Research standard
How ToolThesis reaches a recommendation.
Our job is not to tell you which product has the most features. It is to help you make a better operating and financial decision.
Every conclusion should be traceable to evidence.
We distinguish verified facts, structured testing, professional judgment and modelling assumptions. If we have not used a product, we say so. If data is unavailable, we do not invent it.
Define the decision
We start with the buyer, workflow, team size, budget, operating constraint and decision that needs to be made.
Build the candidate set
Products are identified from market research, buyer relevance and realistic availability—not commission size alone.
Verify current facts
Pricing, functionality, limits, integrations and programme terms are checked against primary sources and dated.
Test realistic workflows
Where access allows, products are tested using synthetic companies, contacts, documents and meetings. Client or employer data is never used.
Model total cost
We calculate subscription, seats, implementation, integrations, training, switching effort and plausible operating savings.
Assess fit and trade-offs
Products are evaluated against the needs of small professional-services firms rather than generic feature counts.
Reach a conditional verdict
Recommendations specify who should buy, who should skip, important alternatives and the immediate next step.
Monitor and correct
Material pricing, product and policy changes trigger review. Substantive corrections are disclosed rather than silently hidden.
Evidence labels
What our language means.
Regular use
Ongoing first-hand experience with the named product and relevant workflow.
Structured hands-on test
A defined scenario tested specifically for the analysis, with the date and important limitations recorded.
Previously used and re-tested
Prior experience refreshed through a current structured test before publication.
Documentation-based analysis
Current primary-source research without a claim of first-hand product experience.
Modelled assumption
An explicit scenario used to estimate economics. It is not presented as an observed customer outcome.
The independence rule
A commission can monetise a recommendation. It cannot determine one.
Affiliate availability is recorded after the relevant product set and evaluation criteria are established. Products without affiliate programmes may still be included when they are relevant.
Use of AI
AI assists the process. Human judgment owns the verdict.
AI may support research gathering, structuring, extraction, draft development, editing, consistency checks and refresh monitoring.
Final recommendations, material claims, evidence labels and conclusions require human review. ToolThesis does not use AI to fabricate product experience, statistics, interviews, quotations or performance results.