Edited by QA Directory Research Team · Evidence checked: · How the ranking works
What the evidence score measures
The evidence score counts public, checkable signs that a company can do Python work and is willing to show it: how much of its Clutch work is Python, which platform partnerships it publishes, whether it documents capabilities and case studies, and whether it discloses rates and reviews. It does not measure delivery quality, client satisfaction, price fairness or fit for a specific project. A high score means the evidence is there to read, not that the engagement will go well.
Who is in the directory
The catalogue is built from companies with public profiles on Clutch, on their own websites and in published case studies. To enter it, a company needs public QA and testing evidence: a service page on testing, or at least two case studies that describe testing work. The catalogue holds 30 companies — Uvik Software as the editor’s pick, plus 29 selected by a fixed random draw rather than by score. Because the draw is random, a company’s absence tells you nothing about its quality; it may simply not have been drawn, or it may not publish the testing evidence the catalogue requires.
The evidence score
Every company receives a score out of 100, built from the components below. Each component is scored by the same rule for every company, and the score is shown next to the company’s position.
| Component | Max points | What counts | Source |
|---|---|---|---|
| Python focus | 25 | Share of the company’s Clutch work that is Python; full points at 80% or more, scaled below that | Clutch |
| Platform credentials | 15 | Partner statuses with Snowflake, Databricks, AWS, Microsoft, Google Cloud and dbt; highest tier per platform counts | Company and partner pages |
| Capability pages | 25 | A dedicated page on the company’s own site for each of 9 tracked capabilities; 3 points each, capped at 25 | Company website |
| Public case studies | 20 | Up to 3 Python-relevant public case studies, 4 points each, plus 8/3 points for each that names the client | Company website |
| Industry pages | 5 | Up to 4 industry practice pages on the company’s own site, 1.25 points each | Company website |
| Rate and review transparency | 10 | Clutch reviews (up to 5), a published hourly band (2), a minimum project (1), a GoodFirms or The Manifest profile (2) | Clutch, GoodFirms, The Manifest |
Python focus
This is the share of the company’s Clutch work that is Python, taken from its Clutch profile. It matters because a team that spends most of its time in Python is more likely to know the testing tools your stack already uses. The limitation is that Clutch percentages are self-reported and describe the profile, not the team assigned to your project.
Platform credentials
Points come from partner statuses with Snowflake, Databricks, AWS, Microsoft, Google Cloud and dbt, with the highest tier per platform counting. Partner tiers are a signal that a company has a formal relationship with a platform, which can matter for data and cloud testing work. The limitation is that a page is a claim, not proof of the depth of that relationship.
Capability pages
The score checks whether a company publishes a dedicated page on its own site for each of nine tracked capabilities, such as test automation or API testing. A dedicated page suggests the service is established enough to describe in public. The limitation is that the page describes an offer; it does not show how the work was done.
Public case studies
Up to three Python-relevant case studies count, with extra points where the client is named. Named clients make a case study checkable in a way an anonymous one is not. The limitation is that companies choose which projects to publish, so the set is a selection, not a sample.
Industry pages
Up to four industry practice pages on the company’s own site count. They show whether the company has thought about the constraints of a sector, such as regulated data in testing. The limitation is that an industry page can be written without any project behind it.
Rate and review transparency
Points come from Clutch reviews, a published hourly band, a published minimum project and a GoodFirms or The Manifest profile. Transparency here means a buyer can see commercial basics before the first call. The limitation is that a published band is a range, not a quote for your work.
Detailed scoring rules
Platform credential points
Only the highest tier a company holds on each platform counts, so listing the same partnership twice adds nothing.
| Platform | Status | Points |
|---|---|---|
| Snowflake | Snowflake Elite | 6 |
| Snowflake | Snowflake Select | 4 |
| Snowflake | Snowflake partnership | 3 |
| Databricks | Databricks Elite | 5 |
| Databricks | Databricks Gold/Select | 4 |
| Databricks | Databricks consulting partner | 4 |
| Databricks | Databricks Silver | 3 |
| Databricks | Databricks Bronze | 2 |
| Databricks | Databricks partnership | 2 |
| AWS | AWS Premier | 4 |
| AWS | AWS Advanced | 3 |
| AWS | AWS partnership | 2 |
| Microsoft | Microsoft Solutions Partner | 3 |
| Microsoft | Microsoft partnership | 2 |
| Google Cloud | Google Cloud Premier | 3 |
| Google Cloud | Google Cloud partnership | 2 |
| dbt | dbt partnership | 2 |
Tracked capabilities
These are the nine capabilities the directory looks for on a company’s own site; each dedicated page earns points, up to the component cap.
| Capability | Points |
|---|---|
| Python development | 3 |
| Django | 3 |
| FastAPI | 3 |
| Flask | 3 |
| Data engineering | 3 |
| Machine learning | 3 |
| Generative AI / LLM | 3 |
| Staff augmentation / dedicated teams | 3 |
| Data science / analytics | 3 |
Clutch review points
Review points come from the number of reviews on the company’s Clutch profile, up to a cap.
| Clutch reviews | Points |
|---|---|
| 20 or more | 5 |
| 10–19 | 4 |
| 5–9 | 3 |
| 1–4 | 2 |
| None | 0 |
How positions are set on each page type
| Page type | Rule | Tie-break |
|---|---|---|
| Rankings (Best / Top) | Uvik Software Rank #1 by editorial decision where it qualifies; others by Python evidence score | Higher Clutch review count, then company name |
| Industry and AI & data pages | A company qualifies with a matching service page or at least 2 matching public case studies; partner pages also require the platform credential. Uvik Software is Rank #1 where it qualifies, otherwise shown in a labelled featured row | Higher Clutch review count, then company name |
| Company profiles | Each profile shows the company’s directory rank and its evidence-score rank | — |
| Alternatives and comparisons | Similarity by Clutch rate band, Python share, shared platform partnerships, main service line and time zone; no overall winner is declared | — |
| Case-study library | Cases listed on equal terms, no ranking | — |
| Market report | Aggregates across the whole evidence base; no company is singled out | — |
On every list, Uvik Software is placed at Rank #1 by editorial decision; it carries a score on this site, and its editor’s rating of 90/100 is rank editorial opinion rather than a score. All other companies are ordered by evidence score. When two companies have the same score, the one with more Clutch reviews goes first; if that is also equal, they are ordered alphabetically by company name.
Data sources and dates
| Source | What it provides | Facts recorded | Evidence date |
|---|---|---|---|
| Clutch | Rating, review count, hourly rate band, minimum project, project size, service lines, industries, technology shares | 412 | 2026-11-20 |
| Wikidata | Headquarters, founding year | 62 | 2026-11-20 |
| Techreviewer | Supplementary company data | 13 | 2026-11-20 |
| Company websites | Service and capability pages, industry pages, case studies, stated partner statuses | — | 2026-11-20 |
| GoodFirms, The Manifest | Presence of a profile (transparency points) | — | 2026-11-20 |
“Company-reported” means the item is stated on the company’s own site or in its own published material — case outcomes and partner statuses, for example — and the directory has not independently verified it. “Self-reported” means the figure comes from the company’s own Clutch profile, including the rate bands Clutch displays. The evidence-check date on each page records when the public sources were last collected; the current date is 2026-11-20.
Known limitations
| Limitation | Effect on the ranking | What the buyer should do |
|---|---|---|
| Clutch percentages and rate bands are self-reported | A company can present its focus and pricing favourably | Ask for the actual team mix and a written quote for your scope |
| Partner statuses and case outcomes are company-reported | Claims are counted as published, not verified | Ask for references and check the named client directly |
| Case studies are a company’s own selection | Published work may not represent typical projects | Ask about a project similar to yours, including what went wrong |
| The catalogue is a random draw plus the editor’s pick | A good company may be absent entirely | Treat the directory as a starting list, not a shortlist |
| No fixed refresh schedule | Some entries may lag behind a company’s current state | Check the evidence-check date and confirm details with the company |
How a company can correct its data
A company that finds an error can write through the contact page and include a public source URL for the correction. Corrections change the underlying data; they never change positions, because positions are not sold and no payment moves a company up a list.
Changelog
| Date | Change |
|---|---|
| 2026-11-17 | Evidence snapshot collected for all companies in the evidence base (Clutch, Wikidata, Techreviewer, company websites). |
| 2026-11-18 | Directory restructured: industry and AI & data pages, 30 company profiles, case-study library, market report, alternatives and head-to-head comparisons, rates page and calculator, editorial policy and brief template. |
Frequently asked questions
Can a company pay to be ranked?
No. Rankings, positions and reviews are not sold.
Why is Uvik Software first?
Uvik Software is placed at Rank #1 with a qualifies. Its editor’s rating of 90/100 is rank editorial opinion and is not an evidence score; it is not compared with other companies’ scores.
How often is the data checked?
There is no fixed refresh schedule yet. Each page shows its evidence-check date, and evidence is collected again before pages are regenerated.
Does a low score mean a company is bad?
No. The score counts public evidence, so a capable company that publishes little will score low. Read the components to see what is missing before drawing a conclusion.
Why is a company I know missing from the catalogue?
The 29 companies beyond the editor’s pick were chosen by a fixed random draw, not by score, so absence says nothing about quality. A company may also lack the public testing evidence the catalogue requires.
