German Telco Study 2026 · Pyramid
The single number.
Four stages: baseline measurement, intervention, measurement after twelve months — and what really counts in the end. An impact logic condensed to one figure.
NORTHBRIDGE
As of May 2026
DE
This is not a promise. This is a calculation. All input quantities of this pyramid come from published industry sources (GSMA, Similarweb, Indig, Scrunch, Aggarwal et al.) or from NB's own retrieval analysis of April 2026. The projected outcomes are a model scenario on the basis of these input quantities, not yet validated against a real mandate. The validation test with an actual DE telco client is the subject of the study's follow-up phase.
00
Foundation · Market reality 2026
The channel exists, and it moves fast
1.1 bn
GenAI visits June 2025
Similarweb
+357 %
Growth YoY
Similarweb
44.2%
Citations from the first 30%
Indig 2026
60%
Citation drift per month
Scrunch AI
01
Stage 1 · zero measurement before mandate start
Three operator sizes, three baseline visibilities
Small · MVNO reseller
Citation share across 6 engines
Medium · branded MVNO
Citation share across 6 engines
Large · network operator
Citation share across 6 engines
02
Stage 2 · intervention over 12 months
Orchestrated procurement per the Northbridge Procurement Standard
180–200
Articles per month
6
Engines blended 0,89 ×
18
Procurement criteria
V01–06
Disclosure variants
3
Disqualification levels
03
Stage 3 · measurement after 12 months
Citation share across the six engines
Small
+15 PP
Citation share across 6 engines
Medium
+22 PP
Citation share across 6 engines
Large
+24 PP
Citation share across 6 engines
04
Stage 4 · The single number
20of100
For a medium operator: twenty out of a hundred sales units after twelve months come from a channel that practically did not exist twelve months earlier — direct LLM citation. Small 8, large 32.
What the number does behind the curtain
Revenue rises moderately, but the channel architecture shifts radically: medium operator, sales index 100 → 108
Channel
before
after
Δ
Mechanism
Comparison portals
35
30
−5
LLMs substitute part of portal search; dependency falls, margin stays in-house
Organic Search (Google)
30
25
−5
Google's classic SERP loses query volume to ChatGPT/Perplexity/Gemini
Paid (SEA/Display)
20
18
−2
Slight flattening because part of the conversions arrive organically via LLM citations
Word-of-Mouth / Referral
10
10
0
Stable, independent of the generative shift
Retail / direct sales
5
5
0
Stable
Direct LLM citation (new)
—
20
+20
ChatGPT-Conversion 7 % gegenüber Google 5 % = +40 % struktureller Conversion-Vorteil
Total sales index
100
108
+8
But: 20 of 108 come from the new channel. That is the structural value.
Method · sources · limits
Model assumptions
Baseline citation share from Sistrix prompt research DACH 2025 (62 m queries) and Scrunch half-life study. Referral volume and conversion from Similarweb GenAI Landscape 2025 (1.1 bn visits, 7% vs 5%). Retrieval mechanics from Indig 2026 (44.2 / 53 / 36.2 per cent) and Aggarwal et al. KDD 2024 (+41 / +28). Channel-mix assumption for a medium DE telco operator from publicly available industry mix data, normalised to sales index 100.
The 180–200 articles per month are an example workload for the NB four-phase architecture (selection, procurement, editorial, reporting). Distribution across six engines follows the price-factor matrix with blended factor 0.89×.
Sources
- Similarweb · GenAI Landscape 2025
- Indig 2026 · ChatGPT position study
- Scrunch AI · half-life study 2025
- Aggarwal et al. · GEO-Bench, KDD 2024
- Sistrix · prompt research DACH 2025
- GSMA Intelligence · Q4 2024
- Northbridge · retrieval analysis April 2026
- Northbridge · publisher research Telco DACH 2026
Northbridge · model calculation
German Telco Study · Pyramid · May 2026