Statistical Validity in B2B Customer Satisfaction Surveys


Share on LinkedIn

I was asked today about the statistical validity of an InfoQuest B2B customer satisfaction survey.

Now we have a very neat spreadsheet for measuring statistical validity. The formula is made up of a number of parts, including the size of the universe, the number of responders and the percentage interval at which 50% were Totally Satisfied. Then out pops the magic confidence number. So, for example, if you had a total universe of 2,000 and you managed to get responses from 320 of them, then you would have a 94.98% confidence limit. Or, as pollsters say, a 5% chance of it being ‘wrong’.

But measuring statistical validity in a business-to-business customer satisfaction survey is just plain wrong on so many levels.

It assumes that all customers are equal in your eyes. They’re not (or they shouldn’t be). Unless you are selling a commodity, and all your customers trade under the same terms and conditions as each other; all receive the same level of attention, support, discount and payment terms, and there is no special packaging or extra stock-holding or additional training. And all your customers purchase more or less the same amount from you. Then they’re not all equal from your perspective.

Now lets look at your customers from their perspective. They all have different needs and wants. You may have segmented them geographically, or by industry sector. But where are they in their particular life-cycle? Are they part of a group, with internal political pressures? Do they have a new boss who wants to make her mark? Are the owners lining up the business for a trade sale in 18 months’ time? Are they looking to implement JIT and have track-side deliveries later in the year? Or consolidate down from a 3-supplier philosophy on critical items to just two suppliers?

These customers certainly aren’t equal. And they don’t deserve to be given the statistical validity treatment.

If you are looking to conduct a B2B customer satisfaction survey then make sure that whatever methodology you choose a) can give you a high response rate from your most important customers; b) doesn’t, I repeat does not comply with the MRS, MRA and ESOMAR codes of conduct (if your surveyor is a member of any of these bodies then you won’t find out who said what) and c) allows you to pose lots of questions and statements – thereby drilling down into the state of the relationship.

Treat each customer as an individual and they’ll respond to you. Treat them as just a number and you can guess the rest.

If you liked this, please share it.

And you may also care to read my very first blog – Treating Customers as Individuals

John Coldwell
From an operations background, John's attitude towards B2B customer satisfaction surveys is that they must be useful. Interesting doesn't interest him. You should be able to grab the feedback by the scruff of the neck and do something with it. For the past 15 years John has been running InfoQuest's full-day senior-team post-survey workshops around the world.