Over the past weeks we published ten industry deep dives: tax firms, law firms, property managers, skilled trades, medical practices, insurance brokers, freight forwarders, hotels, car dealers and staffing firms. Each with the same method: chase claims to the primary source, date-stamp the numbers, and whatever cannot be verified gets cut. Across all ten investigations, five patterns emerged that we did not expect in this clarity.
1. There is a folklore economy
In almost every industry, automation is sold with numbers that crumble under scrutiny. The "80 percent of sales close after the fifth contact" credits an organization that never existed. The "ZDK survey showing workshops miss 45 percent of calls" is untraceable. The "KBV study on 40 percent unanswered practice calls" does not exist. The "6 seconds per CV" comes from a 2012 study with 30 participants; the "75 percent of applications die in the ATS" from a misreading. These numbers travel from vendor page to vendor page until they sound like common knowledge.
The annoying part: it is unnecessary. The defensible numbers are almost always more striking than the invented ones. That roofers answered only 37 percent of inquiries in a real field test, that 44 percent of hotel bookings are processed by hand, that a staffing firm cycles 789,000 contracts across 660,000 workers, all of that survives scrutiny and tells a harder story. Our rule from this, and our vendor test for you: ask for the source. Whoever cannot name it has not checked the rest either.
2. The staffing story is often yesterday's
"The skilled-worker shortage" is the default argument of every automation pitch, and in several industries the blanket narrative is simply stale. In hospitality, registered openings sit at a post-Corona low and the number-one worry is personnel costs. For medical assistants the calculated gap collapsed from nearly 7,000 to a few hundred. Temp staffing is back at 2010 levels. At the same time, where the shortage is real it is brutally concentrated: in construction electrics three of four openings are statistically unfillable, in vehicle technology six of ten.
Together this yields the more honest sales argument: automation does not fill vacancies, it lowers the cost of the remaining work and protects the margin. Where the minimum wage jumps 8.4 percent and returns sit at 1 to 3 percent, nobody needs a shortage narrative from 2021.
3. Industry software automates the center, never the edge
The most reliable pattern of all. DATEV and BMD automate the booking, not the document chasing. Practice software automates the documentation, not the telephone. Broker systems sort the inbox, not the portfolio care. Dealer systems administrate, but not a single DACH vendor ships its own phone agent. The core products are good at what they were built for, and the edge where the hours die, inquiries, phone, follow-up, document flows between systems, stays open across all ten industries. Adoption data matches:
For businesses this means two things. First, activate the AI features of the software you already own before commissioning anyone, including us. Second, the real automation case almost always sits in the integration layer in front of and between the systems, and that layer is exactly what your software vendor will not build for you.
4. Compliance is not the obstacle, it is the moat
Each of the ten industries has one hard legal line: professional secrecy in the Kanzlei world, the C5 attestation for health data, the broker's burden of proof, competition law on service reminders, the high-risk classification of candidate selection. None of these lines forbids automation. Each defines precisely where the human must sit, and thereby becomes the competitive advantage of whoever builds it in from the start instead of retrofitting it. The architecture is the same everywhere and by now carries almost the force of law: draft plus sign-off. The machine collects, structures, drafts and reminds. The human decides, signs and answers for it. From 2 August 2026 the AI-labeling duty adds the first rule that simply applies to everyone.
5. The calendar beats the business cycle
The workload peaks of small businesses are less often cyclical than statutory: the tax firms' quota ladder from October, the property managers' 30 June, the twelve months of § 556, assembly season, filing season. These dates are written into law and do not move, which makes them the most grateful automation targets there are: the backlog that escalates in spring is reliably created the autumn before, and a system that requests documents early and makes deadlines visible early defuses the peak before it exists.
What this means for your business
If your industry was among the ten: the deep dives live in the guide, each with sources. If not: the five patterns very likely apply anyway, because they hang not on the industry but on the structure of small businesses. The fastest reality check remains the 60-second check, and what projects cost is in the pricing overview, with sources. And if someone shows you the 80-percent statistic: you now know what to do.
As of 10 July 2026. All numbers and debunks come from the ten linked industry deep dives, where each is individually documented with primary sources.