The History of CRM

By Victor Hoang · Updated Oct 24, 2026

Picture a salesperson in the early 1980s. Her customers live on index cards, a desk calendar and a good memory. When she changes jobs, or falls sick, the relationships leave with her.

Keeping records about customers is thousands of years older than the acronym. Merchants from Ashur filed their business letters on clay around 1900 BCE. A rotary card file was patented in the US on January 24, 1956. Pat Sullivan's ACT! shipped in April 1987. Salesforce was incorporated in February 1999 to deliver CRM through a web browser. The throughline: every generation of CRM moved the customer file out of someone's head and into a shared place, and every move created a new problem about who keeps that file accurate. The history of CRM is the long effort to make the file belong to the company.

Timeline at a Glance

The Ledger and Card Years (c. 1900 BCE-1950s)

Merchants keep the file, then a desk gadget keeps it

The scene is a trading town in central Anatolia, almost four thousand years ago. Merchants from the city of Ashur run a trading quarter at Kanesh, and donkey caravans bring in tin and textiles. Livius reports that more than 23,000 clay tablets from some seventy archives have been excavated there, most dated to about 1950 to 1840 BCE. Much of it is correspondence. A letter now catalogued by the British Museum on Google Arts & Culture was written by a merchant named Ashur-malik to his brother Ashur-idi and sealed in a clay envelope. Families and agents kept in touch with their trading partners by tablet.

Almost three thousand years later the same need appears in Italy. Francesco Datini of Prato, born around 1335, returned home in 1382 and founded a holding company of limited companies in Avignon, Florence, Pisa, Genoa, Barcelona, Valencia and Palma de Mallorca. The Datini collection holds about 150,000 letters. A firm spread across seven cities was run from one archive, and who owed what to whom was knowable only because someone wrote it down.

Those are precursors, not CRM. Nobody at Kanesh or in Prato tracked a pipeline, scored leads or reported a forecast. What they show is the oldest reason for the practice: when selling happens at a distance, memory is not enough, and the record becomes the asset. The same pressure, applied to a salesperson on the road, is the story told in the history of traveling salesmen, and the written offer that reached buyers who were never met is in the history of catalog selling.

The first mechanical design that looks like a modern sales tool arrived in the twentieth century. US patent 2,731,966, granted on January 24, 1956 to Hildaur L. Neilsen and assigned to Zephyr American Corporation of New York, describes a rotary card-filing device: a circular stack of cards on a rotor, which the user turns by a knob to bring a chosen card into view. It is the design later sold as the Rolodex. The patent's own filing header gives March 26, 1953, so the invention is older than the grant. A rotary file made one person's contacts fast to find. It did nothing for the next person.

The problem this era left was ownership. The card sat on one desk, and when its owner left the building, the company lost the relationships.

For teams today: the oldest CRM rule is also the simplest. If the record lives only with one person, the relationship belongs to that person. Ask what would walk out of your door if your best rep left.

The Contact Manager Years (1980s-1993)

A salesman builds the card file he wishes he had

The turning point is a personal computer on a salesperson's desk. In an Inc. interview in 2015, Pat Sullivan describes selling microcomputers in Dallas in the early 1980s and deciding "I ought to be using one." He taught himself to program and wrote tools for his own selling, one of them a contact manager. A friend with IBM and Computer Associates experience saw a business in the prototype. They quit their jobs, raised $100,000, hired two programmers, and, in Sullivan's words, "we shipped ACT! in April of 1987." That is his own account, given decades later, but it matches the shape of what the product became: Symantec's 1996 annual report describes ACT! as "an easy-to-use contact database with a graphical activity schedule."

What ACT! solved was retrieval and follow-up. A card file could find a name. A contact database could also remember the last call and schedule the next one. For a single salesperson, that was a large gain.

What it left unsolved was sharing. A contact manager on one machine made one person more organized and the team no more informed. Symantec's annual report lists the acquisition of Contact Software International, the maker of ACT!, on June 2, 1993, and says the deal added ACT! to the company's product line. By then, contact management was a product category big enough to be bought. The question the category could not answer was how a whole sales force could work from the same file. The related history of how managers organized and measured field forces before software is in the history of sales quotas and territories.

For teams today: personal tools are where good practice starts, and shared systems are where it has to end. If your best habits live in one rep's spreadsheet, you have a contact manager, not a CRM.

The Siebel Years (1993-2005)

Sales automation grows into an enterprise system

The turning point is the move from one desk to the whole sales organization. Siebel Systems was founded in 1993, and its first annual report on file says it shipped Siebel Sales Enterprise version 1.0 in April 1995 and version 2.0 in November 1995. The company described itself as a provider of "enterprise-class sales and marketing information software systems." By the end of 1996 it had 213 employees. The same filing acknowledges that many personal and departmental products already existed in the sales automation market.

Two years later the scope had widened. Siebel's annual report for 1998 describes "enterprise-class sales, marketing and customer service information systems," and uses the label Enterprise Relationship Management for the category. Net revenues for 1998 were $391.5 million, according to the same filing. The name is worth noticing: the category was still being named. By 2004, Salesforce's prospectus uses the label CRM and cites a July 2003 IDC report that the market for CRM software was approximately $7.1 billion in 2002.

What this generation solved was visibility: a sales manager could finally see accounts, opportunities and activity across a team, and marketing and service could work from the same customer record. What it created was a cost problem. The same prospectus says customers "have been reluctant to deploy traditional CRM applications because of the complexity involved in implementing them" and that enterprise software carries "a high total cost of ownership." Rolling out one of these systems was a project, and many projects stalled. How to avoid that is the subject of the CRM adoption operating model and CRM change management.

For teams today: a system that shows managers everything but gives reps nothing back gets filled with minimum entries. The Siebel era taught the industry that adoption, not features, decides whether a CRM works.

The Browser Arrives (1999-2006)

Subscription CRM challenges the installed system

The turning point is a web browser replacing an installation. Salesforce was incorporated in Delaware in February 1999 and began offering its on-demand CRM application service in February 2000, according to its own prospectus. The business model was a subscription: customers paid per user and logged in through a browser, with no server to install. By April 30, 2004, the prospectus counts about 9,800 customers and more than 147,000 paying subscriptions, and fiscal 2004 revenue of $96.0 million, up from $5.4 million in fiscal 2001. These are the company's own figures from its own filing.

The prospectus also sets out the case against the older approach. It predicts that "businesses will be able to replace their purchased software with subscriptions to a wide range of application services." The same bet is part of the broader move described in the history of cloud computing.

The incumbent response was consolidation. Oracle announced in September 2005 that it would buy Siebel, a deal Computerworld described as valuing Siebel at almost $6 billion, and completed the purchase on January 31, 2006. In the closing press release, Larry Ellison said "Oracle is now the undisputed leader in Customer Relationship Management software." Announcement and closing are different dates, about four months apart.

What the subscription model solved was access: a small company could now run a CRM without a data center. What it created was a new kind of sprawl. Once buying a CRM needed only a credit card, every team could start its own, and the shared customer file fractured again. That is the data problem addressed in CRM as a single source of truth and CRM data hygiene.

For teams today: easy to buy is not the same as easy to run. A cheap CRM still needs an owner, a data standard and a rule for who may create fields.

The AI Layer (2016-present)

The record starts to act on itself

The turning point is the moment CRM vendors began to embed prediction and automation in the platform itself. On September 19, 2016, Salesforce introduced Einstein, an AI layer across its sales, service, marketing and commerce products. Marc Benioff said "Einstein is now every customer's data scientist." Many features carried an added fee, and some came with existing licenses.

On October 29, 2024, Salesforce announced general availability of Agentforce, a layer on its platform for building AI agents that can act autonomously across functions such as sales and service, and can be triggered by changes in data, business rules or API calls. The shift in wording is the history in miniature: the 1956 card waited to be read, the 1987 database reminded a person, and the 2024 agent is described as taking the action itself.

The problem this generation inherits is the oldest one. An agent acting on a customer record is only as good as the record. Teams that never solved ownership and accuracy will find that automation spreads their errors faster. The practical steps are in CRM data hygiene, and how CRM differs from the lead-handling process around it is in lead management vs CRM. For where the record meets the buyer's path, see the history of the sales funnel, and for how selling at a desk produced much of this tooling, the history of inside sales. The operating discipline that grew up around these systems is covered in the history of revenue operations.

For teams today: before you let software act on customers, decide who is accountable for the data it acts on.

Key People

Key Facts

• Merchants from Ashur left more than 23,000 tablets at Kanesh, mostly dated about 1950 to 1840 BCE.
• A US patent for a rotary card-filing device was granted on January 24, 1956.
• Symantec acquired ACT!'s maker on June 2, 1993, according to its annual report.
• Siebel shipped Sales Enterprise 1.0 in April 1995 and reported $391.5 million in 1998 net revenues.
• An IDC report cited in Salesforce's prospectus put the CRM software market at approximately $7.1 billion in 2002.
• Oracle completed its Siebel acquisition on January 31, 2006.

The Pattern Behind the History

Our synthesis, not published research. Call it File, Share, Open, Act:

• File. The customer record starts as one person's private file: tablets, letters, cards.
• Share. Each system tries to move the file to a shared place: the contact database, then the enterprise system.
• Open. Each sharing step is followed by a cheaper, easier way to get one: the personal computer, then the browser.
• Act. The record shifts from something people read to something software uses.
• Repeat the problem. At every step, the weak point is whoever keeps the data accurate.

What It Teaches Teams Today

• Own the file as a company. The record belongs to the team, not the rep who made it.
• Buy for adoption, not features. The Siebel era showed that complex systems stall when users get nothing back.
• Easy access needs rules. A subscription makes starting simple and sprawl simple too. Set field and ownership rules first.
• Fix data before automation. Agents and prediction amplify what is in the record, good or bad.
• Keep the story in mind. Each wave made CRM cheaper and easier to reach. None removed the need for discipline.

To see where CRM sits in the larger story, read the history of sales. For the account-based model that depends on a clean customer file, see the history of account-based selling.