The Pivot Database

The pivot database holds 75 documented company pivots that I scored with the pivot score from The Startup Pivot Playbook and checked against what happened next. Of the 75, 38 succeeded, 24 failed and 13 ended mixed, and none of the eight companies that changed the brand before the product could back it succeeded.

I built it to test one question I kept meeting on about two dozen pivots as a CEO, operator, consultant and advisor: which part of the company should move first. The earliest case is Wrigley’s 1892 move out of soap and baking powder into chewing gum, and the latest are the AI pivots of 2024. Every case, with all six scores and the reason behind each, is in the Pivot Playbook Toolkit.

Cases in the database

A case entered the database only when the public record showed both the move and its result, with about five sources per case drawn from filings, reputable press, founder interviews and company histories. Two cases come from my own records as CEO, an IT operations software company and a legal technology company. Large companies make up 35 of the 75 because they leave filings and dated outcomes, while most startups that pivoted and failed left no record at all.

Table 1 · Composition of the pivot database

Groups by company size sum to 75. The other rows overlap them.

The pivot score

Each case carries a score of 0 to 3 on six dimensions: the four anchors of brand, target market, route to market and product, and the two levers of talent and finance. Finance means capitalization, so a funding round counts only when it serves the new direction. A total of 0 to 4 is an adjustment, 5 to 13 is a pivot and 14 to 18 is a transformation. Applying the capitalization rule in October 2026 changed the finance score of 22 cases and moved 8 of them between classes.

The score also splits the six by how they change. Product, route to market and finance are hard functions, moved through engineering, channel build-out and capital over quarters or years. Brand and target market are soft functions that a leadership decision and a sales campaign can move in weeks. A hard pivot has a hard load of 5 or more out of 9. The coding for every case also records which anchor or lever moved first, whether the brand led the product, the company’s size, the years to first proof and the core asset kept.

Outcome coding

I judged each outcome three to five years after the pivot began, or at the latest date for recent cases, and tied it to one dated, numeric fact such as a revenue figure, an acquisition price or a shutdown date. A success means the new direction kept the business going. A failure means it did not, and a mixed result means the evidence points both ways or the clock has not yet run. Ten of the 22 pivots begun since 2019 are still mixed or running, so the AI-era rates will move as those cases resolve.

Results

Of all 75 cases, 51% succeeded. The rates in Table 2 split that figure by what the company moved and in what order, and the spread runs between 0% and 69%. The product result is the thesis of the book: companies that moved the product first succeeded in 21 of 33 cases, while companies that moved it halfway, with a new tier or a feature, succeeded in 2 of 12.

Table 2 · Pivot success rate by condition

Share of cases coded as successes. Hard and soft pivot rows cover the 63 cases scoring 5 to 13.

Company size changes how much the class matters. Startups did better the further they moved, and moves scoring 10 to 13 succeeded in 6 of 8 cases. Large companies ran the other way, with all four of their adjustments succeeding and moves scoring 10 to 13 succeeding in 4 of 12. The 28 successes with a dated proof point took a median of 2.5 years to reach it, two years for startups and three for large companies.

Table 3 · Pivot success by class and company size

Cases scoring 5 to 13. A soft pivot at a large company has no success on record: Netflix with Qwikster, Blockbuster, Weight Watchers as WW, tronc and Overstock as Beyond failed, and Meta, Peloton and Salesforce with Agentforce are mixed.

Two scores hold most of the risk inside the pivot band. Talent at 3 succeeded in 2 of 12 band cases, and finance at 3 in 3 of 10, while product at 3 succeeded in 12 of 16. AI-era pivots break the wider pattern, with 3 of the 13 hard pivots begun in 2019 or later succeeding, and Chapter 20 of the book sets out why.

Strength of the findings

The brand-first result is the firmest. None of the eight companies that moved the brand ahead of the product succeeded, against 38 of the other 67, and a Fisher exact test puts the chance of a gap that size arising at random below 1%. The product-first gap (21 of 33 against 17 of the other 42) and the hard-versus-soft gap (22 of 38 against 8 of 25) each carry a chance of about 6% to 7% of arising at random, just above the 5% line, so I treat them as strong directions and keep testing them as cases resolve.

I also ran the framework against its own data before publication. Removing any single case never reverses the gap between hard and soft pivots. Shifting a quarter of the finance scores up or down by one point at random, 2,000 times over, left hard pivots ahead in every run with a median gap of 25 points, and a hard-load cut at 5 separates the classes better than a cut at 4 or 6.

Limits of the data

The cases lean toward companies famous enough to leave a public record. Private startups that pivoted and closed without press coverage are under-represented, so these success rates run higher than a venture portfolio will see, and I plan venture bridges at a 25% success rate for that reason. I coded every case myself, and the one-line reason published with each score is the check on that: a reader who disagrees can see why I gave a score and score the case again. Many cells are small, and a rate built on six or eight cases gives a direction with a wide margin around it.

Your own pivot score

Score your own pivot the way I scored the database. Each executive rates the six anchors and levers alone and writes one line of reason for every score, every capital event goes through the capitalization test before finance is scored, and then the group argues out the differences. The Startup Pivot Playbook sets out the score, the order of moves and the five gates that release the money, and the paperback comes with the toolkit free. The free Pivot Field Guide runs a shorter version for a leadership team meeting.

Questions

What is the pivot database?

The pivot database is a set of 75 documented company pivots between 1892 and 2024 that Nitin Kumar scored on brand, target market, route to market, product, talent and finance and matched to dated outcomes. It underpins The Startup Pivot Playbook.

What share of pivots succeed?

In the database, 38 of 75 pivots succeeded, or 51%, with 24 failures and 13 mixed results. The cases lean toward well-documented companies, so a venture portfolio should plan on about 25%.

What should a company change first in a pivot?

The product. Companies that moved the product first succeeded in 21 of 33 cases, companies that moved it halfway in 2 of 12, and none of the eight that moved the brand before the product succeeded.

Do hard pivots succeed more often than soft pivots?

Yes. Hard pivots, which move product, route to market or finance, succeeded in 22 of 38 cases against 8 of 25 for soft pivots, and the gap widens from 18 points at startups to 48 points at large companies.

How long does a pivot take to show proof?

The 28 successes with a dated proof point took a median of 2.5 years to reach it, two years for startups and three for large companies. First paying customers usually arrive earlier.

Where can I see every case and score?

The full database, with each case’s six scores, the reason behind every score and the outcome evidence, is in the Pivot Playbook Toolkit, which comes free with the paperback.

To cite: Nitin Kumar, “The Pivot Database: 75 Scored Pivots,” Pivot Playbooks, October 2026, pivotplaybooks.com/pages/pivot-database. Method and limits as published in The Startup Pivot Playbook (2026).