What your customers really think about AI on the phone
The honest answer first: we know of no independent, representative survey on AI answering small businesses' phones. What there is: a few evidenced data points — including a vendor study where acceptance has conditions —, typical opinion patterns and, above all, a way to find out where your customers stand. That is what counts, not the average.
What is evidenced — and what is not
Four data points hold up, no more:
- Origin matters: for 88 % of companies the AI provider's country of origin is important; "in this context, 93 per cent would prefer AI from Germany" (Bitkom, September 2025, 604 companies with 20+ employees) — whether the 93 % refers to all respondents or to those 88 %, the press release leaves open. That is a business survey, not a consumer survey — but a strong signal for how sensitive the subject of trust is.
- The underlying problem is acknowledged: 60 % of respondents say small businesses are always or often hard to reach by phone (YouGov/Enreach, 2023). So customers know from their own experience what the AI is up against.
- Acceptance comes with conditions: in a survey by the telephony provider Placetel with YouGov and the Deutscher Mittelstands-Bund ("State of Voice AI Report 2025", more than 2,000 consumers), 87 % would accept voice AI "if certain conditions are met" — named are being able to be put through to a person at any time (52 %), clear labelling as AI (41 %) and strict data protection and transparency (32 %). That is a vendor study; the fieldwork dates and question wording are not on its results page.
- Some refuse: the announcement of a podcast with the CEO of the provider fonio.ai says: "In some industries 50 % hang up, in others only 20 %" (Trendingtopics.eu, 2 December 2025). That is a topic line written by the editors, not a study — but it hints at the range.
Figures like "X % of Germans love/hate AI on the phone" without a study, a date and the question asked are not something we repeat.
Four typical opinion patterns
No percentages and no analysis of our own — a grid for reading the transcripts of your own calls:
- The pragmatists: want to get their request off their chest; whether human or machine is secondary, as long as it works. Never lose this group to poor system quality.
- The pleasantly surprised: expect a recorded announcement and get a conversation with an appointment. The "oh, that actually worked" moment is your strongest advocate.
- The politely sceptical: they play along, but keep it brief. What counts for them is that the promised callback arrives on time — then the scepticism can flip.
- The refusers: hang up or demand a human immediately. The way out is built for them ("What to do when a caller refuses to talk to the AI?") — and honestly: some of them would probably have hung up on voicemail too.
Whether the widespread generational cliché ("older people cannot cope with it") is true of your customers only your own transcripts will show — do not assume it.
How to find out for your own business
- Read transcripts instead of guessing: after two weeks the conversations show how your customers react — where they go along with it, where they hesitate, what they ask.
- Count the drop-offs: calls that end in the first seconds are the best approximation of your refusal rate — a number instead of a feeling.
- Ask regular customers directly: ten short conversations ("how was it with our assistant?") deliver more concrete feedback than any gut feeling.
- Run the callback test: does the promised callback come reliably? An assistant that promises callbacks is only as good as the team that makes them — a factor that often gets lost in the debate about AI on the phone.
The honest summary
Our assessment: more than any general attitude to "AI on the phone", what counts for your customers is their last experience with your business. An assistant that picks up quickly, says honestly what it is, takes the request down cleanly and keeps the way to a person open produces exactly the opinion you need — provided your team calls back on time. One that cannot do that confirms every prejudice. The technology decides less than the execution.
Frequently asked questions
Is there really no survey on the acceptance of AI telephony in Germany?
None that is independent and that we know of asks consumers representatively about AI call answering at small businesses. There is a vendor study by Placetel with YouGov (2025, results above), and there are representative surveys on chatbots in customer service: in Bitkom's (May 2025), 62 % of online shoppers would most like to turn to a quickly reachable human contact when something goes wrong after a purchase, and 36 % name a chatbot. Both are pointers, not evidence about your customers.
Should I tell my customers before I introduce it?
Regular customers yes — one sentence in person or in a newsletter helps ("when I am on site, our AI assistant takes down your request from now on — I will call you back"). Walk-in customers learn it from the greeting, which includes the AI disclosure. Informing your callers about how their data is processed is your own task anyway, in your privacy policy; the help centre has non-binding wording for it under "Informing callers".
How quickly does acceptance change?
There are no robust figures on that. Measure the drop-off rate monthly — if it falls, the system is working for you; if it stays high, check the greeting and the knowledge base first, and whether the promised callbacks happen.
Status & sources
As of 2026-11-15.
- Bitkom, press release of 15 September 2025 (604 companies with 20+ employees, telephone survey, weeks 27 to 32 of 2025, representative): "Für eine breite Mehrheit von 88 Prozent der Unternehmen ist das Herkunftsland des KI-Anbieters wichtig. Dabei würden 93 Prozent eine KI aus Deutschland bevorzugen" (the base of the 93 % is not stated)
- Enreach/YouGov, press release of 25 October 2023: online survey April 2023, 2,099 respondents, representative of the German population aged 18+; 60 % say small businesses are always or often hard to reach by phone
- Placetel, "Voice AI im Reality-Check – Das sagen KMU & Verbraucher 2025" (State of Voice AI Report 2025, with YouGov and the Deutscher Mittelstands-Bund; more than 2,000 consumers, over 500 SMEs): 87 % would accept voice AI "if certain conditions are met" (put through to a person 52 %, labelled as AI 41 %, data protection and transparency 32 %). Vendor study; fieldwork dates and question wording not given
- PCtipp, "Mittelstand plant Sprung in die Sprach-KI – zögert aber noch", 4 November 2025: report on the Placetel State of Voice AI Report 2025 (529 SME decision-makers, 2,000 consumers)
- Bitkom, press release "Kundenservice beim Online-Shopping: Mensch schlägt Chatbot" of 22 May 2025 (1,006 internet users aged 16+, including 978 online shoppers, online, weeks 1–3 of 2025, representative)
- Hang-up rates: topic line under "Akzeptanz und Grenzen" in the announcement of the podcast with fonio.ai CEO Daniel Keinrath on Trendingtopics.eu of 2 December 2025 ("In manchen Branchen legen 50 % auf, in anderen nur 20 %") — the publisher's summary, not a quote from the CEO and not an independent study. Opinion patterns qualitative, explicitly without percentage claims
- scoco.ai product information: as of 15 November 2026