Design Management Group

Case study - Palm Pilot

The Palm Pilot was the first successful interconnected organiser, heralding the dawn of the billion-dollar handheld industry. The brainchild of Jeff Hawkins, it was in fact his third such project, and the success of the Pilot clearly owes much to his earlier experiences:

• Project 1 - the GRiDPad
• Project 2 - the Zoomer
• Project 3 - the Palm Pilot

The GRiDPad

In 1986 at the age of 29, Hawkins had left his employment with computer company GRiD for a one-year stint at UC Berkeley, pursuing an interest in the workings of the brain. During this time, he developed some software for recognising hand-printed characters. He returned to GRiD as VP R&D, and was soon involved in a project to produce the GRiDPad, a pen computing device which used Hawkins' PalmPrint software, now licensed to GRiD. The GRiDPad used off-the-shelf components to minimise cost and development time, weighed in at 4.5lbs and sold for $2500. By 1991, Microsoft had announced Microsoft Windows for Pen Computing, and pen computing was being hyped as 'the next big thing', although the GRiDPad was still the only such product in existence.

The Zoomer

Even so, Hawkins felt that GRiDPad was the 'wrong product'; it was too big, too expensive, and aimed at the corporate market. What was needed was a smaller, sub-$1000 product aimed at the consumer market. Hawkins codenamed this idea the 'Zoomer'. By November 1991, Hawkins had resolved to leave GRiD to found a start-up which would develop the Zoomer concept. With $1.3m funding, Palm Computing was born in January 1992.

There were four initial project partners: Zoomer would use an operating system based on GEOS from GeoWorks, Palm would produce the application software, Tandy would distribute, and Casio would manufacture. AOL and Intuit joined the project later. With six partners in the project, rampant 'featuritis' (feature creep) started to become noticeable. Hawkins' attempts to resist 'feature bloat' were to have a formative effect in reinforcing what was to become Palm's 'less is more' philosophy.

The Zoomer appeared in October 1993, shortly after the unsuccessful debut of the Apple Newton. 20000 units were snapped up by the 'early adopters' in the first two months; then sales fell off sharply. Tandy and Casio lost interest and ceased promotion. Although the market may have been prejudiced by the poor showing of the Newton, in reality the Zoomer too was flawed - too big, too slow, too expensive (at $700).

Palm was somewhat surprised to find that 50% of Zoomer users already had a PC and when the PalmConnect connectivity software was launched, the uptake was high. It began to emerge that users considered the device as an accessory to a PC, rather than a PC replacement. All those extra Zoomer functions were superfluous

The Zoomer II

Palm assumed there would be an updated version of the Zoomer, and set about speeding it up with a faster processor. The user interface was streamlined and simplified. Rather than improve the PalmPrint software, Hawkins proposed, in the face of much scepticism, to adopt a new approach which required the user to learn to write in a way which software could interpret (codenamed PowerPalmPrint, later to become Graffiti ). But even with significant improvements to the speed and design of the software, Casio were not convinced and pulled out, effectively killing the Zoomer II.

Palm still had enough funding for another 18 months, although one option was simply to give up and return the investors' money. Palm's mission was to be the No 1 supplier of software to the handheld industry - but no one had yet produced a suitable handheld device. A sympathetic investor suggested that perhaps Palm should take the initiative and develop the whole product itself. This flew in the face of 'conventional wisdom', but for Palm there seemed to be no alternative. Hawkins was secretly thrilled at the prospect of being handed a clean sheet of paper. Project 'Touchdown' was born.

(Palm Pilot: image courtesy palmOne) Images of earlier products GRiDPad from Jerry's computer museum Tandy Zoomer from Zoomer PDA website Casio Zoomer from Pen Computing website

The Palm Pilot

Hawkins enumerated the characteristics of an ideal product:

• Cost: $299 max
• Size: fits in shirt pocket
• Operation: simple; aimed at average consumer, not 'geek'
• Synchronises with PC, does not replace PC
• no hourglass cursor (software should be fast enough not to need it)
• no error messages (system should be robust)
• minimise number of stylus taps required
• simplicity and immediacy (delight the customer)

Given the size target, he carefully reviewed the implications for the product architecture. There would be space for no more than 2 AAA size batteries, a 160x160 pixel screen, and most significantly no keyboard - the device would depend on Graffiti. In a few hours, Hawkins produced a balsa wood scale model with a pasted-on screen mock-up, complete with stylus whittled down from a chopstick. In the weeks to come, Hawkins would be seen interacting with his wooden model, pretending to use it as if it were a real device. (The model was to prove extremely close to the final Pilot form factor). In essence, the brief was for a simple machine, with a small, fast powerful, flexible operating system - Palm OS. The team duly became experts in killing unnecessary features.

Local design firm Palo Alto Design Group (PADG) offered to design the device in exchange for a reduced fee and some Palm stock. The initial models were 'boxy' and PADG assured Palm that later designs would be more 'curvy'. Hawkins disagreed - curves take extra space, and the size budget was one Hawkins would defend aggressively.

Palm had 27 people, $3m and 18 months to deliver. (Apple had had hundreds of people, and took 5 years and $200m to deliver the Newton). Palm could not afford to recruit so it created a virtual team with 2 external hardware engineers, PADG and Flextronics, who would manufacture the device in Singapore.

Once more Hawkins wanted to use existing off-the-shelf components where possible, although Palm had selected the Motorola Dragonball processor chip, still under development. To keep costs down, Palm preferred to use near obsolete PS-RAM memory chips which were one-third the cost of state-of-the-art chips. Palm had to plead with Toshiba, the last remaining supplier, to hold off shutting down the assembly line at least until Palm's product launched.

Hawkins was passionate about preserving the product vision, so when the hardware team announced that, due to component interference, the device thickness would have to be increased by 1mm, Hawkins refused to accept it. A workaround was duly found. By early Autumn 1995, the team had achieved virtually all of its design goals - the device would fit in a shirt pocket, the costs would permit a $299 sales price, and the UI was simple and responsive.

Funding the final phase

Although Palm had enough cash to complete the software and hardware development, it would need another $5m to cover manufacturing and marketing for the first few months until revenue from sales kicked in. Palm's VC investors suggested the involvement of an established corporation such as Motorola, Ericsson and Compaq. Another potential investor was US Robotics (USR), whom Palm had approached to produce a snap-on modem for the Touchdown.

USR's response took Palm by surprise - rather than invest, USR offered to acquire Palm. This was not what Palm had sought, but maybe it was the only way to get the product to market. After much deliberation, Palm indicated that it would agree to acquisition on the understanding that Palm would continue to have independent status and responsibility for the Touchdown project. In August 1995, a figure of $44m was finally agreed.

USR initially wanted Palm to abandon Flextronics and manufacture at its facilities in the US. Palm insisted that such a change would disrupt the schedule and that the USR facilities were unsuitable. In a compromise, Flextronics would manufacture in Singapore and ship units to Utah for packaging.

Launch preparations

By early February 1996, the Flextronics plant in Singapore had started to produce the first 10000 units. As the first units arrived in the US, a string of problems started to emerge: the springs which retained the batteries were becoming deformed leading to loss of electrical contact. The problem was quickly fixed but the schedule slipped. Then some units coming into Utah were 'dead-on-arrival' (DOA). This was traced to a glitch in the new Motorola processor. A software workaround was created but the shipment schedule slipped another month. Then, an 'own goal' was discovered - a piece of 'Easter egg' code (code which generated an amusing display if an obscure key sequence was executed) was erroneously generating its display with unintended frequency. This too was patched.

Then some more units arrived DOA. Production was shut down while the fault was identified. This time, the fault was traced to the recent inclusion of a red sticker on the rear of the battery door, warning users to change batteries quickly to avoid data loss. Unfortunately this label was 1mm thick and when the door slid shut, it was making contact with the batteries, causing them to disconnect from the battery terminals. (Ironically, if Palm had been able to keep their original plan of just using Flextronics, these faults might not have been picked up prior to customer shipment).

Finally, on April 9 1996, the first units were ready to ship to customers. In May, sales held up, and stores began to sell out. Sales dropped off over summer, but recovered in the run-up to Christmas. By the time the 1996 Christmas season was over, Palm had reached over 70% market share in the US, winning over 20 'best product' awards on the way.

(This account is primarily based on information from the book 'Piloting Palm' by Andrea Butter and David Pogue).

Further information

• Butter, A. and Pogue, D. (2002) Piloting Palm: The inside story of Palm, Handspring, and the birth of the billion dollar handheld industry, Wiley, New York
• Jeff Hawkins by Shawn Barnett, Pen Computing Magazine
• Palm historical timeline from palmOne website www.palmone.com
• Palm backgrounder from palmOne website www.palmone.com
• Dillon, P. (1998) The next small thing, FastCompany Magazine, Issue 15, p97

All trademarks acknowledged.