The case-study block on the homepage, the AI page and every location page carries two cards that undo themselves. An untitled "ai document processing platform" card describes "a comprehensive talent solution" for hiring, but reports results for "patient financing", with 62%, 78% and 35% gains attached. The Prior Authorization AI card claims "43% Fewer submission errors" and "63% Admin time reduced". Its own full story has no percentages and says the product is "in final validation ahead of its first Medicare submission". SoftDoes is hiring to sell into compliance-heavy industries, where buyers click through and check.
First action
Remove the untitled document-processing card from the shared block today, and put the Prior Authorization AI card's measured 16-hour result in place of its percentages.
How to measure it
Cards whose every number matches the full case study
- Now
- –
- Target
- 100% of cards
- By
- 14 days
Steps
List every card
Export the 19 cards in the shared block with their three metrics each.
Marketing Lead 2 hours
Match to stories
For each metric, find the sentence in the full case study or the Clutch review that supports it.
Marketing Lead 1 day
Fix the two contradictions
Remove or correct the untitled card; restate the Prior Authorization AI card as hours saved until submission data exists.
Delivery Lead 1 day
Replace generic numbers
Swap duplicated claims like "40% Higher User Retention" for each client's own measure.
Delivery Lead 2 days
Add a measurement note
One line under the block: where the numbers come from and when they were measured.
Marketing Lead 1 hour
Sync the case-studies page
Make the homepage block and /case-studies show the same set of studies.
Senior Front-End Engineer 1 day
Set a rule
No new card ships without the delivery lead signing off its numbers.
Founder and CEO ongoing
Why it works
Run a one-week proof audit of every card in the shared block. For each card, keep a number only if the full case study or a Clutch review supports it, and say how it was measured. Then: remove or rename the untitled card and give it one correct description; change the Prior Authorization AI card to its real result (16+ weekly hours of phone calls replaced by automated intake) until submission data exists; and swap the two identical "40% Higher User Retention" claims for each client's own measure. Add a short "How we measure" line under the block. The block is shared, so one fix updates about 930 pages.
Procurement and compliance buyers read specific numbers as a test of honesty. One contradiction makes every other number look made up, while numbers that trace back to a source make the honest track record (510+ projects, 5.0 on Clutch) believable. It also stops AI assistants and review sites from repeating figures SoftDoes can't defend.
Ready to use
ChecklistCase-study card proof check
[ ] Company name shown (no untitled cards) [ ] Description and outcome describe the same product [ ] Each number appears in the full case study or a Clutch review [ ] Each number says what was measured and over what period [ ] No metric copied from another client's card [ ] Product status matches the claim (no error rates before first use) [ ] Same cards on the homepage and /case-studies [ ] Delivery lead sign-off date
Risks and how to handle them
Fewer impressive numbers on the page
Lead with measured results such as hours saved and billing errors cut, which are concrete enough to sell.
Clients can't confirm older metrics
Show the qualitative outcome instead of a percentage until a 90-day review measures it.
Evidence
- Your websiteHomepage case-study block (19 cards; the same block appears on /services/ai and the location pages): an untitled 2025 card, "ai document processing platform", is described as "A comprehensive talent solution designed to help companies attract, hire, and retain top talent" while its outcome says it "significantly reduced staff steps for presenting and managing patient financing", with "62% Faster Processing Time", "78% Less Manual Work" and "35% Improved Accuracy". Sparkle and DineMate both show "40% Higher User Retention".Source
- Your websitePrior Authorization AI: the card shows "16 Weekly hours saved", "63% Admin time reduced" and "43% Fewer submission errors", but the full case study has no percentages and says the platform "is live and in final validation ahead of its first Medicare submission"; its stated result is document collection moving from "16+ hours of weekly phone calls" to automated fax intake.Source
- Your websiteCase studies page: 17 cards (10 from 2026, 7 from 2025); six 2025 cards (Forbidden Alchemy, Altera, Brizacorp, Color & Harmony, The Modern Gentlemen, OQ Technology) have no industry tags; Grand Central Language Services, a translation and interpretation company, is tagged Education; the untitled document-processing card and the 2024 Bokeyno Motorsports card on the homepage are not on this page or in the case-study sitemap.Source
- HiringCareers page: one open role, Business Development Manager in Mission, Kansas (Kansas City metro), applied for through LinkedIn; the page says SoftDoes will "build production systems for businesses in compliance-heavy industries" and "lead AI implementations".Source