Building a shared services platform for a SaaS holding company.

Every holding company eventually has to decide what it does for its portfolio brands and what it leaves alone. The decision sounds philosophical and is in fact intensely operational. A platform that does too little leaves real cost on the table and forces every brand to reinvent the same wheel. A platform that does too much suffocates the brands it is supposed to serve, slows decision-making to a crawl, and produces the worst kind of overhead — the kind that nobody asked for and nobody can remove.

We have spent a meaningful amount of time inside the Cobalt Glacier portfolio drawing and redrawing this line. The current version is the result of several rounds of subtraction. The principles below are the ones that survived contact with the operators.

The single test for any platform service

Before any function moves onto the platform, it has to clear one question. Would a portfolio operator, given full autonomy and a market budget, willingly buy this service from us at fair market price. If the answer is yes, the service belongs on the platform. If the answer is no, the service is corporate overhead in disguise, and the brand is correct to resent it.

The platform is a vendor the brands choose to use because we are the best price-quality option, not because we are the only one.

The test is harsh on purpose. It forces the platform team to compete on quality, not on mandate. It also forces the platform leadership to remove services that have quietly drifted into mediocrity. Functions that lose the willingness test for two consecutive quarters get unwound and pushed back to the brands.

The seven functions on the Cobalt Glacier platform

Seven functions currently sit on the platform. Each clears the willingness test in every brand. Each saves the brand both time and cost relative to running the function alone. None of them touch the product roadmap, the brand identity, the pricing, or the customer relationship.

1. Finance

Bookkeeping, monthly close, billing operations, revenue recognition, tax, treasury, and the annual audit all run on the platform. Each brand has a dedicated finance partner who acts as fractional CFO, sits in on the brand's operating reviews, and owns the relationship with the platform's shared back office. The brands publish monthly operating metrics on a single chart of accounts so that portfolio rollups are mechanical, not negotiated. This is also what makes the reporting in LP-grade reporting feasible.

2. People operations

Payroll, benefits, employment-law compliance, equity administration, and the people analytics dashboard are platform services. The brands own hiring decisions, performance management, and culture. The platform makes it cheaper to be a small company with a real benefits package and a real equity plan, neither of which is realistic for a twenty-person brand to build on its own.

3. Security and IT

Identity management, endpoint security, vendor security reviews, SOC 2 and ISO 27001 program management, and on-call security engineering all live on the platform. The platform also operates a shared internal incident-response team. A small B2B SaaS brand cannot afford a serious security program on its own; through the platform, every brand operates at an enterprise-grade posture from day one.

4. Data and analytics infrastructure

The shared data platform — warehouse, transformation layer, BI layer, and the metric definitions library — is centrally operated and locally consumed. Brands write their own dashboards and own their own analytical questions. The platform owns the pipes, the schema standards, and the cost optimization. The benefit shows up in two places: marginal-cost dashboards for the brands and a portfolio-level view of operating metrics for the holding company.

5. Design and brand systems

Each brand keeps its own visual identity, voice, and positioning. We do not consolidate brands and we do not impose a master template. What the platform does provide is a senior brand and design council that the brands draw on for periodic refreshes, marketing campaigns, and conference assets. The point is access to a level of design leadership that would be uneconomic for any single brand to retain full time.

6. Talent acquisition

Sourcing, recruiting operations, candidate experience, and the shared employment brand are platform services. Each brand owns the hiring decision and the loop design. The platform delivers candidates faster and at lower cost than any brand could on its own, and it produces a pool of cross-portfolio hires who can move between brands over the course of a long career.

7. Procurement

Vendor contracts above a threshold are negotiated by the platform procurement team. The brands set the requirements and own the vendor choice. The platform delivers volume discounts, MSA standardization, and a renewal calendar that is actually maintained. The arithmetic is straightforward — a portfolio with four brands spending against the same vendors gets meaningfully better pricing than any one brand alone.

What is deliberately not on the platform

The list of functions we have considered and rejected is at least as informative as the list we adopted.

• Product management and engineering. The product is the brand. Centralizing product would dilute the thing we acquired in the first place.
• Sales and customer success. The customer relationship is the brand's most valuable asset. The platform provides tooling and intelligence; it does not own the account.
• Marketing. Each brand owns its own positioning, campaigns, and content. The platform contributes brand-system review and senior design leadership but does not write the marketing.
• Pricing. Pricing is set by the brand operator with input from the platform finance partner. We never set portfolio-wide pricing rules.

How the platform is funded

Each platform service is funded by an internal charge to the brands that reflects, as closely as we can measure, the marginal cost of the service plus a small overhead for governance. The charge is published, predictable, and benchmarked against external alternatives at least once a year. Brands that want to opt out of a service can do so with one quarter of notice. In practice almost nobody opts out, because the willingness test is enforced. When somebody does opt out, that is a useful signal that the service needs work.

Why the platform compounds over a long holding period

On a single-deal time horizon, a shared services platform is a marginal efficiency. On a permanent-capital holding period, it is a compounding asset in its own right. Every dollar invested in the platform earns a return for as many years as the holding company owns the brands. Standards laid down once continue to pay out decade after decade. People hired into the platform develop portfolio-wide institutional knowledge that a single-brand operator could never accumulate.

The discipline we wrote about in The first 100 days is also where the platform earns its keep. The integration moves we make in the first hundred days after an acquisition are almost entirely platform moves: finance onto the chart of accounts, people operations onto the shared benefits program, security onto the shared posture, data into the shared warehouse. The brand, product, and roadmap are deliberately left alone. The platform is why we can make that promise honestly.

The bottom line

A shared services platform is not a corporate function and it is not an integration program. It is a vendor relationship between a holding company and its portfolio brands, governed by a single question — would the brand operator choose this on the open market. When the answer stays yes year after year, the platform compounds. When the answer drifts to no, the service comes off the platform. The discipline lives in the willingness to keep asking the question.

If you are an operator who has built a shared services function before and wants to embed with a portfolio brand or with the platform itself, our Operating Partner program is the door. If you are a founder weighing a permanent-capital home for the business you have built and want to understand what the platform would and would not do for it, start a conversation.