Octopus Energy Shows Why AI Visibility Tracking Becomes a Multi-Market Problem

Octopus Energy is a useful AEO case because the problem was not writing more content. The problem was seeing how the brand appeared across AI answers in multiple markets without turning the process into manual spreadsheet work.

TL;DR

Octopus Energy used Ahrefs Brand Radar to replace a tedious manual process for checking AI visibility across ChatGPT, AI Overviews, and other AI-search platforms. The case does not prove a visibility lift, but it does show a more mature operating model: measure mentions, citations, markets, competitors, and source patterns before reshaping content strategy.

What was the situation?

Octopus Energy is a UK-headquartered energy company with retail energy, renewable generation, and low-carbon technology businesses. Ahrefs' case study says it supplies domestic electricity and gas to 7.7 million households across nine countries.

That market structure makes AEO harder than a single-domain audit. A company with multiple countries, older acquired brands, local competitors, and different regulatory contexts can show up differently in each AI answer surface. A prompt about switching energy suppliers in the UK may not behave like a prompt about renewable energy in Germany or Japan.

The core question was simple: when users ask AI systems about energy providers, does Octopus Energy show up, and is the answer accurate enough to trust?

What happened?

Ahrefs says Octopus Energy had been manually checking AI responses before adopting Brand Radar. Laura Iancu, an SEO Growth Specialist at Octopus Energy, described extracting AI responses from ChatGPT, AI Overviews, and other AI-search platforms as tedious.

Brand Radar replaced that manual process with a tool that could surface mentions, citations, and impressions across AI search and related channels. Ahrefs says the tool helped Octopus Energy present clear findings to executives and global marketing teams, which supported buy-in for content strategy and marketing initiatives.

That is the important result. The case is not a clean before-and-after citation-rate lift. It is a process lift: fewer one-off checks, clearer reporting, and better internal alignment around what AI systems were saying.

Why does this work at the answer-engine level?

AI visibility is distributed across prompts, surfaces, and sources. A brand can be mentioned without being cited, cited through a third-party page, omitted from comparison prompts, or described through an outdated source.

Ahrefs' methodology explains why prompt design matters. Brand Radar uses real search demand, People Also Ask style questions, and semantic fanout to create broader prompt coverage. It then stores AI responses so users can search mentions and citations across platforms such as ChatGPT, Perplexity, Gemini, Copilot, Google AI Overviews, and AI Mode.

For AEO, that is a better model than checking five handpicked prompts once. Visibility should be treated as a distribution. The brand's presence can vary by country, wording, competitor set, source freshness, and platform.

What evidence is strongest?

The strongest evidence is operational, not causal. Ahrefs reports that Brand Radar's output matched what Octopus Energy had been finding through manual checks, and that the tool made the work easier to explain and export.

That matters because many AEO tools fail at the internal-use layer. A technically interesting visibility metric is not useful if regional teams, executives, PR, and content leads cannot understand it. Octopus Energy's case shows the value of plain metrics: mentions, citations, impressions, competitors, and cited sources.

The case also has a credible pain point. Multi-market companies really do face brand-history problems. If older acquired names still appear in AI answers, or if different regions have different citation sources, the fix requires more than one global page update.

What is replicable?

The replicable part is the workflow. You do not need Ahrefs to copy the operating model, though a platform can make it easier.

Start with markets and prompt classes. For each market, define brand prompts, competitor prompts, category prompts, and switching or buying prompts. Run them across several answer engines. Record whether the brand appears, whether it is cited, which URL is cited, which competitor appears, and whether the description is accurate.

Then sort gaps by action. If the brand is absent from category prompts, the issue may be authority or third-party mentions. If it is cited from the wrong page, the issue may be documentation or page targeting. If an old brand appears, the issue may be entity cleanup. If competitors dominate comparison prompts, the issue may be product positioning or review-site coverage.

What is uncertain?

The case does not show that Brand Radar directly increased Octopus Energy's AI visibility. Ahrefs describes monitoring, reporting, and strategy support, not a controlled lift in citations or revenue.

That distinction matters. AEO teams should not report "we improved AI visibility" when the evidence only says "we can now measure AI visibility." Measurement is still valuable. It is just a different claim.

There is also tool-methodology risk. Any modeled prompt database reflects assumptions about what users ask, which engines are tested, and how frequently answers are refreshed. Ahrefs is more transparent than many vendors, but the numbers still represent potential visibility rather than exact audience reach.

How should teams apply it?

Teams should use this case to design AI visibility reporting before designing AI content. The order matters. If you publish first and measure later, you cannot tell whether the work changed anything.

Build a visibility dashboard with five columns: prompt, answer engine, brand mention, cited source, and action needed. Add market, language, and business line where relevant. Review it every two weeks with content, PR, product marketing, and regional leads.

The output should not be a vanity score. It should be a work queue. Which page needs rewriting? Which outdated third-party source needs correction? Which comparison query needs a stronger proof asset? Which brand entity is confusing the answer engine?

What to do Monday morning

1. Pick one market and one product line. Do not start global. 2. Create 20 prompts from sales questions, support questions, category terms, and competitor comparisons. 3. Run the same prompts across three answer engines and log mentions, citations, and answer accuracy. 4. Tag each cited source by type: owned page, review site, news article, Reddit, documentation, or official source. 5. Turn the top five gaps into content, PR, documentation, or product-marketing tasks. 6. Repeat the prompt panel every two weeks before expanding to more markets.

Sources

• How Octopus Energy uses Ahrefs Brand Radar to monitor AI visibility across global markets
• Ahrefs Brand Radar Methodology
• What is Brand Radar, and how to use it?

How this page should be used

This page is meant to act as a durable source page for site owners, content leads, SEOs, and builders working on answer-engine visibility. It should not be treated as a short definition or a loose blog note. The practical job is to help someone make a better publishing, crawling, content, or measurement decision after reading it.

For AEO work, usefulness comes from the combination of a clear answer, visible evidence, specific examples, and a next action. A page that only defines the term may earn a first impression, but a page that gives the workflow is more likely to be saved, linked, cited, and used as source material by humans and answer systems.

The operational model for Octopus Energy Shows Why AI Visibility Tracking Becomes a Multi-Market Problem

The operating model is simple: define the topic, identify the page or query family it supports, remove access blockers, structure the answer clearly, connect it to the rest of the site, and measure whether the intended page is being selected. That sequence matters because later steps cannot compensate for earlier failures.

Layer	Question to answer	What good looks like
Purpose	What job should this page perform?	The title, H1, first answer, and internal links all point to the same source role.
Access	Can the intended crawler or reader fetch it?	The URL returns 200, is canonical, is indexable when intended, and is not blocked by robots, CDN, or firewall rules.
Retrieval	Can one section answer a real prompt?	Headings are specific, the first sentence answers directly, and examples or tables reduce ambiguity.
Evidence	Why should the answer trust this page?	Official documentation, original tests, screenshots, data, examples, or methodology sit near the claims they support.
Connection	Where does this page fit in the site?	The page links to its parent hub, related glossary terms, tools, methodology, and proof pages.
Measurement	How will we know it worked?	The team tracks Search Console query movement, prompt-panel mentions, exact URL citations, and competitor source replacement.

Implementation workflow

1. Choose the prompt family. Decide whether this page is answering a definition, comparison, how-to, tool, diagnosis, checklist, or platform-specific query.
2. Write the short answer first. The opening answer should be clear enough that a reader understands the page before reading the details.
3. Map the follow-up questions. Each major H2 should answer the next thing a serious reader would ask.
4. Add evidence where it changes the decision. Cite official docs for crawler or platform claims. Use original examples or methodology for observed behavior.
5. Add internal links deliberately. Link up to the hub, sideways to related reference pages, and down to tools or templates.
6. Run the publishing checks. Confirm canonical URL, indexability, sitemap inclusion, llms.txt inclusion when appropriate, and mobile readability.
7. Measure after publishing. Watch whether impressions, mentions, or citations land on this exact page rather than a less relevant URL.

What to improve before calling this page finished

A page about Octopus Energy Shows Why AI Visibility Tracking Becomes a Multi-Market Problem is not finished just because it is long. It should make the next step easier. If the reader is learning, it should give them a learning path. If the reader is implementing, it should give them a workflow. If the reader is auditing, it should give them a checklist. If the reader is comparing options, it should give them decision criteria.

• Add a direct answer for the main question the page targets.
• Add a table when the reader needs to compare terms, tools, crawlers, pages, or decisions.
• Add examples when the guidance could otherwise feel abstract.
• Add caveats where the industry tends to overclaim.
• Add a measurement step so the page connects to real outcomes.
• Add internal links so the page strengthens the site’s topical graph.

Common mistakes

The first mistake is treating AEO as a label rather than an operating system. Adding the phrase “answer engine optimization” to a page does not make it a source. The page still needs crawl access, entity clarity, evidence, and a reason to be cited.

The second mistake is confusing source maps with crawler controls. XML sitemaps help discovery. robots.txt controls crawler access. llms.txt can act as a curated source map. Those files should agree with one another, but they do not do the same job.

The third mistake is scaling weak pages. If the core page for a topic is thin, unclear, or unsupported, creating ten related thin pages usually spreads the weakness around. The better move is to deepen the source page, add examples, and use internal links to consolidate intent.

Quality standard for Optimize AEO pages

Every durable Optimize AEO page should meet a higher bar than a short blog post. The page should answer the main query, explain the method, show where the page fits, and give the reader a practical action. For ranking and citation purposes, the target is not simply more words. The target is enough useful detail that the page can compete with larger authority sites while still being more specific, more operational, and easier to use.