Larry Ellison

The pattern they actually represent

In 1970 IBM published a proposed theoretical model for a relational database. Oracle says exactly that, in its own annual report to the SEC. Oracle's thirtieth-anniversary timeline adds the part that matters: no company had committed to commercialising the technology.

In 1977 Ellison founded Software Development Laboratories with Bob Miner and Ed Oates to build it. That founding move is not invention. It is walking into a category the industry's biggest company had defined in public and declining to concede it.

He then did the same thing for forty years, in writing.

Oracle's fiscal 1994 annual report describes the sales motion without embarrassment. Prospective customers "often perform a detailed technical evaluation or benchmark of competitive products", so sales representatives were teamed with technical staff who "help customers run benchmarks against competitive products". Oracle had 3,124 sales and service people in the United States alone that year. The head-to-head was not something the company survived. It was something the company staffed.

The sharpest version came two decades later. In Oracle's third-quarter results for fiscal 2015, Ellison names salesforce.com, cites the new cloud business salesforce.com had itself announced it expected, and calls the year a close race. In the fourth-quarter release he repeats it with the arithmetic filled in: Oracle would sell more new SaaS and PaaS business than salesforce.com planned to, at a growth rate of around sixty per cent against around twenty.

Read the mechanism, not the volume. The claim is numeric, measured against a named rival's own published forecast, printed in a document Oracle files with the SEC, and checkable in ninety days. That is not trash talk. It is a confrontation with a scoreboard bolted to it.

The blockage it speaks to

Read this one if there is a competitor you will not name on your own website.

You know who they are, and your buyers compare you to them every week. Your page says "unlike other tools in the category". The same reflex shows up elsewhere: the customer asks for a discount and you concede it before they finish the sentence. You write "a significant improvement" because "forty per cent faster" is a sentence somebody could test.

You file all of that under professionalism. It is usually something narrower. A named comparison can be lost. A vague one cannot, because it was never scored. Every soft phrase there is insurance against being measured.

Here is what the insurance costs. The comparison happens anyway. Your buyer runs it in a meeting you are not in, with whatever they assembled on their own, and their version is worse than the one you would have handed them. You did not avoid the contest. You forfeited the half of it you could have shaped.

Ellison is the right study here, and not because he was combative. The record shows a third option to a reader who thinks the choice is between dodging the fight and being unpleasant. Make the contest specific, numeric and dated, and it stops behaving like a fight. A vague confrontation runs forever, because nothing in it can be settled. A bake-off ends on a Tuesday.

Three moves you can steal

Name one competitor, in writing, in a document a customer will actually read. A comparison page, a proposal, slide two. Name them, say plainly what they are good at, then state the specific case where you are the better buy. Naming the rival is what makes the rest of the page credible, because the reader has already made that comparison and now knows you have too.

Attach a number and a date to any claim you make about a rival. Oracle's line landed because it could be checked in a quarter. Do the small version: this migration takes two days, theirs takes two weeks, here is the log. A claim with a figure can be settled. A claim without one can only be argued, and arguing is the thing you were avoiding.

Offer the head-to-head instead of hoping nobody runs it. Oracle paid technical people to run benchmarks against rival products during the sale. Your version is free: send the prospect the test yourself, including the criteria where you lose. Being in the room when the comparison happens beats any claim you could make from outside.

Where the pattern breaks

Here is what running this pattern hard actually bills you. A selling culture leans forward. Pointed at a rival, that produces position. Pointed at your own quarter, it produces a number that has not happened yet. In September 1993 the SEC filed a complaint alleging that Oracle had issued inaccurate financial reports between August 1989 and November 1990 "due to inadequate internal accounting controls". The complaint did not allege any fraud. Oracle had already agreed, in February 1993, to settle shareholder claims for $23.25 million, and in October 1993 it consented to a judgment and a $100,000 civil penalty. That is the bill, and it reads as a design note: if you are going to compete out loud, build the counting to the same standard as the selling.

What to do this week

Pick the competitor you have been calling "other tools". Write one page that names them. Three rows: what they do better, what you do better, and who should pick which. The row where they win is what makes the other two believable.

Then take the strongest sentence on that page and put a number in it. Not "faster" — a figure you have measured and could show a stranger. If you cannot measure it yet, run the test this week and write down what it says, even when what it says is worse than you hoped.

Send the page to one buyer who is currently evaluating both of you. Not a broadcast. One person, who will read it, and who may come back with an objection you cannot answer yet. That objection is the most useful thing you will be handed all month.