Case study / Corporate investigation and risk
Ten Websites, One Entity: GREVESGROUP
A corporate investigation group had grown into ten separate websites, each built at a different time by different hands. The brief was not to rank higher. It was that the network had become hard to read, for people and for machines.
At a glance
- Client
- GREVESGROUP, named here with their permission
- Sector
- Corporate investigation, risk and brand protection
- Markets
- India and the UAE, serving matters across more than 125 countries
- Engagement
- Ongoing, across the group network
- Our scope
- Audit, structured data, content, internal linking, AI discoverability, internal tooling
42/100
The score our audit gave grevesgroup.com at the start. Our own scoring model, not a third party tool.
10
Websites in the group network that the engagement now covers.
39
Pages of corporateinvestigators.com given a new structured data graph in August 2026.
243
Approximate total pages on that one site, of which roughly 200 are country pages.
These are counts of work delivered and one score from our own model. None of them is a performance claim. What we have not measured is set out further down this page, in its own section, rather than left out.
What the group arrived with
GREVESGROUP is a corporate investigation and brand protection group. Over the years it had grown the way successful firms often do, by adding a venture at a time. Each division ended up with its own website: the group site, a corporate investigations site, a process serving site, a risk investigations venture registered in Dubai, and others. Ten in total.
Each site had been built at a different time, by different people, on different foundations. One of them is not a content management system at all but a hand built HTML and PHP site of roughly 243 pages. They worked. They just did not agree with each other.
The request that started this was not about rankings. It was closer to a complaint: the group had become hard to describe. A law firm landing on one site got a different account of who GREVESGROUP is than a bank landing on another. Search engines had the same problem, and so, increasingly, did AI assistants being asked about the firm.
One more thing shapes this engagement and is worth stating plainly. Swaranjit Singh worked with this group for years before PIXELS TRAIL took it on as an agency client. That history is why the scope reaches into places a search agency would not normally go, including a case management tool for the investigations team that has nothing to do with marketing.
What the audit actually found
We scored grevesgroup.com at 42 out of 100 on our own model and delivered it as a ranked report. The score mattered less than the three problems underneath it, which were structural rather than cosmetic.
The group described itself differently on every site
There was no single description of the organisation, no shared address or identifiers, and no agreement on what the group was called in the places machines read. Ten sites meant ten versions of the same company.
The structured data described the wrong business
On corporateinvestigators.com the existing markup described a kind of business the site is not. Structured data that is wrong is worse than none, because it is confidently wrong, and it was being served on every page.
Roughly 200 country pages had nothing at all
Country pages make up about 82 percent of that site, and they carried no structured data and no consistent linking between them. They were the largest part of the site and the least legible part of it.
The constraints we had to work inside
Most of the interesting decisions on this engagement came from limits rather than ambitions. These are the ones that shaped the work.
- ▶A static site, not WordPresscorporateinvestigators.com is hand built HTML and PHP across roughly 243 pages. No plugin could do any of this. Every change had to go into the markup itself.
- ▶Ten separate codebasesThe network is not one installation with ten sections. Consistency had to come from a written specification applied by hand, not from shared software.
- ▶Clients who cannot be discussedAn investigation firm cannot publish client names or case detail. The content had to earn credibility without a single example of work, which rules out the usual approach.
- ▶Jurisdiction accuracy matters more than volumeEnforcement and process serving genuinely differ country by country. Content that generalised would be wrong, and being wrong in front of law firms costs more than being absent.
- ▶Improve, do not replaceThere was no budget or appetite to rebuild ten sites, and no good reason to. The work had to raise what already existed.
What we built
The engagement is ongoing and runs across several workstreams rather than as one project with an end date. This is what has been delivered so far.
An audit that ranked the problems
grevesgroup.com reviewed and scored against our model, delivered as a client ready report with issues ordered by what they would cost to leave alone rather than by how easy they were to fix.
One entity graph across the network
A new structured data graph written for the 39 main HTML pages of corporateinvestigators.com, replacing markup that described the wrong kind of business, so the group now presents one consistent identity.
Schema and answer engine work on a second division
Structured data, Open Graph and answer engine optimisation delivered for indianprocessservers.com, a group division with its own audience of law firms.
An internal linking strategy for severg.com
Related investigation services now pass authority to each other instead of sitting as orphan pages that only the sitemap knows about.
llms.txt and llms-full.txt across the network
Machine readable descriptions of the group written for all ten websites, so an AI assistant reading the sites gets a stated account of what the group does rather than inferring one from menu labels.
Ten TSCM industry pages
Technical surveillance countermeasures written for IT, legal, financial, pharmaceutical, manufacturing, healthcare, real estate, retail, telecom and energy. Around 1,000 words each, on a deliberately strict two to three heading structure.
A full content set for the Dubai venture
riskinvestigators.com written from its sitemap: home, about, eight services, 43 country pages, 14 industry pages, FAQs and contact. This one is in delivery rather than finished.
IP and brand protection service design
A three tier service structure with financial modelling, priced in INR and USD, plus an enforcement rate card that separates Indian from international action and weights by the kind of action required.
A case management tool for the investigations team
An Excel build with dynamic dropdowns, RAG status tracking and investigator workload views. Used by case managers, not by marketing, and included here because it was part of the same engagement.
What we decided not to do
These are the choices worth more than the deliverables, because each one was the option that would have earned us more work or looked more impressive in a proposal.
- ✕We did not consolidate ten sites into oneOn paper it is the obvious recommendation, and it would have been a far bigger project for us. But the divisions serve different buyers in different jurisdictions, and several carry their own search history and reputation. Folding them into one domain would have thrown that away to make our job tidier.
- ✕We did not patch the existing structured dataThe markup on corporateinvestigators.com described the wrong kind of business. Correcting it field by field would have been quicker and would have left the wrong shape underneath. We replaced it instead.
- ✕We did not rewrite 200 country pagesThe existing country pages were kept as they were. They did not need new prose, they needed structure and linking, so that is where the budget went. Rewriting them would have been easy to bill and hard to justify.
- ✕We did not chase volume on the TSCM pagesTen pages at around 1,000 words, not thirty thin ones. Technical surveillance countermeasures is a small and specific market, and padding it out would have produced pages no buyer would finish reading.
- ✕We did not add a case studies section to their sitesFor an investigation firm, publishing case detail is a liability rather than an asset. Credibility there had to come from jurisdiction accuracy instead, which is slower to build and does not screenshot well.
What changed
Stated as changes to the sites themselves, because those are the things we can point at. Commercial outcomes are the client’s data and are dealt with in the next section.
- ✓The group reads as one organisationThe same description, address and identifiers now appear across the network, instead of each site offering its own account of who the group is.
- ✓The country pages sit inside a structureRoughly 200 pages that previously carried no structured data at all are now part of a coherent graph, with linking between related pages.
- ✓There is something for AI assistants to readTen sites now carry llms.txt and llms-full.txt. Whether that changes how often the group gets cited is addressed below, honestly.
- ✓Ten industries have a page that did not existThe group served these verticals and had nothing to send a prospect in any of them. Now it does.
- ✓The investigations team works off one toolCase tracking moved from ad hoc files to a single build with status and workload visible in one place.
- ✓The service line has a defensible priceIP and brand protection went from quoted case by case to a tiered structure with a rate card, which is faster to sell and harder to argue with.
What we have not measured, and will not claim
Case studies usually end with a number. This one does not, and the reason is simple: we do not hold the data that would make the number honest. Here is exactly what is missing, so you can judge the work on what is actually shown.
- ⚠We have not published ranking or traffic figures for this engagement. We do not hold the analytics and Search Console history for every site in a ten site network, and reporting a figure from one site as though it described the network would be misleading.
- ⚠The 42 out of 100 is from our own scoring model. It is a useful before and after against that same model and it is not comparable to any third party tool's score. Treat it as our opening assessment, not an industry rating.
- ⚠We have not measured whether the llms.txt files changed how often AI assistants cite the group. This is genuinely hard to measure well, and anyone quoting you a precise percentage for it should be asked how they got it.
- ⚠Lead volume and revenue are the client's figures, not ours. We have not asked permission to publish them, so they do not appear here.
- ⚠The riskinvestigators.com content set is in delivery rather than complete. Read it as scope, not as a finished result.
- ⚠We cannot show you the audit report here. It names and ranks problems on a live client site, which is not ours to publish. We can walk you through how it is built.
How to check this yourself
Most of this engagement is public, which means you do not have to take our word for any of it. Four things you can confirm in a few minutes.
01
View the source of any corporateinvestigators.com page
Open the page source and search for the string application/ld+json. You will find a graph describing a corporate investigation business. What was there before described something else.
02
Add /llms.txt to the group domains
The files are public. One practical note: several of these sites refuse automated requests, so open the URL in a normal browser rather than fetching it with a script or a command line tool, or you will get an error that has nothing to do with whether the file exists.
03
Search for a TSCM industry phrase
The ten technical surveillance countermeasures pages are live and indexable. Search for the combination of TSCM and one of the ten industries listed above.
04
Ask us to walk through the audit
We will not email you another firm’s audit. We will share the structure, the scoring model and how issues were ranked, on a call, which is the part that is actually useful to you.
What we would do differently
The llms.txt files went out before we had any way to tell whether they made a difference. If we started this again we would record a rough baseline first, by asking several AI assistants to describe the group and saving the answers, then repeating it afterwards. It would not be a controlled experiment, but it would be better than the nothing we have now.
The strict two to three heading structure on the TSCM pages did its job, which was consistency across ten pages written over several weeks. It also made a few of them read tighter than they needed to. Consistent structure does not have to mean an identical rhythm, and we applied the rule more mechanically than the content wanted.
The structured data on the static site was applied page by page. A small script generating the graph from one data file would have taken about a day and would have paid for itself by the third round of changes. We learned that on roughly the thirtieth page.
We also took on the Excel case management tool because we were asked and because we could. It was the right thing for the client and it sat oddly inside a search and content engagement. If we were scoping this again it would be a separate piece of work with its own brief.
The services this engagement drew on
If one part of this is closer to your situation than the rest, these pages go into detail on method and pricing.
Other case studies
Where this client relationship started
The 2014 identity and website rebrand for the same group, back when it was running a clipart logo.
The same work, on our own site
We did this to ourselves first. Every claim on that page is checkable in your browser.
An eight year white label partnership
A Dallas marketer sold, we delivered. More than 150 small business websites between 2013 and 2021.
Questions people ask about this one
Why is GREVESGROUP named when your other case studies are not?
Because they agreed to it. The other case studies on this site describe the work without naming the client, because those clients did not give permission. We will not name a client without it, including when naming them would obviously help us win work. If you become a client, the same applies to you.
Where is the traffic graph?
There is not one, and the reason is in the section above on what we have not measured. We do not hold complete analytics and Search Console history across all ten sites, so any single chart we showed you would be describing one site while implying it described the network.
What is TSCM?
Technical surveillance countermeasures: the practice of detecting and removing covert listening devices, hidden cameras and other surveillance from offices, boardrooms, vehicles and homes. It is a specialist corner of corporate security, usually bought before sensitive negotiations, after a leak, or on a scheduled sweep for executive areas.
Does llms.txt actually do anything?
It is an emerging convention rather than a standard that any AI company has committed to following, and nobody should tell you otherwise. What it does reliably is give you one place where your own description of your business is stated plainly, which is useful on its own and costs very little to maintain. We wrote these files for the group on that basis, and we have been clear above that we have not measured a citation effect.
Can you work on a static HTML site, or do you need WordPress?
Static is fine. Roughly 243 pages of this engagement are hand built HTML and PHP with no content management system, and the structured data went straight into the markup. Static sites are often easier to work with than a heavily plugged in WordPress install, because there is less fighting you.
We have a group of sites in a similar state. Is this transferable?
The method is: agree one description of the organisation, express it as structured data, apply it consistently, then fix the linking so the pages that already exist stop being orphans. What is not transferable is the sequence, which depends on which of your sites carries the most commercial weight. That is the first thing worth working out, and it usually takes a conversation rather than a tool.
Who actually did the work?
Swaranjit Singh, who runs PIXELS TRAIL, with support on production. This is a small studio, not a team of account managers, so the person who audits your site is the person who writes the schema.
Can we speak to GREVESGROUP about working with you?
Ask us and we will pass the request on. We will not volunteer a client’s time or hand out their contact details, so the answer depends on them rather than on us.
How long did this take?
It is an ongoing engagement rather than a fixed project, which is why there is no duration here. Individual pieces had their own timelines: the audit was a defined deliverable, the schema rebuild across 39 pages was a few weeks around August 2026, and the content sets run longer.
What would something like this cost?
It depends almost entirely on how many sites and pages are in scope, and whether the sites can be changed through a content management system or have to be edited by hand. The service pages linked above carry indicative pricing in INR with USD equivalents. For a network engagement the honest answer is that it needs a scoping conversation first, because quoting it blind would mean padding the number to cover what we had not seen.
Got a group of sites that do not agree with each other?
That is the problem this engagement solved, and it is more common than it sounds. If you have grown by adding ventures and now have several sites describing the same company differently, tell us what you have and we will tell you what we would look at first.