Machine Learning Archives - Customer Strategy

November 13, 2024

A Big Question in AI: Build or Buy?

Maurice FitzGerald / AI, Customer AI, Customer experience, Machine Learning / Customer Experience, strategy /

AI continues to be a hot topic and just about every corporate IT team wants to be able to say they are working on an AI project. Looks great on their CV / resume. And I suppose a team that delivers even a small AI project will be rewarded more than a team taking care of that new accounting software, when it comes to annual review time.

However, if you manage a product, service, or major function such as sales for your company, the decision to build or buy AI software that will improve customer retention and growth is a big one. Should you buy something off the shelf, or have an internal team build it for you?

I could give you all sorts of data-driven, complex answers to this question and I would rather not. I like to use metaphors. In this case, let’s suppose you own a plot of land and you want a new house. Your extended family is enthusiastic about the new project, and two (of the 30 people on the list) have already worked in construction. Their message is clear “We know how to do this. We can get the whole family together to build it for you. The tools we can buy down at the DIY store are the same as the professional builders use. We guarantee that the result will be as good as what that builder you have in mind can do and we can do it a lot more cheaply.”

So . . . should you build or buy?

The questions to ask are fairly simple:

Have they built this type of house before?
When do you need it done? (You really should not believe it when they tell you can do it as quickly as external teams who have done it all before.)
Do you really think that your family using a bunch of fancy and often quite expensive tools for the first time can use them as well as professionals who have years of experience with those tools?
Does your family team understand all of the latest building regulations?
When it’s complete and you find things that need to be improved, will they still be around to do that?

Customer AI - Build or buy? — Created using Microsoft Bing Designer

The house-building metaphor has its limits in this case. An additional fact that matters a lot for your AI decision is this: According to research by Dr. Robert G Cooper about 80% of AI projects fail, mostly for “dumb reasons”.

I could keep going. The message is clear: there are things you simply should not do on your own if you lack experience and need results quickly. There are indeed exceptions, rare occasions where the level of detailed knowledge required to produce a great result is extreme and only your team has it. Such leading edge (probably meaning ‘bleeding edge’) cases are really rare. When you are planning your first major project in a new area, you generally reduce risks of all kinds by working with experienced third parties. Naturally, I am going to add that if you want to implement an AI solution that uses attitudinal, operational, and customer profile data to improve customer retention and growth, look no further than Customer AI from OCX Cognition for your first major project in the space.

Notes

OCX Cognition predicts customer futures. Our breakthrough Customer AI solution lets enterprises transform what’s possible in customer experience. Reduce your customer risk, break down silos, and drive speedy action – when you can see what’s coming, you can change the outcome. Building on more than 100 years of CX-focused expertise in our small team and thousands in teams we have led, we’ve harnessed today’s advances in AI, elastic computing, and data science to deliver on the promise of customer-driven financial results. Learn more at www.ocxcognition.com.

Maurice FitzGerald is Editor-in-Chief, Content at OCX Cognition. He retired from HP where he was VP of Customer Experience for their $4 billion software business and was previously VP of Strategy and Customer Experience as well as Chief of Staff for HP in EMEA. He and his brother Peter, an Oxford D.Phil in Cognitive Psychology, have written three books on customer experience strategy and NPS, all available from Amazon.

November 12, 2024

Sometimes, executives answering surveys have no idea what they are talking about (Discuss!)

Maurice FitzGerald / B2B, Behavioral economics, Blog, Cognition, Customer experience, Customer Satisfaction, Machine Learning, NPS, Psychology, Strategy / Customer Experience, strategy, Surveys /

There is a somewhat strange phenomenon that I have observed at an anecdotal level over the years. It is limited to B2B surveys and causes all sorts of problems. Or at least that’s what I think, and I would like to understand your views too. Here is what I want to discuss:

When we send relationship surveys to B2B customers, we always try to identify key decision-makers and decision-influencers and we ask them to answer a set of questions. This can be done in person, remotely, or by sending a link to a web-based survey, for example. Let’s imagine a situation where a company selling ERP software believes their customer’s Vice-President of Procurement is a key decision-maker. The company sends a survey request out by email and the procurement VP decides to respond. The survey follows a fairly standard format of an NPS question, a couple of open questions about what should be improved, and then a set of 12 questions asking the procurement VP to provide scores about product reliability, the contact center experience, how easy the software was to learn to use, and so on.

This person may have no idea what they are talking about

You may already see where I am going with this. The procurement VP does not use the software and really does not know what is going on. Nevertheless, they have decided to spend five or ten minutes filling out the survey, so they are going to give you scores for everything. Uninformed scores. Wrong scores. Scores that probably correspond to their overall opinion of your company rather than the actual questions being asked.

Then you decide you need to take action on the bad scores, even though your performance may already be perfect.

In this case, the VP is probably out of synch with the rest of the team, especially those using the ERP software every day. These are the people who have useful suggestions, though they may have only a minor role (or perhaps none at all) in deciding whether or not to renew the contract with the software.

Here is my current opinion

In my possibly-wrong opinion, these scores are therefore a relatively accurate predictor of attitudes and future purchasing behavior at the aggregate level (as is ther overall relationship NPS score) but are useless at the level of the individual questions, in terms of telling you which operational improvements you should prioritize to address those attitudes.

Of course you may well find purchasing VPs who have good knowledge of most operational touchpoints, but I think those are rare people in large companies.

I’m not totally certain what to do about this. Do you have some ideas?

I believe the major customer-touching teams in your company do all need input about what they could do better. This issue is that the input needs to be more accurate than it probably is at the moment. For the overall company relationship surveys that are answered by high-level executives, I doubt that adding a request not to answer the driver questions if you don’t know what you are talking about would be a good idea. Maybe we should supply the procurement VP with relevant operational data? Maybe we should accompany all relationship survey requests with the results and trends from any and all transactional surveys we may do?

For transactional surveys, the main issue we all face is that response rates are very low, so we can’t be sure they are accurate. That’s where analyzing operational data trends for all the customers helps. If the operational data numbers and trends don’t match the transactional survey results, I have to say I believe the survey results are wrong, probably because you are not getting the correct people to take the surveys.

So, what do you think? What should we CX and Customer Success experts do about this horrible situation? Yes, I have some ideas ideas. We need more solutions. Please help!

LinkedIn version of this post here

Notes
OCX Cognition predicts customer futures. Our breakthrough Customer AI solution lets enterprises transform what’s possible in customer experience. Reduce your customer risk, break down walls between teams, and drive speedy action – when you can see what’s coming, you can change the outcome. Building on more than 100 years of CX-focused expertise in our small team and thousands in teams we have led, we’ve harnessed today’s advances in AI, elastic computing, and data science to deliver on the promise of customer-driven financial results. Learn more at www.ocxcognition.com.

February 4, 2022

Don’t Look Ahead!

Maurice FitzGerald / Behavioral economics, Blog, Cognition, Customer experience, Machine Learning, Psychology / Customer Experience, Journey mapping, strategy /

Does it exist? Does looking ahead exist? Ever since PhD candidate Richard Curry and his supervisor Brian Owen created Spectrum AI software, the controversy has been growing. And the Don’t Look Ahead movement has become more and more powerful. What does this mean for the world?

As soon as Richard and Brian completed their analysis of the first live customer data, they rushed off for a meeting with Fortune 500 President Meryl Hannity. The President’s team had requested the meeting and agreed that it was critical. They were surprised to be kept waiting outside while the President celebrated a birthday in her team and met with some important social media personalities. Finally, the next morning, they met. Richard and Brian laid out their case, proving they could look ahead, predicting future customer retention and NPS with 84% accuracy.

“Ah…” said the president’s Chief of Staff Novax Jokeovich, “So you can’t prove looking ahead works. And it is inaccurate. This means it does not exist.” President Hannity nodded her agreement and then had to rush off for an extremely urgent meeting with Chief Flower Arranger Khloé Minaj.

Despite never watching television, Curry and Owen decided they would use the most popular information channel in America to get their message across. It did not go as planned. They were frustrated by the constant questions about views expressed on the topic by social media scientists and by Doctor Facebook. Did they choose the right TV network to get their message across?

As trusted science star Gwyneth Kardashian pointed out recently, the only reliable news source and indeed the most popular source on the planet is Wolf News. Here is what reliable Wolf News spokesperson Tucker Cruz had to say on the topic after the interview: “Obviously, looking ahead to understand customers is impossible. The future does not exist yet, so we cannot possibly look at it.” There you have it: the definitive answer. It simply can’t be done.

Despite the Wolf News disaster, the Look Ahead message gained momentum. More and more customer scientists demonstrated that using the technology does indeed identify customers who want to change suppliers and allows actions to be taken before they leave. The customer scientists contrast this with the use of surveys, saying that these only identify what has already happened, and then only for the small fraction of customers who respond. The reaction to this position has been strong.

Fortune 500 President Hannity’s staff united behind a public position then unanimously voted in the Don’t Look Ahead Act. The act notably includes a statement that the second amendment to the Fortune 500 constitution includes the right, even the duty, to “bear surveys.” As they say, “Surveys were good enough for our grandparents and our parents. They are good enough for us. We have a moral obligation to examine our past, and only our past. Don’t Look Ahead!”

President Hannity herself continues to rely exclusively on surveys. Yes, it is true that she is constantly surprised by clients who choose not to renew their leases with her real estate company, but as she says, “After leaving, they had the courtesy to tell us why they left, and we are working on those problems. The biggest problem is not what the customers are saying. The biggest problem is the ridiculous notion that you can look at operational data to read the minds of customers and predict what they will do before they do it. I have a great mind. People tell me mine may be the greatest mind of all time. Even I can’t do this, so it is impossible. Richard Curry and Brian Owen have published fake science, the fakest of all science. Don’t listen to them.”

Now the protests have reached the streets, chanting “Don’t Look Ahead! Don’t Look Ahead!” The future seems uncertain. Where will things go from here? Let’s use Spectrum AI to find out.

News Flash

We understand that Meryl Hannity has agreed to play all roles in the forthcoming blockbuster based on this story. “We find we make a great team, learning a so much from each other,” she explained, looking in the mirror.