MetroMinds explains what it does within seconds: software that helps leadership teams decide which AI projects to fund. What lets it down is evidence. The benchmarks cite the company’s own library, the site never says who is behind it or who uses it, and the numbers change from page to page. Fix those first, then shorten the route to a demo.
Keep the positioning, the public pricing and the free tools. Fix the benchmark sources, the proof and the numbers first.
Based on the public site as it stood on 26 October 2026.See how Radence can help →A good product story let down by weak evidence. What works, what doesn’t and what to do first.
The site says clearly what MetroMinds is for: helping leaders decide which AI projects to fund. Public pricing, pages written for each executive role and free tools that give a result without a form all take a buyer a long way before they talk to sales.
MetroMinds sells numbers you can defend, but its own numbers don’t hold up well. The benchmark ranges cite the company’s own library, there are no named people or customers, counts disagree between pages, the homepage promises 4,980 use cases and shows 100, and the About page claims EU hosting more firmly than the privacy policy does.
Every page in the main sitemap (89), the three legal documents, all five free tools and 15 of the 100 public use-case pages. What we say about use-case pages is based on those 15.
The Business Case diagram also leaves out the minutes-to-hours step (finding 06). The calculator itself is correct.
Open the findings ↓We started with the first screen only: headings, images and buttons. We wrote down what we thought MetroMinds sold, then read the rest of the site. Five of our assumptions were wrong.
A headline about deciding which AI investments deserve budget, a one-sentence subhead and four short bullets. The primary button opens the use-case explorer; the secondary button books a demo. The large image is a single insurance use-case record with a benchmark table.
/Sells: software for managing a portfolio of AI initiatives, built around a benchmark library.
Helps: executives who hold AI budgets, such as the CIO, CFO, COO and CHRO.
Why buy: fund the right ideas using comparable scores and benefit ranges.
The image and the main button both point at the library, so what stuck was “a database of AI use cases with some portfolio tools on top”.
| First assumption | What the site actually says | Where |
|---|---|---|
| All 4,980 use cases can be browsed. | 100 are public. The full library is for customers. | /tools/use-case-explorer |
| Benchmark ranges are measured results from implementations. | Use-case records say the ranges come from comparable use cases in the MetroMinds library itself. No method page explains further. | /library/banking/keeping-commercial-client-files-current-for-bank-credit-reviews |
| It is software only. | Services are a large part of the offer: Expert Start costs €15,000 (Foundation) or €30,000 (Scale, two modules required), and workshops cost €1,250–€2,250 each. | /pricing |
| The AI register is a record-keeping form. | It also analyses the customer’s application portfolio to find AI inside vendor software. The inputs or integrations this needs are not explained. | /platform/ai-register |
| The free Readiness Snapshot is the readiness product. | It is a six-question preview of a 40-question scan that several colleagues answer inside the platform. | /platform/readiness |
This is our own first reading. Section 03 suggests a quick test to see whether real buyers read it the same way.
We looked at the site the way a busy buyer would: headings, images and buttons first, paragraphs second.
What we saw: The homepage explains a funding decision and shows a single use-case record. The main exploration button draws attention to the library. Visitors also need to see what happens after finding an idea.
Why it matters: A visitor may understand the category yet remember a database rather than a system for comparing and managing investments.
What to change: Lead with a portfolio view and annotate three actions: compare initiatives, build a business case, agree what to fund. Show one idea progressing into a decision.
Seen on the site · visitor interpretation needs testing
Homepage ↗ · Platform ↗What we saw: The reviewed homepage contains roughly 1,123 words excluding global navigation; the CIO page roughly 1,141. Product explanations, sector material, tools and articles compete for attention.
Why it matters: Busy buyers have to read several sections before the offer makes sense. The screenshots show the interface but don’t explain what a buyer would do with it.
What to change: Give each section one job: a heading that answers the visitor’s question, one annotated screenshot and one next step. Move the detail behind clear links.
Seen on the site
Homepage ↗ · CIO journey ↗What we saw: All five tools are useful without signing up, and the tools page links them into one process. On the Readiness Snapshot result, the next steps are a request for detailed advice (name, work email and organisation) or a demo. The result does not preview the paid readiness view or link to matching use cases.
Why it matters: A good free result becomes a dead end if the visitor can’t see what the paid version would add. We checked this on the Snapshot only.
What to change: Add a three-column “Free tool / Paid platform / Next step” explanation. At each result, show the matching platform view and a relevant demo instead of a general request to contact sales.
Seen on the Snapshot result
Tools hub ↗ · Readiness Snapshot ↗Two more first-impression problems, missing proof and a button that promises more than it delivers, are covered in findings 02 and 04.
Show five target buyers the homepage with the paragraphs hidden. Ask what MetroMinds sells, who it is for, what happens inside the product, how the free tools differ from the paid platform and what they would click next. If most can’t answer, the headings and images need more work.
We followed three typical routes through the site.
What works: The funding-decision headline gives an immediate reason to care. The explorer provides something concrete to try.
Where it stalls: The first screen makes MetroMinds look like a library. The portfolio and decision tools come much later.
What to change: Show a complete decision workflow above the detailed sector and article content. Keep a clear product route beside the free-library route.
What works: The CFO page uses relevant language, and first-year costs are public.
Where it stalls: Pricing is well explained but the proof isn’t: no method, no team, no customer results. It also takes effort to work out what comes with the software and what costs extra.
What works: Visitors get a useful result before being asked for personal details. Two different answer sets both produced an immediate result.
Where it stalls: The result offers generic advice by email or a demo. It does not show how the paid readiness scan would extend this result, and the industry list on the advice form leaves out Banking and Insurance.
What to change: Show “Your snapshot now” versus “Your team’s platform assessment”, a relevant example and a targeted readiness walkthrough. Confirm the paid features with MetroMinds first.
Six things to keep through any redesign.
The main message is about deciding which AI investments deserve funding. That is more specific than a general promise to “transform with AI”.
View source pageAnnual licences, optional onboarding, workshop prices, VAT exclusions and first-year examples help buyers judge whether the offer suits them.
View source pageAll five free tools give a real result before asking for an email address or a meeting.
View source pageCFO, CIO, COO and CHRO pages connect the platform to financial, governance, operational and people decisions.
View source pageThe calculator shows its scope, adoption and confidence settings, and says plainly that saved hours only turn into money if the business does something with them.
View source pageThe guided tour works, the interactive demo loads on click and demo companies are labelled as fictional. The design is consistent and reads well on a phone.
View source pageWhat works and what doesn’t, page by page. Each row links to the page so you can check it; the page wording is paraphrased.
| Page | Verdict | What the page shows | Why it matters |
|---|---|---|---|
| / | Good | The headline frames a funding decision for people who hold AI budgets. | The category and buyer are clear within seconds. |
| /pricing | Good | Public plan prices, first-year examples and a statement that realised financial benefits are not assumed within three months. | Buyers can qualify themselves before a demo. |
| /tools/ai-savings-calculator | Good | The full formula, including scope, adoption and confidence, with a worked example labelled as illustrative. | It invites scrutiny instead of asking for trust. |
| /tools/ai-readiness-snapshot | Good | A score and weakest domain appear before any form. | Visitors get value before giving details. |
| /tools/eu-ai-act-check | Good | The Annex III flag is described as a starting point, not legal advice, and the one-in-fifteen screen is disclosed as based on titles. | An honest caveat. It should travel with every use of the one-in-fifteen figure. |
| /for/cfo | Good | Built around the questions a CFO actually asks about an AI business case. | Strong message fit by role; the CIO, COO and CHRO pages follow the same pattern. |
| /library/banking/keeping-commercial-client-files-current-for-bank-credit-reviews | Needs work | The benchmark footnote says ranges come from comparable cases in the MetroMinds library. | It contradicts the real-world-implementations claim elsewhere (finding 01). |
| /about | Needs work | About 10% of the library described as public; no team, legal entity or location; EU hosting stated without qualification. | Public share is really 2%, and the hosting line is stronger than the privacy policy (findings 02, 04, 05, 19). |
| /industries/manufacturing | Needs work | 491 and 496 use cases on the same page; 55 and 56 nonconformance cases. | Undercuts a product that sells reliable numbers (finding 03). |
| /industries | Needs work | Transport & Logistics (495) called the largest industry. | The manufacturing matrix totals 496 (finding 03). |
| /industries/insurance | Needs work | Most insurance use cases said to be in case and exception handling. | The page’s own chart shows 24 of 84, or 29% (finding 08). |
| /pricing | Needs work | The list of included capabilities does not name the AI register or full library access. | Buyers cannot confirm the EU AI Act module is in their plan (finding 09). |
| /services | Needs work | Workshops may be delivered by certified partners. | The partner programme opens in January 2027 (finding 10). |
| /library/education/from-a-raw-power-trace-to-a-cleared-school-field-device | Needs work | Internal calculation names in the business explanation. | Records read as unedited system output (finding 12). |
Open a finding to see what we saw, why it matters, what to change and how to check it yourself.
P1 fix first: accuracy and trust · P2 fix next: clarity and conversion · P3 tidy up. The priorities are our judgement; we had no data on enquiries or revenue.
20 findings shown
Source captions on library, industry, tool and article pages say ranges are based on real-world implementations. The footnote on sampled use-case records reads: “Ranges come from comparable use cases in the MetroMinds Library.” No page in the main sitemap explains source types, sample sizes, selection, dates or review. Article source lists name external publications but do not link to them.
A buyer cannot tell measured outcomes from expert estimates or reusable planning assumptions, and the two statements point in different directions. For a product that sells defensible numbers, this is the first claim a sceptical CFO will test. It is a disclosure gap, not proof that the ranges are wrong.
Publish a methodology page covering source types, number of implementations, dates, how ranges are set, who reviews them and known limits. Replace the record footnote with the real source for each range, and link external article references.
Compare the source caption on /platform/use-case-library with the benchmark-range footnote on any public use-case record. No private methodology was available to this audit.
The homepage, About and pricing pages do not supply named leadership biographies or verifiable customer outcome case studies. A search of all main-sitemap pages found no customer names, testimonials, case studies, logos, certifications or named authors. The legal pages identify OmniMinds Technologies B.V.; demo organisations are correctly described as fictional.
Someone weighing a €17,500-a-year commitment has nothing to go on about the team’s experience or what customers got out of it, and so little to take to colleagues when suggesting a demo.
Add named experts, relevant experience and customer stories shared with permission. Show baseline, scope, measured outcome and limitations. If customer names cannot be disclosed, use clearly labelled anonymised cases with a verification route. Do not invent testimonials or logos.
Review About and the public buying journey. Absence of public proof does not establish absence of customers or expertise.
Manufacturing uses 491 in its introduction and charts, but 496 in its headline counter and matrix, and gives both 55 and 56 nonconformance cases. Government uses 490 and 491 on its landing page; the government article uses 487. The industry overview’s 13 counts sum to 4,952 rather than the advertised 4,980.
The industry overview calls Transport & Logistics (495) the largest industry, while the manufacturing matrix totals 496. The demo portfolio shows 13 accepted use cases on /platform and 11 on /platform/portfolio.
MetroMinds is selling reliable, comparable numbers, so its own numbers have to agree. Some gaps may come from different data snapshots, but the pages don’t say so.
Publish one versioned source of truth for counts and derived percentages. Label any excluded records and historical snapshots. Update narrative copy, charts, directories and widgets together.
Compare the linked pages and their August 2026 source captions, and add up the counts on the industry overview.
The About page describes about 10 percent of the library as public. The library page and explorer state 100 public cases out of 4,980. The homepage’s “Explore 4,980 use cases” button lands in that public sample.
Visitors may expect access to the full collection and discover a smaller sample. The explorer does disclose its limit, but the entry promise should match.
Correct About’s percentage. Label the button “Browse 100 free use cases” and explain that the complete 4,980-case library is available with a platform plan, once MetroMinds confirms which plans include it.
Compare About, the library page and the destination of the homepage exploration button. 100 ÷ 4,980 ≈ 2.01%.
About says the service is hosted in the EU with a data processing agreement. The privacy policy says OmniMinds seeks to keep primary hosting and databases in the EEA, and that some support, email, security or infrastructure services may process data outside it.
The data processing agreement promises encryption at rest only where the hosting or storage provider supports and enables it, and makes the sub-processor list available on request. Multi-factor sign-in for portfolio managers is mentioned only on the CIO page, not in the agreement’s security measures. No security page, single sign-on statement or ISO 27001 / SOC 2 reference was found.
Enterprise security reviews compare marketing claims with contract documents. An overstated hosting line and hedged controls invite a long questionnaire before a demo, or an early no.
Make About match the privacy policy and name the provider and region. Publish a short trust page covering hosting, sub-processors, encryption, multi-factor and single sign-on, backups, testing and any certification plans, each consistent with the agreement.
Compare /about with section 5 of the privacy policy and Schedules 2 and 3 of the data processing agreement. This reviews public statements only; security controls themselves were not tested.
The diagram under “The same logic for every use case” multiplies annual volume, minutes per item, benchmark reduction and hourly cost, followed by scope, adoption and confidence. It does not convert minutes to hours. The calculator’s displayed formula does, and its results matched our independent check.
A buyer copying the diagram literally would get a value 60 times too large: 60 minutes of work at €50 an hour costs €50, not €3,000. That undermines the promise of defensible financial evidence.
Add ÷ 60 after minutes per item, or change the input to hours per item. Use the same formula wherever the method appears.
Compare the diagram on the business-case page with the calculator formula. This is an explanation defect; the calculator arithmetic passed.
Headlines such as the claim that half of all AI ideas end up on the COO’s desk are based on the library’s 4,980 records. The library’s shape looks designed rather than sampled: all twelve AI-domain counts are multiples of 60 (from 780 down to 240), and there are exactly three benchmark metrics per use case (14,940 = 3 × 4,980). The 147.5% claims-handler overlap is the sum of MetroMinds’ own benchmark midpoints.
A designed library can be useful, but its proportions describe how it was built, not where organisations actually place AI. Buyers who notice this may discount the research content as a whole.
Phrase headlines as library findings, for example “In the MetroMinds Library, half of the use cases sit in operations”. Explain how the library was assembled and whether its proportions are meant to reflect the market. Present the 147.5% example as a demonstration of the overlap check, not as market evidence.
Add up the domain counts on /platform/use-case-library and compare the metric total on /platform/business-case. That the library was designed rather than sampled is our inference; MetroMinds can confirm.
The insurance page says most insurance use cases sit in case and exception handling; its own chart shows 24 of 84 (29%), with 45 in other capabilities. The banking page says its use cases are mostly in exception handling, risk assessment and financial risk management; those three total 23 of 81 (28%).
Small wording errors are costly in a product that sells precision, and they are easy for a buyer to check on the same page.
Use the numbers: “Case and exception handling is the largest single group: 24 of 84 insurance use cases (29%).” Apply the same rule to banking.
Compare each claim with the capability chart on the same page.
Pricing and its FAQ list use-case management, prioritisation, business cases and value, readiness, capability and workforce impact, and results tracking. The AI register and EU AI Act module, promoted as one of six platform modules, is not named, nor is access to the full 4,980-case library. Expert Start for Scale is described as two required modules, although Expert Start itself is optional. The FAQ question about larger rollouts is answered with Expert Start details.
A buyer drawn in by the EU AI Act content cannot confirm it is in the plan without asking sales.
List every module and library access per plan. Explain what one Expert Start module covers. Answer the larger-rollout question with Enterprise terms. Consider showing that Expert Start (€15,000) equals the eight workshops at list price, plus a second Decision Board and progress calls.
Compare the six modules on /platform with the included list and FAQ on /pricing. The eight workshop prices on /pricing sum to €15,000.
The Services page says Expert Start and workshops are delivered by MetroMinds or by a certified practitioner from its partner network. The Partners page says the programme launches in January 2027 and is collecting registrations of interest.
A buyer asking who will run their onboarding may find that the network does not exist yet.
Until launch, say workshops are delivered by MetroMinds and that certified partners will be available from 2027.
Compare the practitioner note on /services with the launch date on /partners.
Selecting Functional Readiness, score 2, in all six dilemmas returns the correct overall 2.0/5 and six correct domain scores. The explanation nevertheless names Governance as both the strongest and the weakest domain. Reproduced again on 26 October 2026.
A correct score is paired with advice that appears contradictory, which can reduce confidence in the assessment and its follow-up recommendation.
Detect ties. Say that readiness is even across the six domains and offer a way to choose where to start. If only some domains tie, name the tied set.
Start a new Snapshot, choose level 2 in every dilemma and view the result. No contact details are needed.
Sampled records include transaction_based_time_saving, fte_based_effort_reduction, annual_volume, baseline_minutes_per_item and fully_loaded_hourly_cost in customer-facing prose. Some passages read like generation instructions, use US terms such as K-12 for an EU-focused product, or repeat a long generic account of value, oversight and monetisation.
Business buyers must decode field names before understanding the benefit, and the records read as unedited system output.
Use plain labels and a short assumptions table. Lead with the business situation, what AI prepares, who decides and the measurable benefit. Move technical model names into optional method details. Edit the 100 public records by hand first; they are the shop window.
Read “Where the value comes from” in the sampled automotive, banking and education records. Based on 15 of the 100 public records; the other 85 were not read in full.
Manufacturing’s landing page frames scaling around shop-floor data. The linked feasibility article says that the hardest ideas have stronger data access and integration scores, and that decision and orchestration complexity distinguish them.
The summary and supporting analysis send buyers towards different diagnoses. The issue is the coherence of the explanation; this audit does not validate the scoring model.
Explain the relationship between data readiness and decision complexity. Align the headline, supporting chart and linked analysis around that explanation.
Compare the manufacturing hero with the section on hard ideas having better data and harder decisions in the feasibility article.
Captured main content contained approximately 1,123 words on Home, 1,141 on CIO and 2,315 on the Library page, including widget text and excluding global navigation. The same library statistics and definitions recur across the journey.
Readers must decide what matters before they understand the product. Phone layouts are readable, but stacking the same content makes a long journey.
Give each page one main decision. Show a short portfolio-level example early, then the relevant proof and next step. Put detailed classifications, methods and research behind clear links.
Walk Home → role page → platform → pricing and count repeated explanations, including at a narrow viewport. Word totals describe captured states, not every widget state.
Services lists Business case sprint as workshop 4 and Decision board as workshop 7, yet says workshops 1 to 4 lead to the first decision. Pricing’s Expert Start route lists the first Decision Board as 04 in month one and the Business Case Sprint as 05 in month two.
Review names also vary: Quarterly Portfolio Review (an included 45-minute call), Quarterly value review (workshop 8 on Services) and Value Realisation Review (a €1,750 workshop).
Buyers comparing support packages may be unsure which sequence and which review they are buying.
Publish one named workshop catalogue and a separate schedule for each route, using the same names and numbers throughout. Explain any staged decision made before the deeper business case.
Compare the numbered workshop list on Services with the month-by-month Expert Start route and review descriptions on Pricing.
The featured insurance use case’s metrics are named review time reduction, error rework reduction and straight-through processing increase on the homepage and library page, but manual review time reduction, rework reduction and clean pass-through increase on the insurance page.
The method is shown as six steps on the homepage, four on /platform and five on the CFO page. The AI register records owners by role, never by name, while the CHRO and workforce pages count accountability held by a named person.
Shifting terms make the method harder to remember and harder for a buyer to quote back internally.
Keep a short glossary as the single source for metric names, step names and accountability terms, and use one method diagram everywhere.
Compare the linked pages. The 96% (AI prepares, a person decides) and 80% (named accountable person) figures measure different things and are not themselves in conflict.
Privacy, terms and the data processing agreement use the legacy menu rather than the current Platform / Solutions / Tools / Insights structure. No canonical link was found on these three pages. Discovered legacy marketing aliases resolve to current pages.
The journey feels inconsistent and creates unnecessary maintenance. A .php suffix by itself is not an SEO defect.
Apply the current shared navigation and explicit self-canonical URLs. Keep working legacy redirects and verify status codes when deployment access is available.
Open each legal document and compare its header with the homepage. Browser navigation verified destinations, not redirect status codes.
The Financial Services directory lists two direct-sector records, while its embedded explorer includes banking and insurance records and reports 13 public cases, without explaining the scope.
The public library has 15 directories, listing Banking (8), Insurance (3) and Financial Services (2) separately. Banking and Insurance have their own industry pages, but the explorer and Snapshot industry lists offer only Financial Services.
Different valid groupings can look like another data inconsistency, and a banking or insurance visitor cannot select their own sector in the tools.
Show Banking and Insurance as named sub-sectors of Financial Services everywhere, with direct-sector and roll-up counts labelled separately.
Compare /library, /library/financial-services, the industry pages and the industry lists in the explorer and Snapshot advice form.
The site and footer present the product as MetroMinds. The legal pages name OmniMinds Technologies B.V. (KVK 42130269), and the footer LinkedIn link opens the OmniMinds company page. The About page names neither the legal entity nor its location.
Procurement teams check who they are contracting with. An unexplained second name adds friction and can look like concealment.
Add one line to About and the footer, such as: “MetroMinds is a product of OmniMinds Technologies B.V., [city], Netherlands (KVK 42130269).”
Compare the About page and footer with the header of any legal document. Company registration details were not independently verified.
Articles are bylined MetroMinds only, and the main sitemap gives all 30 blog posts the same last-modified date, 18 October 2026. Several open with a specific scenario, such as a mid-sized manufacturer’s CIO with 48 initiatives or a mid-sized insurer’s steering committee, without saying it is illustrative.
With no customer evidence elsewhere on the site, unlabelled scenarios can read as implied customer stories.
Mark scenarios as illustrative, add named authors or reviewers where their expertise can be shown, and show real publication and update dates.
Open the linked articles and the blog entries in the main sitemap.
No findings match these filters.
Existing wording and a proposed replacement. Current wording is paraphrased except where quoted; open the linked page for the exact text. Square brackets mark facts MetroMinds must supply. None have been filled in.
Current: “Ranges come from comparable use cases in the MetroMinds Library.”
Proposed: “Range source: [source type], [number] implementations, reviewed [month year]. Treat it as a planning assumption until your own data replaces it. [link: How we build benchmarks]”
Why: Names the evidence behind each range and keeps the honest caveat.
/library/banking/keeping-commercial-client-files-current-for-bank-credit-reviewsCurrent: (paraphrased) A button inviting visitors to explore all 4,980 use cases, which opens an explorer holding 100.
Proposed: “Browse 100 free use cases”, with the supporting line “The full library of 4,980 is available to customers on [plans].”
Why: The click delivers what the label promises.
/Current: (paraphrased) Data is hosted in the EU, with a data processing agreement.
Proposed: “Primary hosting and databases in the EEA ([provider], [region]). Some support and email services may process data outside the EEA under standard contractual clauses. [link: See our sub-processors]”
Why: Matches the privacy policy, so a security reviewer finds no contradiction.
/aboutCurrent: (paraphrased) Most insurance use cases sit in case and exception handling.
Proposed: “Case and exception handling is the largest single group: 24 of 84 insurance use cases (29%).”
Why: Accurate, and still makes the point that claims work dominates.
/industries/insuranceCurrent: (paraphrased) The inputs drive the primary calculation, referred to by its internal code name.
Proposed: “Your business case uses three inputs: hours spent on this work, the share of time it takes and the fully loaded cost per role. The benchmark reduction is applied to those hours.”
Why: Plain language a CFO can repeat; the code names move to the method page.
/library/banking/keeping-commercial-client-files-current-for-bank-credit-reviewsA direction to test with MetroMinds, not finished copy.
Compare value, feasibility and risk in one portfolio. Turn promising ideas into business cases your leadership team can review.
Show a real portfolio screen with useful labels and explain the decision it supports.
Show benefit ranges, assumptions and the evidence behind them.
Show the prioritised outcome and a realistic team workflow.
Visual brief: swap the single use-case record for an annotated portfolio screen. Use real product screens and label any fictional data.
| Offer | Explain at a glance | Relevant next step |
|---|---|---|
| Free tools | A sample of use cases and one-off results for one person | Try a tool, then preview its relevant platform workflow |
| Paid platform | What a team does in the platform, and exactly what each plan includes | See product workflow → compare plans → targeted demo |
| Optional services | When workshops or Expert Start are worth paying for, and what they cost | Compare independent start with supported onboarding |
Tools ↗ · Platform ↗ · Pricing ↗ · Services ↗
Based on section 03 and findings 04 and 14. Check the wording against the real product before using it.
We used made-up inputs and sent no forms or meeting requests. Results cover the routes we tried, not every option.
| Tool / result | Test scenario | Observed outcome |
|---|---|---|
| Use-case explorer Passed sampled path | Manufacturing filter → supplier-delay details → Estimate savings | Eight public cases displayed; details loaded; the selected case carried into the calculator. |
| Savings calculator Passed arithmetic | 40,000 items/year · 45 min/item · €55/hour · scope 60% · adoption 70% · confidence 80% · benchmark 10–22% | Low €55,440 · mid €88,704 · high €121,968. Independently recalculated and matched the output. |
| Readiness Snapshot Tie result wrong | Functional Readiness / score 2 for all six domains | Correct overall 2.0/5 and domain scores. Governance incorrectly selected as both strongest and weakest. |
| EU AI Act check Passed sampled branch | Employment use selected; confirmed ranking or scoring people | Returned one potentially applicable high-risk area, duties, dates and an optional checklist request. |
| Workforce impact explorer Passed sampled path | Changed job family to Finance and Capital Management | Three public examples and finance-specific role, task and skill information appeared. |
Gross estimates from made-up inputs, before implementation and running costs.
Public pricing is one of the site’s strengths. The service descriptions around it need tidying (findings 09 and 15).
| Plan | Annual platform | Optional Expert Start | First year with Expert Start |
|---|---|---|---|
| Foundation 2 portfolio managers · 25 contributors | €17,500 | €15,000 · one module | €32,500 |
| Scale 5 portfolio managers · 100 contributors | €32,500 | €30,000 · two modules | €62,500 |
| Enterprise Agreed allocation and scope | Custom | Custom | Custom |
Published as inspected on 26 October 2026. EUR, excluding VAT. Independent Start and the specified Quarterly Portfolio Review are included; Expert Start is optional. Individual workshops range from €1,250 to €2,250 per session and require an active platform licence. Published pricing
Choosing a demo date shows eight time slots. We didn’t book one.
The mobile menu lists all five tools.
The guided tour works, and the Storylane demo loads when clicked.
The five changes we would make first, then a suggested order for the rest.
Ranked by how much we expect each one to help buyers understand and trust the offer.
Leadership, research, data and web: publish how the benchmarks are built, add the team and the company behind the brand, make hosting and security wording match the legal documents, and correct the numbers, shares and formula diagram. Most of this is editing, not design, so it can start straight away.
Done when: every statistic traces to one figures file, and a security reviewer finds nothing on the site that contradicts the contract documents.
Marketing, product and design: an annotated first screen comparing several AI projects, an honest library button, a short free-versus-paid explanation and a plan comparison that names every module.
Done when: most of five target buyers can say what MetroMinds sells and what they would click next from the first screen and headings alone.
UX, product and commercial: after each tool result, show the matching platform view and a relevant demo, and shorten the route from role pages to pricing.
Measure: demo requests from the right kind of buyer and lead quality, by entry page and device, against a baseline taken before the changes.
Click a screenshot to enlarge it. Each one shows the page as it was on 26 October 2026.
89 main-sitemap pages, 100 public use-case detail pages, three legal documents and the login page. 108 distinct pages were opened; 85 use-case detail pages were inventoried only.
| Page / URL | Category | Review state |
|---|
“Inspected” means we opened the page and read it; depth varies. We read 12 of the 30 articles in full and checked the other 18 for sources and statistics. We read 15 of the 100 public use-case pages, one from each of the 15 sector directories. Download the complete CSV ↓
Everything here is based on the public website as it stood on 26 October 2026.
The EU AI Act check uses 2 December 2027 and 2 August 2028 for the high-risk rules. Those dates follow the amended regulation, so we haven’t flagged them.
EUR-Lex: rules for trustworthy AI ↗
Source inventory: Main sitemap · Public library sitemap · Robots guidance
This is a website and buying-experience audit, not a legal opinion, security assessment or certification.
SEO means making pages easy for search engines to find and useful for people searching for an answer. For MetroMinds, the goal is relevant visits that can lead to a product conversation.
The two public sitemaps list the main website and public use-case records. In our saved checks, the platform overview, library and portfolio pages each have a descriptive title, a page description, one main heading and a canonical URL—the address the page identifies as its preferred version.
That’s a good start, although it doesn’t tell us whether Google has indexed them.
Platform overview ↗ · Library page ↗ · Portfolio page ↗ · Main sitemap ↗ · Library sitemap ↗The role pages, tools and industry pages cover useful subjects. The next step is to find out which questions buyers actually search for and give each important question a clear home.
Example to research: someone comparing AI projects may need an explanation of how to prioritise them, a worked example and a link to the portfolio product. “AI portfolio prioritisation software” is a topic to research; we haven’t checked search volumes.
Radence can help: research search questions, map them to pages and rewrite titles, introductions and internal links. Avoid creating several pages that give essentially the same answer.
CFO page ↗ · Portfolio product ↗The reviewed articles name sources but do not link out to the underlying research. Sampled library records also contain internal field names. Add source links, explain terms in plain language and show how an example works in practice.
That gives readers a way to check the claims.
Claims-handler article ↗ · Public library ↗Search Console and Bing Webmaster Tools access would let us check indexed pages, search queries, impressions and clicks. A technical crawl should also check redirects, duplicate pages, page descriptions, structured data and mobile performance.
The three legal pages have no canonical tag, which a developer can add quickly.
Privacy page ↗ · Terms page ↗ · Data processing agreement ↗“AI SEO” usually means improving the chance that an AI search service can find, understand and refer to your pages when answering a question. You may also hear it called GEO. There’s no guaranteed way to get cited.
On each important page, answer the main question in a short paragraph, then explain it with an example. For instance, the business-case page could explain which inputs are needed, how time is converted into money and which assumptions affect the estimate.
Keep important information as page text, with diagrams supporting it. A visitor should not have to interpret a screenshot to find the answer.
Business-case page ↗MetroMinds sells benchmark-based decisions. A public method explaining sources, dates, sample sizes and limits would make those benchmarks easier for readers to assess and reference. Individual records should link to that explanation.
Keep company details, product names, prices and statistics consistent across the website. Add named authors or reviewers where their role and expertise can be verified.
About page ↗ · Pricing ↗ · Benchmark explanation ↗Review whether search crawlers can reach and read the pages MetroMinds wants discovered. Check robots rules, indexing controls and any hosting restrictions. Crawler access for search and access for model training are different choices; review each provider’s controls before changing them.
Google says its AI features rely on normal SEO and need no special markup or AI text file. Other AI services may work differently.
Published robots rules ↗Agree realistic buyer questions, then record which services were tested, the date, the question, the answer and any cited pages. Repeat the checks and look at relevant visits and enquiries where the source is available.
We haven’t measured how often AI tools mention MetroMinds today; this would set that baseline.
Guidance checked against Google Search Central: AI features and your website and Bing Webmaster Guidelines.
A visitor may be interested without being ready for a demo. The contact experience should explain their options and what happens after they get in touch.
Why it matters: someone may want to ask about pricing, a feature or whether the platform suits their business before agreeing to a call.
What to change: the contact page already offers both a question and a demo. Make that choice visible from the demo buttons elsewhere on the site, so people who aren’t ready for a call know they can just ask.
The contact page already does this well: the demo is 30 minutes with a clear agenda, and questions get a reply within one working day. Repeat that next to demo buttons on the role and pricing pages.
Why it matters: people should know what they’re signing up for before they hand over their details.
Review each field: does the team need this information to answer or arrange the meeting? Use clear labels, mark optional fields and explain errors beside the field that needs attention.
Not yet tested: keyboard use, phone entry, error messages and spam protection.
After sending, show a clear success message and explain when to expect a reply. For a booking, confirm the date, time zone and how to reschedule or cancel.
Not yet tested: whether a request arrives, the confirmation email and the calendar invite.
Review the contact page ↗ · Follow the CFO demo route ↗
This section is a commercial proposal from Radence, kept separate from the audit findings. Start with the pages that explain the product and lead to a demo. Agree the scope before taking on a wider website rewrite.
Proposed work: rewrite the homepage, key product explanations and button labels. Explain the difference between free tools, the paid platform and optional support.
You receive: draft page copy and a simple message guide to keep future pages consistent.
Proposed work: plan the homepage layout, shorten repeated sections and mark up real product screens to show what a buyer can do.
You receive: desktop and mobile page designs, with a visual brief for the product examples.
Proposed work: design the next step after a tool result and improve the route from role pages to pricing and a relevant demo.
You receive: a visitor journey map, result-screen copy and designs for the main calls to action.
Proposed work: help present MetroMinds’ method, team and genuine customer examples clearly. Plan short visitor tests and agree which enquiries to measure.
You receive: an evidence checklist, draft page sections and a test plan. MetroMinds supplies and confirms the underlying facts.
This is a starting point for discussion, not a quote. Build work, analytics access and any wider rollout would be agreed separately.
See the suggested homepage direction ↑ · See the order of work ↓