The 90-Day Data Foundation: What Actually Happens When You Hire a Fractional CDO

Most companies that reach out to us say the same thing: “We know we need help with data, but we honestly don’t know what the engagement actually looks like.”

Fair enough. “Fractional CDO” is still a relatively new concept. When you hire a fractional CFO, everyone understands what that means. But “fractional chief data officer”? That could mean anything from building dashboards to implementing a data lake to teaching your team SQL.

So let me remove the mystery. Here’s exactly what happens — week by week, deliverable by deliverable — when we do a 90-day data foundation engagement. I’m going to be specific, because specificity builds trust and vagueness breeds suspicion. If someone tells you they’ll “transform your data capabilities” without telling you what that means in practice, run.

This is the playbook I’ve used across $5-30M ARR companies in SaaS, marketplaces, and e-commerce. The details vary, but the structure stays remarkably consistent.

Before We Start: The Discovery Call

Before any engagement begins, we have a 30-minute discovery call. This isn’t a sales pitch. It’s a diagnostic conversation where I ask questions like:

• How many data sources do you have, and how many of them do you actually trust?
• When your CEO and CFO quote a revenue number, is it the same number?
• How many people on your team can answer a data question without asking someone else?
• What happened the last time you hired for a data role?

About half the companies I talk to don’t need a fractional CDO. Some need a senior analyst. Some need a data engineer. Some just need someone to properly configure their existing tools. I’ll tell you that in the discovery call and point you in the right direction. I’d rather lose a deal than start an engagement that’s wrong for both of us.

For the other half — companies where the problem is strategic, not just tactical — here’s what the 90 days look like.

Weeks 1-2: The Diagnostic Sprint

Goal: Understand exactly where you are, what’s broken, and what matters most.

This is the part most consultants stretch into 6-8 weeks of “discovery.” I do it in two. Not because I’m rushing, but because after doing this 30+ times, I know exactly what to look for and which questions to ask.

What happens:

Stakeholder interviews (6-10 conversations, 45 minutes each). I talk to every key decision-maker: CEO, VP of Product, VP of Marketing, Head of Sales, Head of Finance, and whoever is currently doing data work (analyst, engineer, or the VP who ended up as the accidental “data person”). Each conversation follows the same framework:

• What decisions do you make weekly/monthly?
• What data do you use to make those decisions?
• Where does that data come from, and do you trust it?
• What questions can’t you answer today that you wish you could?
• What’s the most frustrating thing about data at this company?

Current state audit. I get access to everything: your data warehouse (if you have one), your dashboards, your analytics tools, your spreadsheets, your Slack channels where people argue about numbers. I’m looking at:

• Tool inventory — What do you have? (Typically: 3-7 SaaS tools feeding data nowhere, a half-built Looker/Metabase instance, and 15 critical Google Sheets)
• Data flow mapping — Where does data originate? Where does it go? Where does it get lost or corrupted?
• Dashboard audit — How many dashboards exist? How many are actually used? How many show conflicting numbers?
• Metric definitions — Is “revenue” defined the same way everywhere? What about “active user”? “Churn”? (Spoiler: at 90% of companies, it’s not)

Quick wins identification. In every engagement, there are things that can be fixed in days, not months. A dashboard filter that’s misconfigured. A metric that’s wrong because someone fat-fingered a SQL join six months ago. A report that three people maintain manually that could be automated in an afternoon. I find and fix these during the diagnostic — it builds credibility and shows immediate value.

What you get:

A 12-15 page diagnostic report that includes:

1. Executive summary — The three biggest data problems at your company, ranked by business impact. Not 47 recommendations. Three.
2. Current state map — Visual diagram of your data ecosystem: sources, pipelines, storage, outputs. Where it works, where it breaks.
3. Metric discrepancy log — Every metric I found that has conflicting definitions across teams, with the specific numbers and sources.
4. Tool assessment — What you have, what you’re actually using, what you’re paying for that you shouldn’t be, and what’s missing.
5. Quick wins completed — What I already fixed during the sprint, with before/after.
6. Gap analysis — The delta between where you are and where you need to be, mapped to business outcomes.

Real example: At a $12M marketplace, the diagnostic found that “revenue” was calculated differently in Stripe, their internal dashboard, and the board deck. The delta? $340K per year. Three teams had been arguing about “whose numbers are right” for nine months. The fix took four hours once we identified it.

Weeks 3-4: Foundation Design

Goal: Design the target state and build a prioritized roadmap to get there.

This is the strategic architecture phase. Based on what we found in the diagnostic, we design what the data foundation should look like — not in some ideal future state, but in a practical “buildable in the next 8 weeks” sense.

What happens:

Metric definitions workshop (2-3 hours, all key stakeholders in the room). This is usually the most valuable single meeting of the entire engagement. We go metric by metric through every KPI the company tracks and get the CEO, CFO, VP Product, and VP Marketing to agree — in writing — on exactly one definition for each.

This sounds trivial. It is not. I have facilitated metric workshops where the debate about how to define “monthly active user” lasted 45 minutes. But when you come out the other side with a signed-off data dictionary, you’ve eliminated the single biggest source of data distrust in the company.

Deliverable: A data dictionary with 40-60 metric definitions, including: metric name, business definition (in plain English), technical definition (SQL logic or calculation), data source, update frequency, owner, and known limitations.

Data model design. Based on the diagnostic, I design the data model that will power your analytics. This isn’t a generic template — it’s built specifically for your business. For a marketplace, that means modeling supply, demand, matching, and transaction dynamics. For a SaaS company, it means modeling the customer lifecycle: acquisition, activation, engagement, retention, expansion, churn.

Deliverable: An entity-relationship diagram and dimensional model documented in a format your team can maintain and extend after I leave.

Tool stack recommendations. Most companies don’t need new tools — they need to use the tools they have correctly. But sometimes a change is warranted. I’ll recommend specific tools with specific reasons, pricing, and implementation timelines. No vendor kickbacks, no partnerships. I recommend what works.

Typical stack for a $10-20M company: BigQuery or Snowflake for the warehouse, dbt for transformation, Looker/Metabase/Hex for BI (depends on team technical level), Fivetran or Airbyte for ingestion. A data catalog is usually overkill at this stage — the data dictionary in Notion or Confluence does the job.

90-day roadmap. A week-by-week plan for what gets built, in what order, and why. Priorities ranked by business impact vs. implementation effort. High impact / low effort goes first. Always.

Deliverable: A prioritized 90-day roadmap in a format your team can track (usually a Linear or Jira board with clear milestones).

Weeks 5-8: Build and Ship

Goal: Build the core data infrastructure and the dashboards that matter most.

This is where most of the tangible work happens. The diagnostic told us what’s broken. The design told us what to build. Now we build it.

What happens:

Data warehouse setup (or cleanup). If you don’t have a warehouse, we set one up. If you do — and it’s a mess of 200 tables with names like `revenue_v3_FINAL_nick_copy` — we clean it up.

Specifically:

• Set up a proper data modeling layer (staging, intermediate, mart layers using dbt)
• Connect all critical data sources (typically 5-8 sources: billing system, product database, marketing platforms, CRM, support tool)
• Build tested, documented data transformations
• Set up automated quality checks (row counts, freshness, schema changes, null rates)

Core dashboards (3-5 that actually matter). Not 20 dashboards that nobody looks at. Three to five that the leadership team will open every single day. These come directly from the stakeholder interviews in Week 1 — we already know what questions need answering.

Typical set:

1. CEO/Board Dashboard — Revenue, growth rate, unit economics, runway. The five numbers the CEO needs before any meeting.
2. Revenue & Sales Dashboard — Pipeline, conversion rates, revenue by segment, expansion/contraction. What the CRO/VP Sales needs weekly.
3. Product Health Dashboard — Activation, engagement, retention, feature adoption. What the VP Product checks on Monday morning.
4. Marketing Performance Dashboard — CAC by channel, attribution, campaign ROI, content performance. Updated daily.
5. Operational Dashboard — Whatever’s specific to your business. For a marketplace: supply/demand balance, match rates, fulfillment times. For SaaS: support ticket volume, NPS trends, infrastructure costs.

Each dashboard is mobile-friendly, filter-ready, includes comparison to prior periods and targets, and has a “how to read this” guide any new hire can follow.

Self-serve access. Before I leave Week 8, every VP can log in, navigate to their dashboard, and answer their top questions without asking anyone.

Documentation. Everything gets documented as it’s built — data dictionary connected to actual tables, dashboard guides, pipeline documentation, and a runbook for common issues. Concise and practical, not a 200-page binder.

What you get by end of Week 8:

• A clean, well-modeled data warehouse with automated pipelines
• 3-5 production dashboards used daily by leadership
• A self-serve analytics layer that handles 80%+ of recurring questions
• Full documentation that doesn’t require me to maintain

Real example: At a $8M B2C SaaS, we went from “the analyst manually updates a Google Sheet every Monday with numbers from five different tools” to “automated dashboards that refresh hourly, accessible to all VPs, with a self-serve exploration layer.” The analyst’s ad-hoc request volume dropped from 35/week to 8/week in the first month after launch.

Weeks 9-12: Handoff and Scale

Goal: Make sure everything we built survives after I leave. Transfer knowledge, train the team, set up ongoing operations.

This is the phase most consultants skip entirely. They build something, hand over a Confluence page, and disappear. Three months later, the dashboards are broken and nobody knows how to fix them. I’ve built my entire practice around making sure that doesn’t happen.

What happens:

Team training (structured, not informal). Not “let me walk you through the dashboard real quick.” Three tracks:

• Analyst training (8-10 hours): Data model, dbt project, dashboard maintenance, how to add metrics, how to debug pipeline failures. They should be able to maintain everything without me.
• Leadership training (2-3 hours): Dashboard usage, request queue process, metric interpretation. Focus on self-sufficiency.
• Data culture session (1 hour, all-hands): How to think about data decisions, when to trust dashboards vs. when to ask for deeper analysis.

Knowledge transfer package. Architecture diagrams, decision log, finalized data dictionary, maintenance calendar, vendor contacts, known issues and workarounds. Everything organized so your team never has to guess.

Hiring plan (if needed). Most companies at $10-20M need one dedicated data person long-term. I’ll recommend the exact role (analyst, analytics engineer, or data engineer — they’re different), write the job description, provide interview questions, and give compensation benchmarks. The wrong hire wastes 6 months; the right specification prevents that.

Ongoing advisory setup. For most clients, I transition to a lighter-touch advisory role after the 90 days: 4-8 hours per month, available for strategic questions, quarterly reviews, and “is this normal?” sanity checks. This is optional — some companies don’t need it, and I’ll tell you if you don’t.

What you get by end of Week 12:

• A fully trained team that can operate independently
• A comprehensive knowledge transfer package
• A hiring roadmap for your next 12 months of data investment
• (Optional) An ongoing advisory relationship for strategic guidance

What You Don’t Get

Transparency means being clear about what this engagement is not:

You don’t get a 200-page strategy deck. You get a 12-15 page diagnostic, a prioritized roadmap, and working infrastructure. Strategy is embedded in the decisions, not in a PowerPoint.

You don’t get 6 months of “discovery.” Two weeks. If someone needs 6 months to understand your data problems, they either lack the experience to recognize patterns, or they’re padding the engagement.

You don’t get a permanent dependency on me. The engagement is designed around handoff. If I build something only I can maintain, I’ve failed.

You don’t get a Fortune 500 data platform. We’re building a foundation right-sized for $5-30M ARR. Designed to scale, but not overbuilt. Overbuilding is as dangerous as underbuilding.

You don’t get ML or “advanced analytics.” Not because they’re not valuable — but you need the foundation first. Predictive models on messy data is like a high-performance engine in a car with no wheels.

The Numbers: What This Actually Costs and Returns

Let me be direct about economics.

A typical 90-day engagement costs $36,000-$60,000 depending on company complexity, number of data sources, and how much infrastructure needs to be built. That’s $12,000-$20,000/month for senior data leadership — roughly what you’d pay a mid-level data analyst in salary alone, except you’re getting a decade of cross-company pattern recognition instead of a single person learning on the job.

What you get in return:

• Analyst productivity: 80% ad-hoc drops to 40% = $50,000-$65,000/year in recovered strategic value
• Decision speed: Questions that took 3-5 days now take 30 seconds via self-serve. Saves leadership 15-25 hours/week
• Metric trust: When CEO, CFO, and VP Product quote the same revenue number, board meetings get shorter and decisions happen faster
• Turnover prevention: Analysts who do strategic work stay. Fixing the ad-hoc trap saves $68,000-$105,000 per turnover event
• Hidden findings: Every engagement surfaces at least one significant financial insight during the diagnostic. Range: $50K-$340K in annual impact

Conservative ROI for a $15M ARR company: $150,000-$250,000 in year-one value against a $36,000-$60,000 investment. A 3-5x return that compounds after the engagement ends.

Is This Right for You?

This works best for companies at $5-30M ARR, past product-market fit, with 1-3 people doing data work (or nobody dedicated), experiencing data chaos — multiple sources of truth, metric arguments, analyst burnout.

It’s not the right fit if you need a full-time CDO, if your primary need is a specific technical build (ML model, real-time pipeline), if you’re pre-PMF, or if you want someone to just “build dashboards” without addressing strategy.

One Last Thing

I started doing this work because I kept seeing the same $10M companies make the same $500K mistakes with data. They’d hire an analyst who’d drown. They’d buy a BI tool nobody used. They’d build dashboards on top of broken pipelines. They’d argue about numbers in board meetings. And every time, the fix was the same: stop treating data as a tool problem and start treating it as a leadership problem.

That’s what a fractional CDO actually does. Not just build dashboards. Build the foundation that makes every other data investment work.

—

*If this sounds like where your company is right now, let’s have a conversation. Book a 30-minute discovery call — I’ll tell you honestly whether a 90-day engagement fits your situation or if there’s a simpler path.*