Integration strategy is reshaping how we work. The average growing business uses somewhere between 30 and 80 software tools. Each was bought to solve a specific problem. Most do their individual jobs well. And almost none of them talk to each other in any meaningful way.
The result is a phenomenon every operations leader recognises: data lives in multiple places. Reports require manual consolidation. Customer records exist in three different shapes across three systems. Decisions get made on partial information because pulling the full picture would take half a day. The tools, individually impressive, collectively produce a tax on every action the business takes.
Integration strategy is the discipline of converting this collection of tools into something that actually behaves like a coherent system. It is rarely glamorous, almost never championed by a vendor and one of the highest-leverage technology investments a growing business can make.
What an integration strategy actually solves
A useful integration strategy delivers four things that an unintegrated tech stack does not:
1. A single source of truth for each data type. Customer data lives in one system, period, with other systems reading from it. Financial data lives in another, with the same discipline. There is no question about which version of a record is correct, because there is only one version.
2. Automated information flow between systems. When a deal closes in the CRM, the customer onboarding workflow triggers in the operations platform, the invoicing record creates in the accounting system and the welcome sequence starts in marketing automation – all without manual intervention.
3. Real-time visibility across the business. Reporting that draws from every relevant system, refreshed continuously, available to whoever needs it. Decision-making accelerates because the information cost of decisions falls.
4. Reduced operational risk. Manual data entry between systems is the source of most operational errors. Eliminating it eliminates a class of problems entirely.
each of these has direct commercial value. Together, they transform the operational efficiency of the business.
The four integration patterns
Most integrations fall into one of four architectural patterns. Knowing which pattern applies to which integration is half the design work.
Point-to-point. System A talks directly to System B through their APIs. Simple to set up, fast to deploy. The problem is scale: with N systems you can have N×(N-1)/2 integrations to maintain. Sustainable for small stacks; chaotic at scale.
Hub and spoke. All systems integrate through a central platform (often an iPaaS like Workato, Boomi or MuleSoft, or a custom integration layer). Each system has one integration to maintain – to the hub. Dramatically more sustainable at scale; requires investment in the hub itself.
Event-driven. Systems publish events (“a deal was won”, “an invoice was paid”) to a central event bus. Other systems subscribe to events they care about. Highly scalable, decoupled, robust. The most architecturally elegant pattern; requires more sophisticated engineering.
Data warehouse + reverse ETL. Operational data flows into a central warehouse (BigQuery, Snowflake, Redshift) where it can be analysed and combined. Reverse ETL tools then push enriched data back to operational systems. Excellent for analytical use cases and increasingly for operational ones too.
For most growing businesses, a pragmatic blend of hub-and-spoke (for simple integrations) and data warehouse (for analytics and complex transformations) is the right architecture. Pure event-driven is overkill until you reach a certain scale; pure point-to-point becomes unmanageable past a handful of systems.
The systems that matter most
Not all integrations are equally valuable. The high-leverage ones cluster around the systems that the rest of the business depends on:
Customer data. Whatever system holds the master customer record – usually the CRM – needs to talk cleanly to every other system that touches customers. Marketing, support, billing, product, operations.
Financial data. Accounting and ERP systems need to receive accurate data from the systems generating financial events: sales, billing, expense, procurement, payroll.
Identity. Single sign-on across the business’s tools is one of the highest-leverage integrations available. Better security, smoother user experience, easier offboarding when people leave.
Communication. Email, calendar, messaging, video – these systems generate enormous amounts of context that, when integrated with the operational systems, produces dramatic productivity gains.
Operational systems. Whatever tools the business actually runs on – fulfillment, project management, knowledge bases – should be integrated with the customer data layer so that operational status is visible from the customer view.
These five clusters, integrated well, produce most of the value of an integration programme. Other integrations add value, but at diminishing returns.
The “integration debt” problem
just as code accumulates technical debt, integration architecture accumulates its own form of debt – undocumented connections, brittle workarounds, integrations that haven’t been touched in years and that nobody fully understands. This debt grows silently and bites publicly, usually at the worst possible moment (a system upgrade, a vendor change, a security audit).
Healthy integration architecture is documented, monitored and periodically pruned. Connections that are no longer used should be removed. Connections that depend on deprecated APIs should be migrated. The integration layer should be treated as a first-class system in its own right, with the same operational discipline applied to it as to any other production system.
Most businesses do none of this. The first time anyone looks at the integration layer is when something has just broken.
The strategic decision: build, buy or hybrid
Three legitimate approaches to building an integration capability:
Buy an iPaaS platform. Workato, Boomi, MuleSoft, Tray.io and similar tools provide pre-built connectors to most major SaaS systems. Fast to deploy, lower engineering investment, ongoing platform fees. Best fit for businesses with mostly-mainstream SaaS systems and limited engineering capacity.
Build a custom integration layer. Bespoke code, typically in a serverless or microservices architecture, calling APIs directly. More flexibility, no platform fees, deeper integration with custom systems. Best fit for businesses with engineering capacity and integration requirements that off-the-shelf tools can’t meet cleanly.
Hybrid. Use an iPaaS for common, standardised integrations and custom code for the complex or strategically important ones. Pragmatic, scales well, recognises that not every integration is the same. This is what most mature businesses converge on.
The wrong approach is the absence of approach: ad hoc integrations made by whoever needed them at the time, with no shared infrastructure, no documentation, no monitoring. This is where most businesses currently sit and it is also where most of the operational tax is being paid.
The strategic point
integration is invisible when it works. When it doesn’t, it manifests as a thousand small frictions that the business mistakes for unrelated problems. Building a real integration strategy – and the architecture to support it – is one of the lowest-profile, highest-return technology investments available.
The businesses operating with genuinely connected tech stacks move faster, decide better and scale more cheaply than those with the same tools held together by manual work.
Tired of paying the integration tax in your business? We design and build integration architectures that turn collections of tools into connected systems. Get in touch.