Baseline Review 2026: Qualitative Stock Research Tool Tested
Tested on 13 October 2026 using the Consumer plan.
Baseline is worth considering if your equity research stalls at understanding a business rather than finding its financial ratios. It combines qualitative stock screening with substantial company reports, competitive assessments and investment debates. In our test, its best feature was its ability to move from a question like “which businesses have exceptional pricing power?” to a shortlist, then to the reasoning behind each score.
This review covers the Consumer workflow we actually used: screening, company research, source checks, peer charts, saved screens, watchlists and notes. We examined NVIDIA, ASML, Coca-Cola and PHINIA, and tested JPMorgan coverage. Baseline deserves a prominent place in our equity research software comparison, particularly for business-quality research. Its coverage gaps, uneven refresh dates and occasional factual inconsistencies mean it still needs to sit alongside original filings and a financial modelling tool.
Our verdict on Baseline
Baseline is a useful research accelerator for fundamental investors who want to discover and understand companies through a consistent qualitative framework. The explanations often do the difficult work of identifying what drives a business, which risks matter, and which evidence would strengthen or weaken the thesis. The strongest sections give a researcher a practical agenda for further investigation.
The Consumer plan was $39 per month when checked. That is easier to justify if you regularly investigate unfamiliar companies than if you mainly need historical financial tables or adjustable valuation models. Start with the free allowance, search the exact companies you follow, and inspect the sources behind a few decisive claims before subscribing.
| What worked well | What needs care |
|---|---|
| Qualitative filters, detailed score explanations, company and segment education, bull/bear debates, cited evidence panels, custom peer charts, persistent screens and notes. | Selective coverage, differently dated reports, inconsistent statements between sections, uneven citation availability and limited modelling/export functionality in the Consumer surfaces tested. |
What Baseline does and how its scores work
Baseline applies a common research framework across a global universe. Its main lenses are Business Quality, Competitive Dynamics and Risks, with more specific assessments underneath. These include pricing power, switching barriers, mission criticality, end-market quality, management, market leadership, accounting and the potential effect of advanced AI. Business-model and event tags provide another route into the universe.
The 1-5 scale requires careful reading. Higher values are favourable, including in the risk categories: a high risk score indicates a more contained risk profile. Baseline’s methodology identifies 0 as missing data. A score is an analytical judgement under the platform’s rubric, not a probability of outperformance or a percentage margin of safety.
Baseline View adds a separate synthesis of the company’s likely business trajectory over three to five years. A good business does not automatically deserve the highest trajectory assessment. We found this distinction useful, although the synthesis needs to stay aligned with the underlying reports as they change.
Qualitative screening is the clearest reason to use it
The screener displayed 3,644 stocks during our session. Alongside country, sector and market capitalisation, filters covered characteristics such as pricing power and barriers to switching. Tags included subscription revenue, market structure, founder leadership, corporate events, hidden assets and activist involvement. These let a researcher express a business hypothesis more directly than a conventional list of valuation ratios.
We selected a pricing-power score equal to 5 and restricted countries to the United States and the Netherlands. Baseline returned 57 companies. That result included NVIDIA and ASML, but also companies outside semiconductor equipment. It was a workable starting universe for investigating strong pricing power, not a conclusion that every result was attractively valued.
We expanded the Business Quality columns, temporarily hid the Sector column and tested ascending score sorting. In our two-stock shortlist, ASML’s 4.6 appeared before NVIDIA’s 4.7. These controls worked. The table can get wide when detailed columns expand, so a focused selection is easier to use than showing every dimension at once.
Company reports explain the business before presenting a thesis
Business Overview contains Stock Story, Segment Overview and Key Debates. ASML was a useful test because its reported segment structure differs from the product and end-market breakdowns investors commonly discuss. Baseline explained that the business has one reportable segment and that technology, geography and customer end-use are alternative views of revenue. It explicitly warned against adding those overlapping views together.
The Key Debates report, generated on 7 October 2026, examined the durability of AI-related capital spending, whether technology leadership can improve profits, and the effects of China restrictions and competition. Each debate separated context, the bull argument, the bear argument, a current assessment and the evidence that could resolve the disagreement. That structure is particularly useful when preparing questions for an earnings call or deciding what to monitor next.
Coca-Cola demonstrated that the scoring framework can make distinctions within a familiar business. Its report assigned 4/5 pricing power while acknowledging that consumers can easily switch drinks. It treated brand preference, package mix, and distribution separately from contractual customer lock-in. That is more useful than equating a famous brand with an unqualified maximum score.
Sources are useful, but auditability varies between reports
NVIDIA’s detailed pricing-power report contained clickable source markers. Selecting one opened a side panel with the relevant excerpt and an external source link. The financial-results citation led to NVIDIA’s own release. A different claim about estimated B200 economics relied on third-party analysis. The interface makes that distinction discoverable, but the reader still has to make it.
Company-reported gross margin and an analyst’s estimated hardware margin answer different questions. A linked estimate should not acquire the authority of a reported financial statement simply because it appears in the same report. We would keep those evidence types separate when building a model or writing an investment memo.
Auditability was less consistent in the NVIDIA Baseline View synthesis. We did not find the same clickable source markers in that report. Its underlying detailed sections can help, but a reader cannot assume every statement across every module has an equally direct evidence trail.
What our factual spot-check found
We checked 13 selected factual claims across four covered companies against primary references. Twelve matched those references. One blurred an acquisition’s announcement and completion dates. This was a targeted spot-check of claims encountered during research, not a random sample or an estimate of the platform’s overall accuracy.
| Company and checks | Result | Primary reference |
|---|---|---|
| NVIDIA: Q2 FY2027 revenue of $96.2 billion, 106% year-on-year growth, $89 billion Data Center revenue and 75% gross margin | Four claims matched. | NVIDIA Q2 FY2027 results |
| Coca-Cola: 2025 organic revenue growth of 5%, price/mix growth of 4%, concentrate-sales growth of 1%, and comparable operating margin of 31.2% versus 30.0% | Four claims matched. Concentrate sales must not be confused with unit-case volume. | Coca-Cola 2025 results |
| ASML: its 1984 founding and the timing of the Cymer acquisition | The founding matched. The Stock Story used 2012 for the acquisition; completion was 30 May 2013. Baseline’s separate Segment Overview correctly used 2013. | ASML history and completion announcement filed with the SEC |
| PHINIA: 2025 sales of $3.48 billion, net earnings of $130 million and $242 million returned to shareholders | Three claims matched. | PHINIA 2025 results |
The ASML discrepancy is a small historical error, but it demonstrates a broader diligence problem: two polished sections can disagree. The acquisition was announced in 2012 and completed in 2013. A reader should not need to reconcile those dates, yet our test did.
Check generation dates at the section level
Baseline displays generation dates, which is valuable. The difficulty is that a company’s sections do not necessarily share the same date. NVIDIA’s Baseline View was generated on 2 September 2026 and referred to a 4/5 pricing-power assessment. Its detailed Pricing Power report, generated on 21 September, showed 5/5. That can reflect an update arriving in one section before another, but it leaves the synthesis out of step with the current detailed score.
PHINIA illustrated a wider date spread. Its Accounting report was generated on 28 March 2026, while its AGI Scenario report was dated 22 September. ASML’s Key Debates was more recent, dated 7 October. The correct conclusion is that freshness varies by module. We would check the date of each decisive report and look for subsequent filings or events before using it.
Baseline’s published methodology discusses its rubric and a vendor-hosted factual-verification exercise. That transparency is helpful, but it should be read separately from our hands-on spot-check. Neither exercise establishes that every current report is correct or that the scores predict investment returns.
Risk research includes accounting and AI scenarios
PHINIA’s accounting report looked beyond a generic risk list. It discussed the company’s spin-off heritage, related-party arrangements, a tax settlement, revenue recognition and adjustments. These are useful prompts for checking the financial statements. We inspected the report’s structure and dates; we did not independently validate every accounting conclusion it contained.
The AGI Scenario report asked how a business would fare if cognitive work became cheaper. For PHINIA, it distinguished engineering and administrative tasks from physical manufacturing, qualification requirements and distribution. This can help expose an overlooked risk or resilience argument, but it remains a scenario analysis. The assessment is not evidence that a particular AI adoption path will occur.
Charts add valuation context without replacing a model
The company overview provides price charts, peer comparisons and historical market narratives. The metric selector offered Price, P/E, EV/EBITDA, EV/Sales, EV/EBIT and P/CF. We switched NVIDIA to a three-year EV/Sales view, added Microsoft to the existing AMD and Broadcom peer selection, and inspected the peer average.
This makes it easier to ask whether a company’s multiple has changed relative to a chosen peer group. It does not resolve whether those peers have sufficiently comparable margins, growth or business models. The displayed EV/Sales series was labelled NTM, so an investor should also understand the forward-estimate basis before comparing it with trailing numbers elsewhere.
We did not find a Consumer workflow for editing DCF revenue growth, discount rates or terminal assumptions. We also did not establish quote latency, dividend-adjustment conventions or a reproducible return calculation for every chart view. Treat the chart as research context rather than an execution feed or an independently verified performance record.
Saved screens, watchlists, notes and copying
We created a watchlist containing NVIDIA and ASML, filtered the table to those two companies and saved the resulting screen. The named preset remained available at its saved URL after reloading. We also wrote an NVIDIA note recording the mismatch between the synthesis and the newer pricing-power score, then confirmed the note remained after we navigated away and returned.
These features connect discovery to follow-up work. A saved screen preserves the question, a watchlist preserves the companies, and notes preserve the analyst’s own doubts. The notes in this test were clearly labelled as review material.
The report copy control worked on ASML’s Key Debates, returning text that included its title, generation date and source-marker numbers. This is useful for a working note, but we did not validate preservation of clickable source links through every paste destination. We did not find or test an Excel/CSV download, spreadsheet add-in or API in the Consumer interface.
See more of our reporting in Google Top Stories, AI Overviews and AI Mode.
Coverage: global does not mean complete
The universe included companies from multiple markets, and ASML provided a substantive non-US test. PHINIA showed coverage wasn’t confined to the largest technology companies. However, searching JPMorgan opened a notice saying JPM was not on Baseline yet, with a button to request coverage. We left that request unsubmitted.
This is a meaningful limitation for someone choosing a main research subscription. Search your actual holdings and candidate list before paying. The size of a global universe is less useful than whether it covers the companies and sectors you need. Our sample is too small to infer a general coverage rate for banks or smaller companies.
Baseline pricing and what each plan includes
The pricing page showed the following on 13 October 2026. Prices were in US dollars. We tested Consumer access, not Enterprise custom research.
| Plan | Published price | Scope and status observed |
|---|---|---|
| Free | $0 | 10 result deep-dives a month, every covered stock and column, search, filters and sorting. No card required. |
| Consumer | $39 per month | Unlimited result deep-dives, every covered stock and column, saved screeners, watchlists and stock notes. Billed monthly, cancel anytime. |
| Professional | $79 per month displayed | Marked “Soon” and unavailable to select in our test. Do not treat it as a currently purchasable plan. |
| Enterprise | Custom | Advertised Deep Research, custom fields and custom dashboards. These features were not tested. |
The free allowance is described as result deep-dives. We did not measure precisely how the free counter treats repeated visits or individual report sections, so we would not translate that into a promise of 10 complete company dossiers. Consumer removed that practical limit during our research session.
Baseline versus broader research platforms
Baseline deserves a more prominent specialist position for qualitative discovery: find companies with a particular business characteristic, understand their economics, and identify the debates worth investigating.
Within our wider comparison, Koyfin retains the broad-workspace role. BAMSEC addresses filing-intensive work, while Fiscal.ai and TIKR serve structured financial analysis. We did not rerun those platforms in this Baseline session. Their positions here provide workflow context, rather than a new simultaneous benchmark.
For a model-heavy analyst, the missing piece is still a reliable path from financial statements to assumptions, sensitivities and valuation outputs. For a researcher starting with an unfamiliar company, Baseline can make the initial reading and question-setting stage substantially more organised. We did not run a timed productivity study, so we can’t claim measured hours saved.
Who should use Baseline?
Try Baseline if you regularly build new company shortlists, want a consistent framework for judging business quality, or need a structured first pass before deeper diligence. Its combination of filters, explanations, debates and persistent research notes is particularly useful for that job.
Choose another starting platform if your essential workflow is comprehensive financial history, refreshable Excel models, editable intrinsic valuations or guaranteed coverage of a particular sector. Before buying Baseline, use the free plan to test your own companies, open several citations and compare the dates across their reports. If those checks pass for your universe, the Consumer plan offers a focused qualitative research layer at an accessible monthly price.
How we tested and what remains untested
Our desktop-browser test used an authenticated Consumer account on 13 October 2026. We exercised screening, sorting, column controls, saved presets, watchlists, company notes, selected company reports, source panels, custom peer charts and report copying. We checked 13 selected factual claims against company or SEC references and recorded both an unavailable company and cross-section inconsistencies.
We did not test Enterprise Deep Research, custom fields or dashboards, team administration, mobile usability, API integrations, service-level support, portfolio backtests or trading execution. We did not validate every report, forecast, chart or score. Screenshots show the tested interface and dates, not guarantees that the same data or coverage will remain unchanged.
You Might Also Like:
AI Portfolio Analysis Tools
Best Equity Research Software
Best AI Trading App For Beginners
