How we score
Every number in the Data Health Check either comes from a published source linked below, or is labeled as our own judgment with the reasoning. If we can't back a dollar figure, we don't show one.
The six areas and their weights
Your score is a weighted average of six areas. The weights are our weighting, not an industry standard. They follow a "foundations first" idea: data has to be recorded and trustworthy before reports or automation can do much with it. That ordering is similar to Monica Rogati's widely cited AI Hierarchy of Needs (collection, then data flow, then cleaning, then analytics), though the percentages below are ours.
| Area | Weight | Why |
|---|---|---|
| Data Capture | 20% | Foundation. Nothing downstream works if it isn't recorded. |
| Data Quality / Trust | 20% | Foundation. Reports built on data people don't trust don't get used. |
| Centralization | 15% | Recorded data is only usable if you can get to it in one place. |
| Visibility / Reporting | 15% | Reports turn stored data into something people actually look at. |
| Decision Usage | 15% | The payoff: numbers shaping real pricing, spend, and staffing calls. |
| Automation | 15% | Saves time, but it speeds up whatever the foundation already is. |
We'll refine these weights as more businesses complete the check and as we compare scores with what we find in paid Diagnostics.
How the Health Score is calculated
Each question has four answers worth 0 to 3 points. An area's score is the points you earned divided by the points possible, times 100, so every area runs 0 to 100 no matter how many questions it has. Your overall score is the weighted average of the six area scores.
Health Score = Σ (area weight × area score)
The industry question counts inside one area: Visibility for home services, restaurants, and healthcare; Decision Usage for fitness, professional services, and nonprofits; Data Quality for retail; Data Capture for everyone else. On that question, "Don't know" scores 0 (not measuring it is the gap), the weakest range scores 1, the middle 2, and the strongest 3. The hours question has a fifth answer, "Not sure," which scores 1.
Tiers: 0–39 Getting Started · 40–59 Building · 60–79 Established · 80–100 Advanced.
How the dollar ranges are calculated
Dollar figures are ranges. The low end uses the low end of each answer band and the high end uses the high end. We round to the nearest $500 below $10,000 and the nearest $1,000 above. Your "biggest opportunities" put dollar items worth more than $1,000 a year first, then fill in with the areas where a gap weighs most on your score (weight × gap).
1. Manual data work (all businesses)
$31.50 is the median hourly wage for general office clerks, $21.64 (BLS, May 2025) [1], divided by 0.687, because wages are 68.7% of total employer cost for office and administrative support roles (BLS, June 2026) [2]. We used the clerk median, not a higher-paid role, to stay conservative. 48 working weeks is our assumption. The hours question excludes time spent fixing errors, and we don't claim all of this time can be automated.
2. Shrink from process errors (retail)
27% is the share of retail shrink that comes from process and control failures and errors rather than theft [6]. If you don't track shrink, we use the 1.6% retail average from the same survey and say so.
3. No-shows (healthcare-adjacent)
A meta-analysis found patients who got appointment reminders were 25% less likely to no-show, with a confidence interval down to 18% [10]. We use 18% for the low end. We only price this when you give us your own no-show rate. If you don't track it, we show the published average [9] as context but no dollar figure.
Your headline total adds items 1 and 2 or 1 and 3. If that total starts below $2,000 a year, we point to your biggest opportunity instead of a dollar figure.
What isn't priced, and why
Data Capture, Centralization, Visibility, and Decision Usage have no dollar figure. There's no reliable published figure to turn them into dollars for a small business.
Time lost to bad data isn't priced. The widely repeated "workers spend 27% of their time fixing bad data" figure traces back to a claim about sales reps and contact data, not small-business staff, and we couldn't find a verifiable source that fits.
Missed calls, member churn, restaurant labor, billable utilization, and donor retention show your rate and an industry benchmark but no dollars. For each, there's no credible published figure for how much of the gap a business can realistically win back.
Revenue under $500K or "Prefer not to say" means no revenue-based dollar figures. We'd rather work from your own numbers.
Assumptions for open-ended answers
- "10+ hours" is read as 10 to 15 hours. "Not sure" is read as 5 to 10 and marked as an estimate.
- "$10M+" revenue is read as $10M, to keep estimates conservative.
- "Over X%" is read as X to X+10 points, except retail shrink "Over 2%," which is read as 2% to 3%.
- "Under X%" is read as 0 to X. A range that starts at zero isn't priced.
Sources
- U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: General Office Clerks, median pay May 2025. bls.gov
- U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation, Table 4, June 2026. bls.gov
- Invoca, See How Much Missed Sales Calls Cost Home Services Businesses, 2024. invoca.com
- Health & Fitness Association, 2025 Fitness Industry Benchmarking Report, 2025. healthandfitness.org
- National Restaurant Association, Restaurant labor costs are well above historical averages (2025 Restaurant Operations Data Abstract), 2025. restaurant.org
- National Retail Federation, National Retail Security Survey 2023, 2023. nrf.com
- SPI Research, 2026 Professional Services Maturity Benchmark, as summarized by Certinia, 2026. certinia.com
- Fundraising Effectiveness Project (AFP and GivingTuesday), Q4 2025 report, 2026. afpglobal.org
- Dantas, Fleck, Cyrino Oliveira and Hamacher, "No-shows in appointment scheduling: a systematic literature review," Health Policy 122(4), 2018. sciencedirect.com
- Robotham et al., "Using digital notifications to improve attendance in clinic: systematic review and meta-analysis," BMJ Open, 2016. ncbi.nlm.nih.gov
- Monica Rogati, "The AI Hierarchy of Needs," Hackernoon, 2017. hackernoon.com