How we compare QHSE software

How AI is used

This is the site's statement on AI use; every byline links here.

Drafted with AI language models: articles, software profiles, buyer guides, comparisons and FAQs. Drafts are built from vendor documentation and websites, public regulatory and standards texts and other publications. Pages link the sources they cite; some articles do not yet link theirs.

Generated automatically, without page-by-page editing: the short answers under /answers, the replies of the on-site chat assistant and, on a /compare page where neither product has a researched profile, the comparison summary. The chat replies and those summaries are written when you ask for them, by Google Gemini models through the Lovable AI gateway.

What the editor checks: Dimitris Mitsios is responsible for what is published. He checks selected pages against their sources and records each check with its date and what was checked; a page with a logged check shows it in the footer. No check has been logged yet.

What the editor does not do: test products hands-on, re-check every page or every source when a page is updated, or review the automatically generated text listed above. Not every page has been individually reviewed. AI drafts can contain errors, outdated facts or wrong attributions; where a page links its source, that source is the reference, and corrections are welcome.

No language model chooses or orders the Get Matched shortlist: it follows the fixed tag rules described below. AI-generated or AI-summarized text is not customer testimony.

How published prices are recorded

A product pricing page lists only amounts printed on the vendor's own pages, as the product's research record quotes them. Each row gives the unit, term and minimum, the page it was read from and the check date. These are list prices, not quotes. An amount shown with a dollar sign but no currency code keeps the bare sign. Default views of a pricing calculator are labelled as calculator views, not list prices.

The entry cost is arithmetic on those rows: the published minimum at the published rate, or a monthly package plus its one-time licence for the first year. Where a plan is listed with annual and monthly billing, the page shows how much less the annual option costs over 12 months. Neither figure is a quote; both exclude taxes, add-ons and services, and neither is worked out where the evidence does not establish the currency.

Peers are the other researched products in the same category that list an amount in the same unit and period, in alphabetical order, each at the lowest amount it lists in that unit, in its own currency and scope. A product whose prices buy something else, are calculator views or have no established currency is not a peer. Commercial partnership does not change this rule. Each product's own page gives its source.

A pricing page is indexed only when it adds to the product profile: at least two sourced price points, or one amount with its unit and a purchase minimum. The profile's pricing section then summarises that page and links to it.

Editorial evaluation framework

Added 19 September 2026. These dimensions guide product research and buyer evaluation. They are not computed product scores and do not replace the rating rules or quiz matching rules below.

Functional depth: 25%

Can the proposed product complete the required workflow, including exceptions?

Usability & adoption: 20%

Can representative users complete the task on their devices with appropriate training?

Implementation & TCO: 15%

What configuration, migration, support and total cost of ownership does the proposed scope require?

Integrations & API: 15%

Which connections are included, how does data move and who maintains them?

Security & trust: 15%

What evidence supports access controls, data handling, recovery and contractual commitments?

Customer voice: 10%

What dated, attributable customer evidence is relevant to the use case and product version?

A documented feature is not a hands-on result. Comparisons show documented scope, pricing conditions and fit, not an editorial score or an aggregate rating. Any future weighted score needs a disclosed rubric, evidence, date and treatment of missing data; these weights alone do not establish one.

How lists are ordered

Contextual product lists use “Best documented match”. It is computed from the researched record only and shows how closely our sources document a product for the subject of the list, not how good the product is. No rating, review count, traffic figure or payment is an input. Products are compared tier by tier:

1. Scope. First, products whose primary category is the subject of the list. Next, products for which a category tag set by our research, or a capability the profile marks as documented with a source, names the subject. Last, products that are on the list for another reason, such as a catalogue tag our research has not confirmed, an industry, a company size or a search. The number of tags a product carries does not count, so tagging it for more processes does not move it up.
2. Documentation. Three bands: a researched profile with sources and evaluation dimensions; a researched profile without both; a catalogue entry with no researched profile. We use bands rather than a score so that more research on one vendor does not lift it above others with a complete record.
3. Buyer transparency. A tie-break count of three things the profile documents: a published price (the same rule the pricing index uses), a free trial and an offline workflow for the assessed use.
4. Name. Remaining ties are broken alphabetically.

Lists that have sort controls, such as the software directory, also offer “Published price first” and “A–Z”; in every sort, Tekmon included, each entry keeps the place that sort gives it. There is no rating sort. Legacy catalogue ratings and review counts are unsourced and are excluded. Each list carries a one-line note that links here.

Measured market interest

The general software directory uses “Most searched — 12 months” as its default only while a separate evidence file passes every gate. The current status is active: 188 of 212 researched products have eligible measurements, against a minimum coverage of 50%. The current scope is All locations, English, on Google. If coverage, declared scope or freshness later fails, the option is hidden and the directory falls back to “Best documented match”.

The market-interest score reserves 75% for rolling branded-search demand, 15% for search momentum and 10% for privacy-safe reader demand. It renormalizes only the components that are actually available and publishes confidence beside each score. The current import therefore has 75% confidence because it contains reviewed search demand but not the two secondary signals. A product needs recent data from an identity-reviewed query to qualify.

Branded-search demand uses Google Ads historical metrics for 1 September 2025 through 31 August 2026. Google groups close variants, so the importer selects one reviewed query per product and never adds alias volumes together. Known contaminated broad terms are replaced by an explicitly reviewed vendor-qualified query. Blank measurements remain unavailable rather than becoming zero. Google Trends may later supply the secondary momentum signal because its public interface uses sampled, normalized interest rather than absolute volume; see Google's Trends methodology. AI-generated popularity tiers, payments, ratings and review counts are never inputs. The current machine-readable status and component scores are published in the market-interest dataset.

Sponsored placement

Tekmon pays for sponsored placement. It enters an organic list only when the same published rule admits it, at the position that rule gives it and with the same styling as every entry. Directory filters use their stated category, process, industry or company-size criteria. Process guides and workflow lists on reference pages additionally require a sourced documented or conditional capability for the specific workflow. Conditional scope is labelled. Alternatives use their shared-category or documented-feature eligibility rule and their existing similarity order; a profile's short peer list matches shared recorded attributes and displays its selected peers alphabetically.

The sponsored placement is additional: a card labelled “Featured · Sponsored” or a banner labelled “Ad”. The label and payment disclosure distinguish it from organic membership. Choosing A–Z in a directory's sort control keeps the partner's alphabetical place without a separate card. Tekmon never counts toward the minimum number of independent candidates required for indexing. Product lists have no numbered rankings, FAQs do not recommend a product, and comparisons between Tekmon and another product remain noindex. The separate rule for worked pricing examples continues to exclude the partner.