What drives business performance, measured.
PIMS® links the market position, operations, and cost and capital structure of more than 4,300 business units to their financial results. The data comes from confidential management accounts, collected between 1967 and today and never published.
Used to set targets, value investments and evaluate AI models.
12,675 three-year observations · 477 variables each · 95.7% complete · 1967 to date
Who works with PIMS® data
AI labs and evaluation providers
Test how well models judge strategy and performance, on thousands of real business records no model has seen.
Strategy and consulting firms
Set targets and test recommendations against the outcomes of structurally comparable businesses, not general case studies or industry averages.
Investors and private equity
Compare a company's actual return with its PAR. Below PAR can signal unrealised potential, above PAR a likely decline. For due diligence or a whole portfolio.
Company leaders use the same data through Strategic Benchmarking. Researchers can request the brochure.
Recorded where decisions are made, and followed over time.
Every variable has a maintained definition, so a business measured in 1985 is directly comparable with one measured today, and an industrial supplier with a consumer brand.
The data AI models have never seen.
AI models learn from public text. Confidential management accounts are not part of it, and neither is what those businesses went on to report. PIMS® holds both, for 12,675 real cases.
Unseen by any model
The records have never been published and are not part of any AI training dataset to date. Cases held back for evaluation stay unseen.
A real outcome for every case
Each record links a business's position and choices to the results it reported years later. Judgement questions get an answer that actually happened.
Thousands of real cases
Each of the 12,675 observations can become a task. No formula gives the answer, so a model has to judge, and tasks stay hard as models improve.
Real businesses, not fictional worlds
Agent environments can be built around an anonymised PIMS® business, with later years held back. Its figures fit together the way a real business's do.
What a task built from PIMS® data looks like
A task is one test question for a model, with a known answer.
The answer is what the business actually reported.
Anonymised extracts and a data dictionary are available under NDA, or we build and run task sets in our own environment.
Talk to the data team →
What's in the data
477 variables per business unit, in five groups.
What each group covers
The data at a glance
One strategic business unit (SBU) over one three-year window: its market position, operations, cost and capital structure, and the financial results that followed. The database holds 12,675 such observations.
More than 4,300 units, each tracked year by year for 3 to 12 years. Together that is more than 25,000 business-years of panel data.
North America, Europe and Asia, across every major industry group. North American, manufacturing-intensive businesses from before 2000 are the most heavily represented.
1967 to date, with contributions continuing.
95.7 per cent: 5,783,513 values in 6,045,975 possible fields.
Every one of the 477 variables has a maintained definition, formula and collection protocol. The data dictionary is available under NDA.
Pre-revenue and very early-stage businesses. Their results reflect start-up choices more than competitive position and would blur the link between position and performance the data is built to measure.
No. The records have never been published and are not part of any AI training dataset to date. They cannot be reconstructed from public sources.
Agreed with each client: anonymised extracts, evaluation task sets, analyses run by our team, or full access under licence.
pims.ai | malik holds the database and all rights to it, so licences and exclusive arrangements can be agreed directly.
Models built on the data
Six decades of analysis on the data condense into nine basic findings on business strategy. The models below turn them into estimates for a specific business.
Profit Potential Analysis (PAR®)
Estimates the return on sales a business can sustain from 15 structural factors across competitive position, market environment, and cost and capital structure. They are leading indicators, so PAR predicts profitability rather than reconstructing it.
Variation in long-term profitability, approximate shares
Strategic Peer Analysis (ROLA®)
Finds the businesses most structurally similar to the one analysed, its strategic peers, and splits them into winners and losers to show which differences account for the gap.
Cause-Effect Model
Maps the direction, size and order of the links between strategic variables and outcomes: which levers move profitability, growth and cash flow, directly or through other variables.
PAR in practice
How a 4-point gap was closed without a price war
Fictional example
Return on sales
- 1
Situation
A packaging business unit sells to food manufacturers. Return on sales has sat at 6% for three years. Management is split between cutting prices to win share and investing in quality.
- 2
PAR
Its structural profile supports 10%. The 4-point gap comes from inside the business, not from the market.
- 3
Strategic peers
The winners among its closest peers differ in two places: customers rate their quality higher, and they invest less relative to sales.
- 4
Decision
Fix product quality and delivery reliability first. Delay the planned capacity expansion. No broad price cuts.
- 5
Result
Three years later return on sales is 8.5%. More than half the gap is closed.
Ways to work with the data
Every arrangement is shaped around how the data will be used and agreed under licence. Three routes are open:
01
Evaluation and analysis
Work with results, not records. Record-level data stays with us.
- Custom analysis for your question
- Your AI model run against held-out PIMS® cases
- Aggregated benchmarks and model outputs
02
Licences and extracts
Work with the data yourself, under NDA and licence.
- Anonymised extracts for defined research questions
- Evaluation task sets with answer keys
- Secure workspace in an environment we control Coming soon
03
The full database
Long-term or exclusive rights to the complete database, including use for model training.
- All 4,300+ business units and 477 variables
- Joint research or products, with our team
- Terms agreed directly with pims.ai | malik
Six decades of research
1960s
General Electric
PIMS® (Profit Impact of Market Strategy) starts as a study of what drives business success.
1970s
Harvard Business School
A multi-company programme, later run by the Strategic Planning Institute in Cambridge, Massachusetts. Participants include Siemens, Nestlé and P&G.
1987
The PIMS® Principles
Buzzell and Gale summarise the findings. Further PIMS® research is published in Harvard Business Review, Strategic Management Journal, Journal of Marketing and Long Range Planning.
Today
pims.ai | malik
Maintains and extends the database from London, with statistics and machine learning.
PIMS – the broadest and best management database in the world – is the only source of strategic principles based on fact rather than opinion.
Check whether PIMS® fits your question.
For a sample extract or the data dictionary under NDA, talk to the data team. The brochure covers coverage, variables, models and access options.
PIMS® Data brochure
Request the specification and access guide.
What is in the data, why it is different, the models built on it with a worked example, and the ways to work with it.
We use your details to send you the brochure. No newsletter. Privacy Policy
Talk to the data team
What would you like to do with the data?
Tell us briefly and we will set out the options. A sample extract and the data dictionary are available under NDA.
We use your details only to reply to your enquiry. Privacy Policy