Show all working

Where does business really get done, and deals actually get made? For what seemed to be an eternity, the simple answer has been a bit more complex than it first appeared. Certainly, some of the work might have been carried out in a potential client’s office, presenting well-worn, carefully crafted pitches and plans in meeting rooms. But was that the be-all and end-all? Far from it, as the economy of the long lunch (and even longer dinner) would often then take over, with a drop or two of alcohol loosening tongues and allowing both sides to properly size each other up before a final decision might be taken. Lunch and dinner were not celebratory ‘clinching the deal’ affairs, they were the real pitch.

So much so that most firms became extraordinarily good at this kind of thing, mastering the art of wining and dining clients and explicitly hiring staff who excelled at bonhomie and the other soft skills required to win a client over long past going home time. The trouble is that the world has changed quicker than most people expected, and for many firms looking to win new clients, that lunch or dinner might now come far too late.

Decisions are taken earlier than ever before according to 6sense, a sales-research firm that last year surveyed thousands of business buyers and found that 94 per cent of them had ranked their preferred suppliers before speaking to any of them, and bought from the one at the top of the list 77 per cent of the time. Most firms, in other words, no longer even make it to the dinner table and have little more than zero chance of deploying their most suave raconteur, whose skills now go largely unused. (In an age of upskilling and reskilling, this is clearly one of those groups of hitherto indispensable staff who might want to think about a career change).

It was not always like this, of course. For most of the last century the seller owned the evidence, whether it was an expensively-produced brochure, an impressive stand at a trade fair, or a loyal, hand-picked client who could always be relied upon to say nice things to potential customers. What changed the game was the internet (back when it was just that, ‘the internet’, a monolith that had yet to fragment into apps and platforms). It made due diligence and checking up on potential partners far easier than it had ever been, and if they didn’t check out, that expensive dinner counted for nothing. Firms no longer held all the cards, their potential clients did. As long ago as 2012 a survey by CEB and Google found that business buyers were already 57 per cent of the way through a purchase before contacting a supplier. Then came the review sites, from Glassdoor to Clutch, and the cards fell out of a seller's hands altogether.

ChatGPT, which first arrived in November 2022, then did something that few people noticed at the time (this correspondent certainly didn't): it reduced to a pittance the cost of adding polish. Companies across the world had spent two decades learning to look the part (such as developing great websites complete with flawless English, or whatever language their target market worked in). Well, that can now be done in a few minutes, and buyers know it. Michael Spence, who shared a Nobel prize for economic science with George A. Akerlof and Joseph E. Stiglitz in 2001 knew back then why this was of game changing importance. A signal (such as a firm from a non-English speaking country producing a pitch document or website in perfect English) only works if it is expensive (or at the very least difficult) for a weaker and less able rival to fake (Spence’s actual example was a university degree. But the comparison holds: a beautifully written and presented website no longer passes the test of being expensive or difficult. Anyone can do it).

A firm's reputation is increasingly being assembled by a machine from what other people have said about it.

What's more, the first reader of that website is increasingly not a person at all. Responsive, a software firm, found a year ago that a quarter of business buyers now use generative AI more than conventional search when researching suppliers, and nearly two-thirds use it at least as much, and these AI agents and chatbots are not altogether interested in what a company has to say about itself. Semrush, a marketing-analytics firm, found Reddit and LinkedIn among the five most-cited domains on ChatGPT, Google's AI Mode and Perplexity. A firm's reputation is being assembled by a machine from what other people have said about it (or, perhaps even worse, from the fact that nobody has said anything at all).

Borrowed credibility no longer fares any better, as the example of Builder.ai, a London start-up, demonstrates rather well. It made great play of Microsoft being amongst its investors who together put up 445 million US dollars only for it to collapse in May 2025 after its revenue turned out to have been inflated by as much as 300 per cent and it was revealed that the start-up had masked manual labour as machine learning. A wall of famous logos at the bottom of a website, once a must-have (and often faked), is now about as cheap a signal as there is.

Which begs the question as to what does work, and the answer is the rather mundane ‘get back to school’. Anything that shows working rather than just the answer (as every maths teacher, certainly in my experience, used to insist) is a very good, if not altogether exciting, start. Some software companies now publish ‘trust centres’ consisting of public pages listing their security certificates and audit reports before anyone asks to see them. Then there are those trust and rankings sites, as well as referees who will actually deliver a good reference if asked. (A word of warning though: Northwestern University's Spiegel Research Center discovered that the likelihood of purchase peaks at an average rating of 4.2 to 4.5 stars and falls as it nears a perfect five, which potential customers, not unreasonably, assume is too good to be true. That report looked at consumer goods alone, admittedly, but the principle is valid, and an honest account of a project that went wrong is worth more than yet another glowing testimonial).

For firms going beyond their own borders, this kind of thing is particularly important because fundamentally it is the difference between cognitive trust (I have seen your work) and affective trust (I like you, which is broadly what the lunch or dinner is for), something that Erin Meyer of INSEAD, a leading business school, has pointed out, not least when it comes to a country such as Germany, where business firmly runs on the former. Edelman's trust barometer found earlier this year that 81 per cent of people in Germany extend trust mainly to those close to them, and that trust in domestic companies is well above trust in those from abroad. For a software developer or business process outsourcing firm in Belgrade or Nairobi that means being proactive and doing the cognitive work up front without being asked.

None of which means the business dinner is dead (and thank the Lord for that). Affective trust still turns a first contract into a second, and there are few executives in the world who do not enjoy a long, wet lunch. But a seat at the dinner table now has to be earnt long in advance, mainly by offering (proactively and without being asked) evidence that it is deserved.