Which of my numbers are solid and which are shaky?
You look at a dozen metrics every month. Revenue, conversions, ROAS, margins, pipeline value, churn. Some of those numbers come from your bank account. Others come from platforms that have every incentive to look good. Here's how to tell them apart.
The short answer
Numbers from your bank and accounting software are solid. Numbers from ad platforms are shaky. The reliability of a metric depends entirely on where it comes from. Revenue collected in Stripe is a fact. Conversions reported by Google Ads are an estimate. Understanding this distinction is the difference between making decisions on concrete data and making decisions on projections dressed up as facts.
Not all metrics are created equal, and treating them the same is dangerous
You are in a monthly review meeting. Someone says “revenue is up 12%.” Someone else says “Google Ads drove 340 conversions.” Both sound like facts. Both are delivered with the same confidence. But one comes from your accounting software (verified, reconciled, tied to bank deposits) and the other comes from Google's attribution model (estimated, inflated, not reconciled against anything).
When you treat all numbers as equally reliable, you end up making important decisions based on the weakest data in the room. You scale spend because Google says conversions are up. You cut a channel because Meta says ROAS dropped. But neither of those numbers was verified against money that actually entered your bank account.
The first step toward better decisions is knowing which numbers to trust completely, which to trust directionally, and which to verify before acting on them.
The three tiers of business metric reliability
- TIER 1: VERIFIED
Revenue collected, expenses paid, cash in bank. These numbers come from your accounting software (QuickBooks, Xero) and your payment processor (Stripe, Square). They are reconciled against actual bank transactions. When your P&L says you made $52,000, that maps to real deposits. This is the highest-confidence data you have.
- TIER 2: DIRECTIONAL
CRM pipeline, email open rates, deal stages. Your CRM (HubSpot, Salesforce) tracks deals and customer interactions. These numbers are real data points, but they rely on your team logging things accurately and on time. Pipeline value is a projection, not a fact. Email open rates are directional (Apple Mail privacy changes inflate them). Use these numbers for trends and patterns, but do not bet exact dollars on them.
- TIER 3: CLAIMED
Ad platform conversions, attributed revenue, platform ROAS. These numbers come from Google Ads, Meta Ads, TikTok, and similar platforms. They use attribution models designed to make the platform look good. They include view-through conversions, extended attribution windows, and no cross-platform deduplication. These numbers typically overclaim by 20-50%. Never use them raw for budget decisions without verifying against Tier 1 data.
How to audit your own metrics and grade their reliability (step by step)
- 1List every metric you use to make decisions
Open a spreadsheet. Write down every number you look at regularly: revenue, expenses, conversions, ROAS, CPA, pipeline value, churn rate, open rates, etc. Include the source system for each one.
- 2Classify each metric by its source tier
For each metric, ask: does this come from reconciled financial data (Tier 1), from a CRM or internal system (Tier 2), or from an ad platform (Tier 3)? Mark each one accordingly.
- 3Identify which Tier 3 metrics drive actual spending decisions
This is where the risk lives. If you are scaling or cutting ad spend based on platform-reported ROAS without cross- referencing your books, you are making decisions on the least reliable data you have.
- 4Cross-reference one Tier 3 metric per month against Tier 1
You do not have to audit everything at once. Pick the highest- stakes Tier 3 metric (usually Google Ads conversions or total platform ROAS) and compare it to actual revenue from QuickBooks or Xero. Note the gap percentage.
- 5Build a “confidence column” into your monthly reporting
Next to every metric in your monthly report, add a column: Verified, Directional, or Claimed. This simple label changes how people in the room interpret the number and what decisions they are willing to base on it.
How much time it takes to maintain a reliability-graded dashboard
The initial classification takes about an hour. After that, monthly maintenance is 15-20 minutes: pull one or two Tier 3 metrics and cross-reference against your books. The hardest part is changing the habit of treating all numbers as equally trustworthy.
Total time: 15-20 minutes per month after the initial setup. The real challenge is not time. It is discipline. Most people look at a number, see it presented confidently, and accept it without asking where it came from.
Or let Bottomline grade every metric for you automatically
Bottomline pulls data from your accounting software, payment processor, CRM, and ad platforms. It shows you every metric with a reliability indicator, so you always know whether a number is verified, directional, or platform-claimed.
No guessing about which numbers to trust. Every metric in your monthly report comes with a confidence grade based on where the data actually originates. Verified numbers stand on their own. Claimed numbers get flagged so you know to verify before acting. The result: better decisions, fewer surprises, and a clearer picture of what is really happening in your business.