Helping your business grow through SEO & PPC.

Sacha Redler

Peec AI

We help brands to measure and improve their visibility in AI search.

CPO & CMO

idealo internet GmbH

• Member of the Executive Team (Geschäftsleitung).
• Responsible for 10+ websites in 5+ countries.
• In charge of SEO, Content Marketing, and GEO
• Grew SEO traffic (in main market): PDPs+ 21%, PLPs +100%, Magazine +150%
• Generated and earned 50M+ backlinks.
• Grew Google Discover traffic by 1600%

Managed a business unit with 30+ people. Reported to the CEO.

Vice President of SEO

comparison shopping, price comparison, css, ecommerce, marketplace, platform, aggregator, seo, content marketing

• Responsible for 30+ websites in 10+ countries.
• Restructured the SEO and Content Marketing teams.
• Developed two individual contributors into teamleads.
• Made it to the Google Core Web Vitals Hall of Fame.
• Took idealo from the 24th largest to the 12th largest website in Germany (measured by Searchmetrics SEO Visibility).
• Largest winner of the July 2021 Google Core Update in Germany (according to Sistrix).
• 2nd largest winner of the November 2021 Core Update in Germany (according to Sistrix).
• 2nd largest SEO Visibility growth in Germany in 2021 (according to Sistrix).
• Largest organic traffic growth for an ecommerce site in Germany in 2022 (according to SEMrush & Digitaleffects).
• Owned multiple Objectives within company-wide OKR process.
• Developed models to forecast SEO traffic and EBITDA-contribution that met requirements from KKR and EY.
• Significantly contributed to litigation vs Google at the European Court of Justice.

Managed 23 FTE (27 ppl). Reported to the CMO.

Head of SEO

Searchmetrics

Leading the Product department (Product Management, Product Marketing, Quality Assurance, Project Management, User Research, and Technical Writing) of a B2B SaaS platform for Enterprise-level SEO and content marketing.

Reporting to the CEO/COO. Managing 12 FTEs.

VP Product

In charge of packaging & pricing, competitive intelligence, market analysis, sales enablement, internal product education, and product solutions. Led release-coordination and go-to-market for dozens of features.

Reporting to the CTO/Founder. Managing 2 FTEs.

Director Product Marketing & Solutions

Led go-to-market for a new product offering (packaging, pricing, competitive research, collateral, sales pitch, etc.) from 0 to multiple million euros in annual recurring revenue.

Presented at conferences and other industry events all over Europe.

Spent multiple months working in Silicon Valley and New York. Formed deep understanding of the US market.

Senior Product Marketing Manager

Product Marketing Manager

Heartbeat Labs GmbH

I supported Fernarzt, a B2C telemedicine service for prescription drugs, with their search strategy during a relaunch. Fernarzt was sold to investment group MARCOL in August 2020.

Search Strategy Consultant

kpunktnull – Beratung für das Digitale Zeitalter

Consulting work and project management in various industries: automotive, pharmaceutical, FMCG, food.
 • Dashboard for digital marketing KPIs.
 • Website relaunch of a global enterprise.
 • Digital media strategy for a Super Bowl ad.
 • Influencer marketing on Youtube.
 • Multiple digital strategies and social media strategies.

Digital Strategy Consultant & Project Manager

University of Münster

Individual contributer to various research project focussed on social network analysis and social media analytics. Projects were financed by/conducted for IBM, Microsoft, Vodafone, Federal Ministry of Education and Research.

Won an award for the best contribution from a young scientiest.

Research Group Communication and Collaboration Management

Landwehr Online Reputation Management

• Ran a bunch of websites
• Generated traffic (millions of visitors) via SEO, Social Media, and Content Marketing
• Built and operated multiple social media accounts with combined >100,000 followers
• Coordinated design, web development, and content creation work with various external freelancers and agencies
• Ran one the top 50 blogs in Germany for a few years

SEO & Blogger

seoFactory GmbH

I built and led the search engine optimization department of this internet marketing startup. I conducted mainly training for employees, trainees, and interns in Muenster and for the seoAcademy in Crete (Greece).

Co-Founder & Head of SEO

artaxo GmbH

I started the "Ranking Konzept SEO News" blogs. As the blog's chief editor I covered online marketing news and conducted interviews with online marketing influencers. We won an award for Best News SEO Blog.

Blogger

Malte Landwehr is the CPO & CMO of Peec AI, the leading solution for AI Search Visibility. And recognized by Business Insider as one of the fastest growing companies in Europe.

Malte is a former VP SEO, VP Product, and management consultant. He once built and ran one of Germany's 40 largest blogs.

Malte Landwehr has 20+ years of SEO experience and is a regular conference speaker and podcast guest. He has been recognized as a Leader (Venturi's Voice, Econsultancy), Visionary (Adello), and expert in Enterprise SEO & GEO.

Malte Landwehr

I am seeing more and more websites creating a .md copy of each article, adding an llms.txt file, creating /ai pages, or providing json files with metadata.

Do any of them work? No. Or rather, yes but.

I picked 5 domains that use some of these methods and crafted prompts specifically tailored to finding the information contained in these files. Sometimes with very explicit 20+ word quotes and the simple request to find the source of this quote.

1️⃣ llms.txt
Across almost 18,000 citation, there were 6 pointing to llms.txt files. That is 0.03% percent. Not nothing. But also not much.

The only case where this happened was a properly used llms.txt by the way. It contained information about how to use an API and where to find additional API documentation.

The kind of butchered llms.txt files recommended by GEO bros, where you are supposed to stuff in all your content received a total of zero citations.

2️⃣ .md files
Websites using .md copies of all their content were cited 3,500+ times. Zero of these citations pointed to the .md version. ZERO. 

The SEO in me would never allow such a level of content duplication.

There was one exception! 2% of Github citations pointed to .md files. But on Github, these are the regular URLs, they are linked internally, and there is no other way to access the content.

3️⃣ /ai and similar pages
For one website, 0.5% of citations pointed to such a file.

For a second websites, it was a whooping 16%. However, that was mainly because their /ai page contained significantly more information than their /about page.

And please keep in mind that I specifically asked for the information contained in this file.

4️⃣ Metadata json
For another brand, 85 out of 1800 citations (5%) came from a metadata json file.

Important: the file contained information I could not find elsewhere on the website. And I specifically asked for these pieces of informstion.

5️⃣ Summary
When asking questions about your brand, LLMs will sometimes reference llms.txt, /ai or similar files.

However, the same effect could be achieved with a well structured about page!

And these pages and files do not get any special treatment from LLMs. You could create a 12345.txt file and it would be cited if it contains useful and unique information.

Beyond online reputation management, and explaining AI agents how to use your API or MCP, llms.txt & Co have no practical SEO/AEO/GEO application.

ChatGPT is now the most blocked bot on the internet!

In 2020, I started a yearly habit of checking on BuiltWith, which bots are blocked most often across the internet.

This year, I have observed the biggest change ever.

Last year, there was 1 AI/LLM crawler in the top 10. This year, it is 6!

And none of the SEO tools are in the top 10 anymore.

Both our Series A announcement ($21 million) and our out-of-home campaign in Berlin led to hundreds of questions.

Since we are too busy filling 40 open roles over the next 6 months, we cannot find the time to answer every “what’s going on at Peec AI?” message.

We are doing a live AMA on Monday. We will talk about:
• Speed & Execution
• Product decisions
• How to get 1,400 paying customers in 10 month
• Why we moved the whole company to a Château in France for a week
• Organizing LAN parties for 100+ developers
• Why Marius is only average in League of Legends
• What it feels like to be the oldest person in the office (me)
• Anything you ask

If you’ve got questions about AI search – or just want the unfiltered version of the story – join us.

📅 Monday, 7 PM CET
👉 https://luma.com/si9uutf6

2023: Agencies dreamed of increasing their margins with AI.
2025: Agency insolvencies have doubled.

I’ve spoken to a lot of agency owners in recent months. Four patterns keep coming up.

1️⃣ AI efficiency gain is already priced in.
Pre-AI, updating titles and meta descriptions for the top 10 URLs took about an hour. Agencies hoped AI would let them do it in 5 minutes and still bill the full hour. Instead, clients now expect the output to jump from 10 pages to 25 pages for the same price. The “AI margin boost” never appeared.

2️⃣ Teams still lack broad AI skills.
Instead of running proven N8N workflows, juniors start from scratch with a fresh prompt every month. AI helps, but not nearly enough. They still need the full hour, sometimes more, to produce the expanded output clients expect.

3️⃣ Breakout agencies changed the benchmark for everyone.
These are typically small, founder-led agencies. They invest  20+ hours to build a real pipeline. Once set up, they press a button and update titles and descriptions for the top 10,000 URLs. Twice a week. Fully automated. Including QA with an LLM-as-judge. Their capabilities reset client expectations and increase pressure on everyone else!

These three observations are not limited to SEO. The same pattern shows up in image editing, ad creatives, CRM campaigns, translations, subtitles, and more.

4️⃣ Strategic blindspots
Lastly, in SEO specifically, the shift from SEO to GEO has created a split. Some agencies embraced AI search early and are adding multiple new clients per month. Others insist “it’s just SEO” - a false take that has been disproven again and again.

The same is happening with PR agencies. Some cry that consumers are no longer reading the content they produce for their clients’ blogs. Others have fully embraced the relationship between AI Search and owned/earned content.

What else is driving the surge in agency insolvencies? Which dynamics do you see in the market right now?

Want to be more visible in AI Search? Increase your Source Coverage!

LMM-based search and answer engines often perform web search (grounding) to generate answers. The websites they cite have an outsized influence on the generated answers.

Here is what you can do today:
1️⃣ Pick 10 relevant prompts where you would want your brand to show up in the answer.
2️⃣ Run each of them 5 times.
3️⃣ Each time, write down all the sources.
4️⃣ For the top 10 most-cited URLs, go and check if your brand is mentioned in there.

If you are not mentioned in the top sources, getting in there is your new goal!

If you get your brand added to at least 3 of the top 10 sources, you are almost guaranteed to see an increase in AI Search Visibility the next day.

Perplexity just increased usage of reddit as a source by 380%!

Here is the scoop:

For the last couple of months, reddit was pretty stable and showed up between 0.3 and 1.2 times per chat. Always depending on the industry of course.

On November 25, there was a spike by 70%. Suddenly reddit showed up 0.5 to 2.4 times per chat.

Then followed a drop by -75%. This brought the usage of reddit down to 0.2 to 0.5 times per chat. Significantly below the previously stable base.

10 days later, on December 6, there was another spike. Usage of reddit among the sources increased by 380%. We now see reddit 0.7 to 1.8 times per chat. A level 89% higher than the old base.

For some industries this means reddit is now 3x more dominant among the sources than it used to be!

There was also some movement around the usage of Google, Youtube, and LinkedIn in the same timeframe. But with a consistently negative trend.

I asked LLMs to toss a coin almost 3,000 times. Every single LLM failed to perform a random coin toss.

The results were:
• Heads 58%
• No result 32%
• Tails 6%
• Failed 4%

(Failed mostly means no AI Overview appeared. And once Perplexity replied with an empty answer.)

1️⃣ Head bias
All LLMs have a bias towards heads. This makes sense if you think about how the terms heads and tails appear in documents and how LLMs work.

The least amount of randomness happened in GPT-3.5-Turbo and Llama-3.3-70B-Instruct. Both replied with heads 100% of the time.

The most balanced were AI Overviews with only 72% heads and 28% tails. But AI Overviews did not show up at all in 39% of cases and refused to pick a side in 35% of cases. So really, we are talking 18% heads vs 7% tails when looking at all chats.

Ignoring refusals, ChatGPT, DeepSeek R1, and AI Mode all picked heads 81% to 87% of the time. Gemini was 90% heads and Perplexity 93%.

2️⃣ Refusal to make a guess
If we include refusals to pick a side, AI Mode (56%) was the worst, followed by Gemini, Perplexity, and AI Overviews.

ChatGPT was the most eager and picked a side in 94% of cases.

3️⃣ Prompts
The prompt itself has a huge impact. Two prompts with a similar task, can lead to vastly different results.

For example 61% heads vs 100% heads - from the same LLM.

How I asked, also changed the rate of unclear outcomes (no side picked) from 2% to 75%.

Big impact: “playing a game”, “tell me” vs “do you think”.

No measurable impact: changing the order of “heads or tails” to “tails or heads” in the prompt.

4️⃣ Additional observations:
Deepseek tried to introduce randomness by counting the number of words in my prompt and then deciding for heads or tails based on an even or uneven word count.

5️⃣ Disclaimer
I understand I can ask an LLM to write a Python script that gives a random answer. But that is no fun.

Things to consider when evaluating the importance of LLM-based answer engines vs traditional web search engines:

• CTR is 80% to 98% lower
• Even when LLMs send you less than 1% of traffic, they might be responsible for  20% of leads.
• While search engines primarily deliver clicks, answer engines facilitate decisions.

Do you remember Lily Ray's experiment about gullible LLMs?

She published an article with claims about the best SEO at building sandcastles, hula hoop, and similar nonsense. Lily then checked which AI search tools reproduced the fabricated information, revealing stark differences in how easily each model is fooled.

I have been monitoring those prompts in multiple LLMs since August. 11 prompts. 111 days. Time for a recap.

The most believable fact seems to be Jono Alderson's ability to eat spaghetti. In 77% of cases, LLMs mentioned him when asked “Who is the best SEO at eating spaghetti?”.

The least believable fact are Crystal Carter's karaoke skills. She was only mentioned in 21% of LLM answers. Sorry Crystal!

Funnily enough, Rand Fishkin was also mentioned in 21% of answers for the question “Which SEO would win a karaoke contest singing 90s pop?”.

The main culprit here are the Google-systems (Gemini, AI Overviews, and AI Mode). They often return a list of popular SEOs in their answers. Mostly Rand. But also Barry Schwarz, Neil Patel, Brian Dean, Danny Sullivan, and Marie Haynes. They did this especially often for the questions about karaoke, high fives, hula hoop, haikus, and trampolines.

Overall, ChatGPT is the least gullible LLM. Mainly because it does not directly search for “Who is the fastest SEO on roller skates?”. 

Instead, it creates fanout queries like “Who is the fastest SEO on roller skates fastest roller skater speed records roller skating speed record world fastest roller skater” or “fastest person on roller skates speed skating on roller skates top speed records for roller skating record speed on roller skates”.

And for these, Lily’s article is rarely used as a source.

1️⃣ Peec AI made it into Stripe City, Stripe's Black Friday campaign that features businesses using Stripe. We are in good company with Cursor, Lovable, Replit, bolt, ElvenLabs, Hugging Face, and Manus.

2️⃣ We are still hiring. Right now I am looking for:
• AI Search Evangelist:
Explain to the world why AI search matters.

• SEO & AI Search Strategist / Consultant:
Help our key clients and strategic partners to be successful in AI search.

• Customer Success Manager (SEO)
Enable our Enterprise customers to get the most value out of Peec AI.

This Wednesday, Marius Meiners will grill Lily Ray (Amsive), Ethan Smith (Graphite), and me about AI Search.

Tune in if you ever wanted to know if
• GEO is just SEO,
• what worked well in 2025,
• what common advice is wrong,
• and what to do in 2026.

Have your own questions? Drop them here in the comments!

https://luma.com/62sgwqck

graphical user interface, text, application

Grokipedia is about to become one of the top 10 wikis most often recommended and cited by LLMs.

I look at three different metrics:
• Visibility: Is Grokipedia recommended?
• Citation Share: How many wiki-related citations come from Grokipedia?
• Citation Rate: If Grokipedia is a source, how often is it cited?

1️⃣ Visibility
Grokipedia used to be the number 18 to 24 most recommended wiki in November. This improved to 11 to 17 most recommended wiki in December.

Grokipedia performs best in ChatGPT, where it jumped 13 positions and is now close to the top 10.

2️⃣ Citation Share
Grokipedia also improved a lot in ChatGPT when it comes to citations. It used to be responsible for 0.17% of wiki-related citations in November. This improved by 300% to 0.68% in December.

3️⃣ Citation Rate
Whenever Grokipedia makes it into the sources, it has a citation rate of 1.5. Meaning it is cited 1.5 times on average

Only WikiMD, New World Encyclopedia and EverybodyWiki achieved higher citation rates of up to 2.0.

For comparison, Wikipedia and Fandom have a citation rate of 1.0. The worst performing wikis sit at 0.5.

Charlie Marchant (Exposure Ninjas) challenged me to bring real-world examples of companies‘ AI search strategies.

I dugg into some of the research data-sets we are collecting at Peec AI and deconstructed how brands in the B2B space do it.

Examples. Playbooks. Everything.

Tomorrow we will discuss my findings in an exclusive webinar.

https://lnkd.in/ewzQZxNW

taplio

The Battle

Sacha Redler

Malte Landwehr

Followers

Posts

Engagements

Likes

Performance ScoreMalte Landwehr's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Sacha Redler

Malte Landwehr

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Malte Landwehr's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreMalte Landwehr's LinkedIn Performance