	Query concurrency limit per tenant	Configure based on tenant scale (typically 5-10 concurrency)
Storage space quota	Data storage upper limit per tenant	Configure based on business needs
Permission management mode	Tenant admin self-management	ISV admin → tenant admin → tenant users

3.5 Days 13-14: Launch Acceptance and Optimization Adjustment

Key tasks:

1. Execute performance stress testing, verifying response time and stability under high-concurrency scenarios
2. Execute user experience verification, ensuring embedded BI features fully integrate with the business system’s interface style and interaction logic
3. Complete user training documentation and operation guides
4. Official launch

---

IV. Five Key Findings from 200+ Partner Experiences

Finding 1: 90% of ISVs Choose L2 Integration Mode

Among Hengshi’s 200+ ISV partners, approximately 90% chose L2 deep integration mode. The reason: L1 URL embedding, while simple, makes it difficult to fully integrate interface styles; L3 ChatBI embedding, while offering the best experience, requires deep integration with instant messaging tools and has higher initial launch costs. L2 is the “best cost-performance” choice—it can deliver under the ISV’s own brand while maintaining full control over interface experience.

Finding 2: Metric Modeling Quality Determines Integration Success or Failure

Repeated proof from Hengshi partners’ hands-on experience: the key bottleneck for 2-week launch is not technical integration but metric modeling. If the ISV’s metric standards are unclear and dimensional relationships incomplete, subsequent dashboard creation and ChatBI configuration will encounter “wrong data, inaccurate results” problems. It is recommended that ISVs spend 1-2 weeks sorting out core metric lists before integration to ensure standard consistency.

Finding 3: Multi-Tenant Isolation is the First Priority for SaaS Scenarios

For SaaS vendors, multi-tenant data isolation is a mandatory verification item before embedded BI launch. Hengshi’s hands-on experience shows that the focus of multi-tenant isolation verification is not “whether isolation is possible” but “whether performance is acceptable after isolation”—ensuring each tenant’s response time does not exceed 3 seconds when 100+ tenants query simultaneously.

Finding 4: Post-Launch Activity Determines Renewal Decisions

Hengshi partner data shows: after embedded BI features launch, if activity in the first 30 days is below 20%, renewal risk significantly increases. It is recommended that ISVs focus on promoting BI feature usage in the first month after launch—through training, guidance, and incentives—to ensure at least 30% of active users use BI features at least 3 times within 30 days.

Finding 5: ChatBI is the Most Effective Tool for Activity Improvement

Among ISV partners that launched ChatBI functionality, BI feature activity increased by an average of 40%. The reason: ChatBI has the lowest usage barrier (just ask in natural language), the most natural scenario (directly ask for data in work groups), and the most instant feedback (results returned in 2-30 seconds). It is recommended that ISVs plan L3 ChatBI embedding soon after L2 integration launches.

---

V. Typical Case Analysis

5.1 Fenxiang Xiaofei: CRM SaaS Embedded BI Practice

Fenxiang Xiaofei is a leading CRM SaaS vendor in China, serving hundreds of thousands of enterprise customers. Against the backdrop of customers universally requiring CRM systems to have data analysis capabilities, Fenxiang Xiaofei chose Hengshi Technology’s embedded BI solution.

Integration solution: L2 deep integration + multi-tenant architecture

Key decisions:

• Choose embedded over in-house: Estimated in-house BI cycle is 12 months; Hengshi integration launches in 2 weeks
• Choose L2 over L1: Need to deliver BI features under the company’s own brand; L1 iframe embedding cannot meet brand integration requirements
• Launch core scenarios first, then expand: Phase 1 covers “sales trend analysis on customer detail pages” and “management-level business briefs,” with subsequent gradual expansion

Implementation results:

• CRM data analysis feature activity increased by 40%
• Customer renewal rate increased by 15%
• Saved an estimated 12-month in-house development cycle and hundreds of yuan in R&D investment

5.2 Zhiyuan Hulian: Collaborative Office Embedded BI Practice

Zhiyuan Hulian is a leading collaborative office software vendor in China, with its A8+ product serving large enterprise groups’ office collaboration needs.

Integration solution: L2 deep integration + data big screens

Key decisions:

• Collaborative office scenarios need a “report portal” rather than an “independent analysis tool”: Hengshi’s data application concept supports encapsulating a group of dashboards to form a report portal, naturally aligned with the collaborative office “portalization” concept
• Data big screens used for management-level reporting scenarios: Hengshi supports multi-screen adaptive visualization engine, meeting Zhiyuan customers’ data big screen needs

Implementation results:

• Data analysis features in collaborative office products became an important differentiator in customer selection
• Customer satisfaction in data big screen scenarios significantly higher than traditional reporting methods

5.3 Mingdao Cloud: Zero-Code Platform Embedded BI Practice

Mingdao Cloud is a leading zero-code application platform in China, where users build various business applications through zero-code methods.

Integration solution: L2 deep integration + zero-code analysis templates

Key decisions:

• Zero-code platform BI integration must also be “zero-code”: Hengshi’s application template design supports ISV preset analysis templates; users selecting a template automatically generates a complete data analysis experience
• Zero-configuration data connection: Mingdao Cloud user data is stored in Mingdao’s self-built database; Hengshi automatically retrieves data through standard APIs, requiring no user data connection configuration

Implementation results:

• Mingdao Cloud users can create complete data dashboards in 5 minutes—zero-code experience fully aligned with platform positioning
• BI features became a differentiated competitive advantage for Mingdao Cloud

---

VI. Future Evolution: From Embedded BI to Embedded Agentic BI

6.1 Three Evolution Stages of ISV Integration

Stage	Core Deliverable	Key Characteristics
Stage 1: Embedded Dashboards	Dashboards, charts	Static display, pre-defined analysis scenarios
Stage 2: Embedded Analytics	Interactive analysis, drill-down, linkage	Dynamic exploration, user self-service analysis
Stage 3: Embedded Agentic BI	ChatBI, AI Agent	Natural language interaction, proactive service, closed-loop actions

Currently among Hengshi’s 200+ partners, approximately 60% are in Stage 1, 30% in Stage 2, and 10% have begun entering Stage 3. With the opening of Hengshi Sense 6.0’s Agentic BI capabilities, it is expected that Stage 3 proportion will rapidly increase to 30% within the next year.

6.2 ISV Action Recommendations

• Stage 1 ISVs: Plan Stage 2 upgrade path as soon as possible—the transition from static dashboards to interactive analytics is the key turning point for activity improvement
• Stage 2 ISVs: Evaluate ChatBI embedding feasibility—L3 integration mode is the most effective means of activity improvement
• Stage 3 ISVs: Follow Hengshi’s Agentic BI scenario templates—upgrade ChatBI from “general Q&A” to “industry-specific analytics assistant” to further enhance user stickiness

---

Conclusion

200+ partner experiences prove: the core value of embedded BI is not “saving money on in-house development” but “getting data analysis capabilities to customers faster.” When customers directly see sales trends in the CRM system, view business briefs in the collaborative office platform, and create complete data dashboards in 5 minutes on a zero-code platform, data analysis transforms from an “extra tool” into “part of the business.”

Hengshi Technology’s three-layer embedding architecture and 2-week launch implementation path provide SaaS/ISV vendors with a clear path from the “in-house development dilemma” to “rapid integration.” At the end of this path lies embedded Agentic BI—data analysis is no longer a technology to be learned, but a capability to be awakened at any time.