The baseline moves.

Before you talk to a company, you already expect something. Deliver less and people talk about it, in a bad way. Deliver exactly that and nobody says a word: there’s nothing to tell. Deliver more and people tell their friends.

The catch: that expectation is not fixed. It moves every time someone has a good experience. Anywhere.

Who sets your customer’s standard?

Model · 4 min readBenchmark against the best, not your peers.Shows where the bar really is: against the best, not the brand next door.Not your competitor. In 2011 I wrote that it is hard for customers to understand why their travel agency solves a problem on Twitter within thirty minutes, while their bank takes two weeks to answer an email. A customer doesn’t keep separate scores per sector. The best experience they had last week becomes the floor for everyone.

You see it everywhere. KBC Mobile has been named the best banking app in the world, more than once. Every other app in Belgium now gets measured against it, banking or not. If a taxi or dinner arrives a few taps after you open Uber, waiting months for the government feels absurd. And the crew on your flight is compared with the waiter who remembered your name last night, not with another airline’s crew.

Kids show it best. A child who grew up swiping an iPad to watch what they want, when they want, looks at a TV that won’t swipe and a programme that starts when it starts, and doesn’t understand why.

At InSites Consulting we saw it in a satisfaction study for a client (I still can’t say which one). The client hadn’t got any worse. Their scores dropped anyway. New players in their market had invested in smart digital tools and great service, and the baseline had moved past them.

Today there’s a name for this: liquid expectations. Expectations flow from the best experience people know into every other category. Accenture uses the term; it’s a useful label for something that was already in this piece.

Fast beats good.

In 2012, brands had just taught people to expect an answer on Facebook and Twitter within a day. Research from NM Incite, written up in Harvard Business Review, showed how fast that went: more than half of Twitter users expected a reply within two hours. And 33% would recommend a brand that answered quickly but didn’t solve the problem, almost twice as many as for a brand that answered slowly but did solve it.

I liked that. It meant you were allowed to fail, as long as you showed up. Most people would rather see you fail than see nothing at all. The companies that made mistakes in public learned fast, and every case they solved widened the gap with the companies that weren’t there yet.

KLM was one of the first to take this seriously. In 2012 their social team passed 20,000 interactions a week. At the peak it was 120,000 to 150,000 questions a week, a volume no team can keep up for long. The baseline didn’t just move. It multiplied.

AI raised the bar and lowered the floor.

Now the first reply comes from a bot, instantly, at three in the morning. Speed is solved. And the result? The Dutch consumers’ association, the Consumentenbond, asked more than 10,000 people about their last chatbot conversation with a company: 71.2% got no answer or no help at all. Only 10.3% were helped.

KLM started using AI in its social customer service in 2017, and was open about what it learned. The fastest replies were AI suggestions an agent could send as they were. Manual replies came second. The slowest were AI suggestions an agent first had to fix. AI only saves time when the first answer is right.

So the baseline moved twice. People now expect an instant answer, because that is what AI promises. And they are used to that answer being useless. The bar went up, the floor went down. The space in between is where you can stand out: fast, and right, and with a person when it matters.

Part of getting it right is knowing which question can wait. At KLM we talked about this a lot. A cancelled flight or a rebooking needs an answer now. Whether a plane has a horn can wait. Most companies still sort on something else: a phone menu, or how much you spend, with a Platinum line for the best customers. That decides who gets helped first, not what needs help first. Sorting on urgency is still hard, and I haven’t seen anyone do it well yet.

Updated by Polle in October 2026. Two posts in one: “Managing rising customer expectations” (June 2011) and “No web relations? You’re screwed.” (June 2012). The tools and names from back then are gone. The point is still here.

Sources: Consumentenbond, “Chatbots helpen klanten vaak niet verder”, 26 August 2026. Benmark and Singer, “Turn customer care into social care”, HBR, December 2012. KBC newsroom on the Sia Partners ranking. Goes well with: Managing expectations and Acts, not ads.