Proposal

Scalable Reporting System
for Bloom Digital

Reporting infrastructure that grows with your agency — from 13 clients to 30+.

Prepared by Nick Valiotti · Valiotti Analytics
March 2026

Where You Are Right Now

From our conversation, the picture is pretty clear. You've got 13 clients, a growing team, and reporting that's still duct-taped together with screenshots and Google Sheets. You tried Whatagraph — it didn't stick. Supermetrics before that. Same story.

The issue isn't that these tools are bad. It's that none of them let you blend data across channels. So when a client asks "is our retention actually improving, or are we just spending more on ads?" — nobody has a quick answer. Your team ends up pulling numbers from four different platforms, manually piecing together a narrative in a slide deck, and hoping the math adds up.

That's not sustainable at 13 clients. It's definitely not going to work at 25.

What I'd Build for You

A reporting system that your team actually uses — organized around business questions, not platform tabs. Built on infrastructure you own, so you're never locked into another vendor that disappears or breaks.

One template. One data model. Duplicate across every client.

The Stack

Layer Technology Why
Data Warehouse BigQuery Your data lives in your project. No lock-in. If we part ways, everything stays with you.
Data Ingestion Fivetran Pre-built connectors for Shopify, Meta, Google Ads, Klaviyo. When APIs change (and they do), Fivetran handles it — not your team.
Presentation Looker Studio Free. Client-shareable. Your branding. Pulls directly from modeled BigQuery views.
Quick note on BQ Data Transfer Service vs. Fivetran. Your brief mentions BQ DTS, and it does work for Google Ads. But the Shopify and Klaviyo connectors launched in early 2026 and are still rough around the edges. Fivetran has been running these connectors in production for years across thousands of ecommerce companies. When you need reports that are correct on Monday morning, that matters. I'd say start with Fivetran, and we can always migrate individual sources to BQ DTS down the road once they stabilize. Looks like a cost decision — but it's really a reliability one.

The Report

Seven pages. Organized around the questions your clients actually ask — not around which ad platform the data came from. Each page has KPI scorecards, trend charts, and space for your team to add strategic commentary before the monthly review.

# Page Key Question It Answers
1 Executive Summary What moved, what didn't, and where should we look?
2 Business Outcomes Is the growth real, or are we just spending more?
3 Acquisition Where are new customers actually coming from, and at what cost?
4 Retention & Lifecycle Are email and SMS pulling their weight, or are we over-relying on paid?
5 CRO & On-Site We're driving traffic — is the site actually converting it?
6 Opportunities & Threats What should we do next month, specifically?
7 Appendix The nitty-gritty: campaign-level and flow-level breakdowns for whoever manages the channels.

The Data Model

Under the hood, six BigQuery views power everything. Same structure for every client — daily granularity, consistent naming, no surprises. When you onboard a new client, you duplicate the views, point Fivetran at their accounts, and you're live.

View Grain Key Fields
business_daily Day revenue, orders, AOV, new vs returning customer revenue
acquisition_daily Day × channel spend, CAC/NCAC, sessions, new customers, channel mix
retention_daily Day repeat rate, returning revenue %, email rev, SMS rev
lifecycle_program_daily Day × program campaign vs flow performance, placed-order rate
cro_daily Day × page LP CVR, PDP→cart rate, checkout completion
reporting_targets Month revenue goal, MER goal, NCAC goal, retention goal

Each client gets their own dataset — no cross-contamination, no risk of sharing the wrong numbers. Parameterized data sources with unique shareable links mean you can drop a report link into a client Slack and they'll only ever see their own data.


Investment

Option A — Template + Pilot
$8,500
3–4 weeks · 1 client
  • Fivetran setup: Shopify, Google Ads, Meta Ads, Klaviyo
  • Emotive SMS via CSV→BigQuery pipeline
  • 6 BigQuery modeled views
  • 7-page Looker Studio master template
  • Client data isolation architecture
  • Field dictionary & metric definitions
  • Duplication SOP for your team
  • 2-hour training session (recorded)
  • 14 days post-launch support
You get a working system for one client and a playbook to roll out the rest yourself.
How per-client pricing works. Option B includes up to 5 standard data sources per client (Shopify, Google Ads, Meta, Klaviyo, Emotive). If a client runs additional channels — Bing, TikTok, Pinterest, etc. — each extra connector is $500. This way you only pay for the complexity that's actually there, and the scope stays clear for both of us.

What You'll Pay Monthly (After the Build)

Item Monthly Cost
Fivetran (13 clients × 5 sources) $300–500
BigQuery storage & queries $50–100
Looker Studio Free
Total $350–600/mo

For context — probably less than what Whatagraph was costing you. Except now you actually own everything.

Optional: Ongoing Support Retainer

After the post-launch period, if you want me to stay involved — handle new client onboarding, add connectors, troubleshoot issues — I offer a $2,500/mo retainer. No commitment, cancel anytime. Most agencies find it useful for the first 2–3 months until the team is fully self-sufficient.

The Math

If this saves your team 10+ hours/week on manual reporting and helps you keep even one extra client per year — it's paid for itself before the first invoice is due.
Where the value comes from Conservative numbers
Time your team gets back (no more manual report assembly) $2,000–3,000/mo
Keeping one client who would've churned over bad reporting $36,000–84,000/yr
Capacity to take on 2–3 more clients without hiring $72,000–200,000/yr
On the budget question. I know you mentioned the Whatagraph refund as a starting point. I'd rather be upfront: $2,500 gets you a dashboard, not a system. And you've already been through two rounds of "let's try this cheaper tool" — each time costing more in team hours than the tool itself. Option B works out to roughly $1,350 per client for a reporting system you'll use for years. This proposal is priced to solve the problem once.

Why Me

I run Valiotti Analytics and currently work as Fractional Head of Data for a $6M ecommerce marketplace — same stack, same scale, same kind of cross-platform reporting challenges you're dealing with.

A few things that are directly relevant:

You're not hiring a freelancer who'll figure it out as they go. This is literally what I do every week.

Next Steps

  1. Share this with Louis, take your time
  2. We hop on a 30-minute call to go through any questions
  3. Pick Option A or B
  4. Contract through Upwork — I can start within 48 hours of signing

The stack and approach are firm — that's what makes this actually work long-term. Timeline and payment milestones, happy to adjust.

Looking forward to it, Danielle.