Key findings
- Sitecore says the combination connects Scrunch's answer-engine insights and Agent Experience Platform with SitecoreAI content workflows; financial terms were not disclosed.
- In Geolix.ai's English fintech case, ChatGPT cited the target brand's own domain in 22.9% of answers but mentioned the brand in only 6.9%. Citation, visibility and recommendation require separate workflows and metrics.
- After four articles were published, average ChatGPT own-domain citation rate moved from 9.7% before publication to 26.7% afterward, while brand mention moved from 2.2% to 8.3%. The change is observational, not proof of causation.
- A Chinese-market case found only 12% Jaccard overlap between citation domains used by Chinese-engine and Western-engine groups, showing why a global content stack still needs market-specific source operations.
What happened
Sitecore announced that it had acquired Scrunch, which it described as an AI customer-experience platform. Sitecore said Scrunch would surface buyer queries, brand-representation gaps and competitive positioning, while SitecoreAI would help teams manage and activate content across web, social and other channels. The official announcement frames the transaction as a move from insight to action and says Scrunch recommendations will be automated within SitecoreAI content-management, content-marketing and digital-asset-management solutions.
Sources: Sitecore acquisition announcement
Sitecore also reported vendor case-study results: Akamai's AXP-enabled pages achieved a 364% increase in brand presence for non-branded prompts and a 218% increase in citations compared with selected non-AXP pages; Runpod associated its AI-search work with a 400% increase in paying customers. These figures come from the acquisition announcement and should be treated as company-reported case evidence, not independent market benchmarks.
Why integration matters: Geolix.ai evidence from the operating layer
Geolix.ai's English fintech case monitored nine English purchase-intent questions for a GEO service brand focused on fintech in Singapore and APAC. From July 20 to 31, 2026, ChatGPT API, Gemini and Perplexity generated 15,495 valid answers and 223,196 citation records. The same target brand produced very different outcomes by engine.
| Engine | Valid answers | Mention rate | Top-1 rate | Own-domain citation |
|---|---|---|---|---|
| Perplexity | 3,150 | 75.3% | 51.9% | 85.5% |
| Gemini | 3,195 | 38.6% | 25.6% | 25.5% |
| ChatGPT API | 9,150 | 6.9% | 0.8% | 22.9% |
ChatGPT is the clearest example of why enterprise integration cannot stop at a visibility score. The brand's domain entered the evidence set in 22.9% of answers, but the brand name appeared in only 6.9%, a 3.3-to-1 gap. A content system must therefore distinguish at least four states: the page is crawlable, the page is cited, the brand is mentioned, and the brand is recommended. Each state points to a different action and owner.
The content-release observations show the value of a connected feedback loop. Four articles launched on July 22. Average ChatGPT own-domain citation rate was 9.7% before publication and 26.7% afterward; average brand mention rate was 2.2% before and 8.3% afterward; cited own-domain URLs increased from two to an average of 7.2. A second six-page batch produced 2,081 citations for the leading new page and 1,891 for its Chinese version within four days. Multiple pages and actions changed at once and there was no control group, so these figures show timing and association, not a single-page causal effect.
The enterprise stack still has a market-coverage problem
A DXP can improve governance and execution without automatically covering every retrieval ecosystem. In Geolix.ai's low-code SaaS case, ten Chinese purchase-intent questions were run across six engines from June 15 to 19, producing 1,233 valid answers and 15,353 citations. Mention rates ranged from 10.3% on Perplexity to 81.9% on DeepSeek. Chinese sources represented 90.4% to 100% of citations across all six engines.
Source types diverged within that Chinese-language environment. DeepSeek drew 62.5% of citations from cloud and official developer ecosystems; Doubao drew 31.1% from user-generated and video sources; Qwen drew 72.6% from user-generated and video sources and 25.1% from search self-reference. Chinese-engine and Western-engine groups shared only 68 domains, a 12% Jaccard overlap. These are case-specific results, but they show why APAC execution needs local source mapping, not only centralized publishing.
What the combined Sitecore-Scrunch proposition must deliver
- Prompt-level evidence: raw answers, cited URLs, engine interface, language, location and collection date.
- Metric separation: crawlability, citation, brand mention, recommendation rank and business traffic reported independently.
- Action routing: each gap assigned to content, technical, PR, compliance or market operations with an owner and due date.
- Controlled facts: legal entity, licences, availability, fees, eligibility and risk statements governed across versions.
- Market-specific distribution: source and channel plans for ChatGPT, Gemini, Perplexity, DeepSeek, Doubao and Qwen rather than one universal publishing rule.
- Retesting and auditability: a fixed question set, change log, approval record and post-publication observation window.
Where Geolix.ai fits
Geolix.ai occupies a narrower layer than a global DXP. It is a Singapore-based managed GEO service focused on fintech and monitors eight Western and Chinese engines. Its role is to define market-specific buyer questions, interpret answer and citation gaps, coordinate on-site and external-source work, and repeat measurement. A large enterprise may use both layers: a DXP for content governance and global infrastructure, and a specialist service for bilingual prompt research, source-ecosystem mapping and regulated-market execution.
What buyers should ask now
- Which engines, interfaces, languages and regions are actually sampled, and what is the collection success rate?
- Can the system show raw answers and URLs behind every aggregate score?
- Does a detected gap create an accountable task with evidence, approval and retest dates?
- Can the workflow distinguish a citation gain from a mention or recommendation gain?
- How are regulated claims reviewed before content is changed or distributed?
- Which local publishers, developer ecosystems, communities and video platforms are covered in China and APAC?
Methodology and limitations
Geolix.ai figures come from an anonymized, read-only production-monitoring export dated July 31, 2026. The two cases are not a matched language experiment: they differ in category, brand, prompts, date range, repetition level and engine set. In the English fintech case, Google AI Mode succeeded on only 1 of 450 scheduled runs and Google AI Overviews on 62 of 450, so both were excluded. The export has no factual-accuracy field. Before-and-after publishing results are observational and should not be interpreted as causal estimates.
Frequently asked questions
Does the acquisition mean GEO is now part of every DXP?
No. It shows that one major DXP provider sees AI-discovery measurement and execution as close to core content operations. Buyers must still verify product, engine and market coverage.
Does connecting content workflows guarantee better AI recommendations?
No. Integration can shorten the path from diagnosis to action, but results still depend on source quality, engine behavior, distribution, compliance and repeated testing.
Do enterprise platforms remove the need for specialist GEO services?
Not necessarily. Platforms can provide infrastructure and governance; specialists may still be needed for industry judgment, bilingual source ecosystems and accountable execution.