Your own agency as a reference project: try PR GEO yourself first

Why the trial run starts in-house

Agencies develop other people’s brands every day, and their own often gets neglected. A trial run with PR GEO reverses the perspective: you are the client, you supply facts, approve texts and wait for results. You quickly notice where a project stalls. Perhaps there are no up-to-date photos, perhaps coordination between two founders takes longer than expected, perhaps your own positioning is less clear than it looks on the website.

These experiences are valuable because clients will later get stuck at the same points. If you have been through the process yourself, you calculate more realistically, write better briefings and anticipate questions. In a sales conversation, you then talk about something you know, rather than about a service from a brochure.

A trial run needs a clear framework: a clearly defined topic, a manageable publication plan and a measurement before and after the articles. Around 90 days is a sensible timeframe. Our secondary study divides it into four phases, which the following steps are based on.

Treat the trial run like a real client project. One person in the team is responsible for it, there is a written briefing and a budget of hours, and the founders or management approve texts just as a client would later. Otherwise, your own project runs on the side and gets postponed at the first client request.

Step 1: Record the starting point

Before a single article appears, you document how your agency currently shows up in search and AI answers. Without this snapshot, there will be nothing to compare later.

Start with the technical side. Check whether your website is accessible, whether the most important pages are in the index and which crawlers your robots.txt allows.

For your trial run, this means: if your website should be accessible for search in ChatGPT, OAI-SearchBot must not be accidentally excluded. You decide on GPTBot separately and deliberately.

Next, compare your agency’s facts everywhere they appear publicly: website, industry directories, company profiles, social media accounts. Do the name, services, address and contact persons match everywhere? Whatever you correct here before the first articles appear, you will not have to laboriously factor out of the evaluation later.

Then comes the question set. Write down around 20 questions that potential clients really ask, for example “Which design agency in Leipzig designs packaging for small food producers?” or “What does a corporate design for a young company involve?” Ask each question in the systems that matter for your target group, such as ChatGPT, Gemini, Perplexity, Microsoft Copilot and the AI Overviews in Google Search. The AI visibility self-check gives detailed instructions for this first round.

For each answer, enter the same details in a log:

Count mentions and source links separately. A system can mention your agency without referring to your pages, and vice versa.

Step 2: Topics from your own positioning

The topics of the trial run should show what your agency wants to stand for, not everything it can do. Ask yourself: for which client question do you want to be seen as the go-to expert a year from now? That gives you two or three articles, not a list of services.

Productive sources for topics like these are:

• questions that come up again and again in first meetings
• misunderstandings that you regularly explain to clients
• your way of working, for example how a workshop or a brand development process runs
• completed projects whose presentation the client has approved in writing
• a well-founded assessment of a development in your industry

Check every draft against the questions Google sets out for self-assessing helpful content³: does the text offer original information or original analysis? Are the sources clear? Is it evident where the expertise comes from? General phrases such as “holistic brand management” do little here, whereas a concrete process with traceable decisions does a lot.

Step 3: Publish and keep track of the effort

Now the articles are published. Whether you organise them individually or book them through a package is secondary for the trial run, and our comparison of individual booking and packages describes the pros and cons. Documentation matters more than quantity. For each article, record:

1. How many hours did your agency put into the topic, briefing, coordination and approval?
2. How many rounds of corrections were there, and why?
3. Which medium was chosen, and why does it suit your target group?
4. When did the article appear, at which address and as which type of publication?
5. Is it recognisable as a paid article, if it is one?

These notes later form the basis of your costing. The article on the coverage report with AI observations shows how records like these can become a stand-alone evaluation module.

Step 4: Measure afterwards without reading too much into it

After a few weeks, ask the same questions again: in the same system, with the same wording, in the same language and region. Only then can the results be compared. A simple metric is the mention rate: answers that mention your agency divided by all valid answers in the test set. Always state the absolute numbers as well and how often you repeated each question.

What you can show clients afterwards

At the end of the trial run, you have more than an experience. You have material that holds up in client conversations:

• the actual process with all steps and roles
• the real hours per article as a basis for your costing
• published articles as work samples for topic choice and writing
• your log template for measuring before and after
• an honest list of what went well and what you would change

What you should not show is a success story that claims more than your data supports. Even if your agency is mentioned more often after the test, nothing follows from this for a client’s brand in a different industry. Say so openly. It is precisely this restraint that distinguishes advice from an advertising promise.

For sales conversations, a one-page summary with five points works well: starting point, approach, internal effort, observed changes and the limits of what can be concluded. That way, you and the prospective client can see together what the process requires, and they can judge for themselves whether they want to take on this effort for their brand.

The article PR GEO for advertising agencies describes how to fit the service into your portfolio afterwards and clarify responsibilities. If you would like to start the trial run with a ready-made package, you will find the three pr4geo packages on the offers page.

Frequently asked questions

How long should a trial run with your own agency take?

Plan for around 90 days: two weeks for the starting point, two weeks for topics and material, about a month for publications and a month for the evaluation. This split follows the 90-day plan from our secondary study and can be adapted to your agency’s day-to-day work.

Do I have to book a package for the trial run?

No. You can organise individual publications yourself or use a package. What matters is that you document effort and observations equally carefully in both cases.

Can I use the results as a reference?

Yes, if you present them factually: starting point, approach, measurement method and observed changes. Do not claim a cause you cannot prove, and do not derive any promise for clients from it.

What do I do if nothing has changed after 90 days?

You still know how much effort the process takes and where your own content has gaps. Check whether facts are missing, whether the topics match the questions and whether the media reach your target group. Even a result without change is an honest basis for your consulting.

Sources

2. OpenAI: Overview of OpenAI Crawlersaccessed 9 November 2026
3. Google Search Central: Creating helpful, reliable, people-first contentaccessed 9 November 2026

Next step

For your agency or a client project

pr4geo delivers editorial work, publication and reporting. You keep the client and the strategy.

Book pr4geo for your agency or a client projectSee the three pr4geo packages

Provider disclosure: pr-geo.com is a knowledge portal of NexTao GmbH. pr4geo belongs to our group, which is why we expressly recommend it here. No package guarantees a mention in AI answers. Our principles