Transparency

How we prove that Pro Plus AI saves time.

We want to be more than a website agency. meisterkontor positions itself as a digital efficiency showcase for the region — we practise what we sell and we publish what underpins our numbers. This page shows, lever by lever, which studies back the claim "in a typical case, four to eight hours per week saved", how we extrapolate the numbers and where the limits sit.

As of 2026-05-31. We update the evidence when new studies appear.

01 — Projection

Three values, three rationales.

The claim "four to eight hours per week" sits deliberately in the mid range of the study-backed spread. The conservative figure (low) and the typical case (mid) are our marketing anchors — the high end is plausible but not promised.

3.1 h

Low — robust against critique

Even if only half the AI levers run cleanly and a few levers see little volume, the St. Louis Fed lower bound of 2.2 hours per week per AI user plus appointment and mail synergies is reliably achievable.

6.0 h

Mid — typical case

Sits in the corridor between St. Louis Fed (2.2 h per person) projected onto a 1- to 3-person business with multi-user tool adoption (owner plus one office assistant) and the KMU bureaucracy time of 32 hours per month (per KfW 2025).

9.7 h

High — best in class

HERO/Statista 2025 quotes 10 to 20+ hours per week for digital end-to-end solutions. Because meisterkontor doesn't replace an ERP but delivers eight add-on levers, we realistically reach half that value — around 10 hours for businesses with high enquiry and appointment volumes.

The marketing claim "in a typical case, four to eight hours per week" sits between the low and the mid value — deliberately below the high end. That keeps it both compliant with German competition law (no misleading advertising under Section 5 UWG) and within the real expectation corridor.

02 — Lever by lever

Eight levers. Eight pieces of evidence.

Each of the eight Pro Plus AI levers has its own study base. Where no peer-reviewed study exists specifically for trades, we use the closest reliable proxy and name it explicitly. Vendor studies are labelled as such and discounted in the projection.

  1. AI mail triage

    Study: Brynjolfsson/Li/Raymond, QJE 2025 (RCT, 5,000 support agents)

    Plus 13.8 % more enquiries handled per hour with AI assistance, plus 35 % in the bottom quintile — exactly the situation of an owner without specialist training. At 6 h/week mail workload and 30 % reduction: roughly 1.5 to 2 h saved per week.

    Low
    1.0 h/wk
    Mid
    1.8 h/wk
    High
    2.5 h/wk

  2. AI chat widget

    Study: Comm100 AI Live Chat Benchmark 2026 (220 M interactions, vendor data)

    44.8 % of all chats are fully resolved by the bot; for tightly scoped small teams up to 89 % resolution rate. Exactly the pattern for a trade business with 5 to 10 recurring standard questions. Handoff satisfaction 99.4 %.

    Low
    0.3 h/wk
    Mid
    0.7 h/wk
    High
    1.2 h/wk

  3. Appointment pre-qualification + online booking

    Study: Bitkom Handwerk 2025 (n=504) + HERO/Statista 2025 (vendor-bias adjusted)

    Only 48 % of trade businesses offer online booking today — the majority still take appointments by phone. At 3 h/week phone time and a 60 % reduction through online slots plus AI pre-qualification: roughly 1.5 to 2 h saved per week.

    Low
    1.0 h/wk
    Mid
    1.8 h/wk
    High
    3.0 h/wk

  4. Review reply generator

    Study: Proserpio/Zervas, HBR 2018 (TripAdvisor vs. Expedia, quasi-experimental)

    Businesses that start responding to reviews gain plus 12 % more reviews and plus 0.12 stars on average; a third of hotels gained half a star within six months. The lever is primarily revenue, not time — we keep the time saving deliberately low in our model.

    Low
    0.1 h/wk
    Mid
    0.2 h/wk
    High
    0.4 h/wk

  5. Seasonal reminders for regulars

    Study: Win-back aggregate (e-commerce data, no peer-reviewed trade study)

    There is no peer-reviewed trade study on reactivation conversion. Closest proxy: e-commerce win-back with 10 to 20 % reactivation rate. At 100 regulars and 15 % conversion: about 15 additional appointments per seasonal wave. Pure time saving via AI-prepared mass mails: 0.3 to 0.5 h/week averaged.

    Low
    0.2 h/wk
    Mid
    0.4 h/wk
    High
    0.7 h/wk

  6. No-show reminders

    Study: Ulloa-Perez et al., Permanente Journal 2022 (RCT, n=158,669 appointments)

    An additional SMS reminder reduces no-shows by 7 % (primary care) to 11 % (mental health). Older Cochrane consensus: 20 to 30 % vs. no reminder. The lever is primarily revenue (fewer cancellations), not time. At 50 appointments/week and a 20 % to 14 % reduction: about 3 fewer no-shows per week.

    Low
    0.1 h/wk
    Mid
    0.3 h/wk
    High
    0.5 h/wk

  7. Gallery auto-tagging for SEO + accessibility

    Study: Vendor anecdotes (AltText.ai, PhotoTag.ai — no peer-reviewed study)

    Vendor data quotes 50 to 70 % time saving on tagging vs. manual work. Conservative assumption: 60 % reduction. At 1 h/week manual tagging: 0.5 to 0.7 h saved per week. Plus compliance — accessibility alt texts get assigned correctly without extra effort.

    Low
    0.3 h/wk
    Mid
    0.6 h/wk
    High
    1.0 h/wk

  8. Weekly digest

    Study: St. Louis Fed (Bick/Blandin/Deming), February 2025

    GenAI users report 5.4 % work-time savings across tasks. Pure time saving for the digest: 0.1 to 0.3 h/week. The main lever here isn't time saved — it's that the owner makes data-driven decisions that otherwise wouldn't happen at all.

    Low
    0.1 h/wk
    Mid
    0.2 h/wk
    High
    0.4 h/wk

  9. Sum across eight levers

    Low
    3.1 h/wk
    Mid
    6.0 h/wk
    High
    9.7 h/wk

03 — Method

How we extrapolate — and where we round down.

Source selection

We prefer peer-reviewed studies and large representative surveys from public research institutions (KfW, Bitkom, St. Louis Fed, ifo). Vendor studies (HERO Software, Comm100) are documented with a bias note and their values are discounted against independent sources in the projection.

From per-task to per-week

Where studies report percentage values (e.g. 5.4 % work-time saving at St. Louis Fed), we project them onto a 40-hour week. For business-specific levers (mail, appointment, gallery) we start from typical micro-business assumptions — 6 h/week mail, 3 h/week phone bookings, 1 h/week gallery. These assumptions are made explicit in the lever-by-lever section.

Where the limits sit

For seasonal reminders in trades and for photo tagging there are no peer-reviewed studies. We work with the closest reliable proxies (e-commerce win-back data, medical no-show RCTs) and say so. Actual time saving depends individually on business size, enquiry and appointment volume, and consistent use of all eight levers.

What we deliberately don't promise

No "up to 20 hours" claims (that's the HERO number for ERP replacement, not for an 8-lever add-on). No "guaranteed time saving" — studies talk about estimates and ranges, never guarantees. No absolute promises without the qualifier "in a typical case" or "depending on your business".

04 — Sources

Who wrote all this.

Each link leads directly to the primary source (PDF or official page). Where only a secondary reference or PMC mirror is available, that's noted alongside.

Primary studies — peer-reviewed or large representative surveys

  • KfW Research (April 2025): "Sieben Prozent der Arbeitszeit im Mittelstand für bürokratische Prozesse." Fokus Volkswirtschaft Nr. 495, n=10,000. PDF
  • Bitkom Research (August 2025): "Handwerk 2025 — Digitalisierung des Handwerks." n=504, CATI, representative. Overview · Report PDF
  • Bitkom Research (February 2026): "Künstliche Intelligenz in Deutschland — Studie 2025." Representative business survey. PDF
  • Bick, A. / Blandin, A. / Deming, D. (February 2025): "The Impact of Generative AI on Work Productivity." Federal Reserve Bank of St. Louis, On the Economy. Article · Working paper PDF
  • Brynjolfsson, E. / Li, D. / Raymond, L. R. (Quarterly Journal of Economics 2025): "Generative AI at Work." RCT with 5,000 customer-support agents. QJE paper · NBER PDF
  • Proserpio, D. / Zervas, G. (HBR February 2018, Marketing Science 2017): "Online Reputation Management: Estimating the Impact of Management Responses on Consumer Reviews." HBR article
  • Ulloa-Perez, E. et al. (Permanente Journal April 2022): "Pragmatic Randomized Study of Targeted Text Message Reminders to Reduce Missed Clinic Visits." n=158,669 appointments. Full text
  • Hasvold, P. E. / Wootton, R. (JAMIA 2011): "Use of telephone and SMS reminders to improve attendance at hospital appointments: a systematic review." PMC full text
  • ifo Institut (November 2024): "Entgangene Wirtschaftsleistung durch hohen Bürokratieaufwand." Press release · Study PDF

Vendor studies — with bias note

  • HERO Software / Statista (Summer 2025): "Handwerk digital." CATI, construction trades only. Note vendor bias — figures at the upper end. Overview
  • Comm100 (February 2026): "AI Live Chat Benchmark Report 2026." 220 M chat interactions, vendor data, very large n. Article

Questions about the evidence?

On request we share the full research file with the lever-by-lever reasoning — just write us a short note about your case.

Write an enquiry