The 99.99% Truth: How Songfinch Finally Trusted Their Own Data

Songfinch was making spend decisions while their core systems disagreed on something basic: how many orders happened. Onward fixed the trust problem by reconciling ad-platform reporting to Stripe via a single source of truth, getting Songfinch to a 99%+ backend reconciliation rate and unlocking decision-grade contribution margin at the ad set level.

Key Facts

• Client: Songfinch
• Challenge: Three systems, three “answers,” zero confidence in what was real
• Onward work: Custom ETL + unified warehouse + standardized business logic
• Primary outcome: 99% backend reconciliation rate vs Stripe (Onward internal analysis)

Why it mattered: Leadership could finally make budget calls using verified data, not platform claims

The trust problem that stalls growth

Three numbers, one bank account

One platform says you sold 440 songs today. Another platform claims they drove 200 of those. Then you check Stripe and see 300 total.

So which one do you believe?

Before partnering with Onward, Songfinch was stuck in that exact loop. Leadership had plenty of data, but not enough truth to bet budgets on.

What “Reporting Theater” looks like in the real world

We call this “Reporting Theater.” The dashboards look strong, the spreadsheets say you are winning, and your bank balance quietly disagrees.

Part of the issue is structural. Ad platforms use their own attribution rules and models to decide how much credit an ad interaction gets for a conversion. That is useful for optimization inside each platform, but it can create conflicting “truths” across systems.

Privacy changes have also reduced signal over time, which makes clean reconciliation even more important. Apple’s App Tracking Transparency (ATT) requires apps to ask permission before tracking users across other companies’ apps and websites, and that impacts what platforms can reliably observe.

What was broken for Songfinch

The root cause was not “no dashboard”

Songfinch did not need prettier charts. They needed trustworthy definitions and a system that could reconcile reality across sources.

As Songfinch CEO John Williamson (Songfinch) put it: “Our data sources are a little messy… it’s sort of all intertwined.”

That mess shows up as duplicates, missing orders, mismatched currencies, inconsistent campaign naming, and returns not being handled consistently. Over time, it creates a decision tax: every budget call becomes a debate about whose spreadsheet is “right.”

{{video}}

The cost of not trusting your data

The damage is not just reporting frustration. It is margin erosion.

When you cannot reconcile performance to real revenue, you tend to pause campaigns that are actually profitable and scale the ones that only look good inside platform attribution. That is how brands “grow” while quietly losing money.

What Onward built

Engineering a single source of truth

At Onward, we do not guess, we engineer certainty. The job here was to get Songfinch out of Reporting Theater and into an automated truth.

We built a custom ETL (Extract, Transform, Load) pipeline that:

• Extracted raw ad spend from platforms like Meta and Google, plus transactional data from Stripe, plus behavioral clickstream signals.
• Transformed the data with strict business logic (format normalization, currency alignment, consistent order definitions, and returns handling).
• Loaded everything into a unified data warehouse, so finance and growth were looking at the same reconciled numbers.

This was not “connect a few APIs.” It was a measurement system designed to be auditable.

Why this approach is different than typical agency reporting

Most agencies live inside platform dashboards because that is the fastest path to a weekly report. It is also the fastest path to conflicting answers.

Onward treats measurement like infrastructure. If the source data is not reconciled back to Stripe (and cost logic is not standardized), then it is not decision-grade. That is how you earn the right to scale.

The results

99%+ reconciliation, and decisions stopped being debates

The impact was immediate once everything flowed through consistent logic.

• 99% backend reconciliation rate vs Stripe: Songfinch could finally trust that the pipeline matched the system of record.
• Fewer phantom orders and missing revenue: Leadership stopped wasting cycles arguing over totals.
• Ad set level contribution margin: Songfinch moved beyond vanity metrics and into profit-aware optimization at the level where media decisions actually get made.

As John Williamson (Songfinch) explained: “The dashboard is really helpful, because Onward and their team have made that the single source of truth.”

Once that trust exists, the questions change. You stop asking “Did Meta work?” and start asking “Which specific creative and audience combination produced the best contribution margin, and should get more budget next week?”

Where this can fail

Common failure modes we plan for

1. Undefined business logic: If “an order” means different things across teams (refund timing, cancellations, taxes, shipping, discounts), the warehouse will just centralize disagreement.
2. Garbage in, garbage out: Messy campaign naming, broken UTMs, and inconsistent product IDs will reduce the fidelity of any reconciliation effort.

No operating cadence: A single source of truth only matters if it becomes the default in weekly budget, creative, and forecasting decisions.

The takeaway

If you are scaling a business, you cannot afford to have marketing and finance debating whose spreadsheet is correct. You need a system that makes the data trustworthy enough to drive spend decisions.

Do you know which of your platforms is actually telling you the truth, or are you just choosing the number you like best?