Return to site

Mass of One

Tailoring services for some time was more of a marketing exercise than an actual practice. How can we convince people that our services our unique to each client when they might actually be the same product, just sold for the client’s unique needs?

For a short time, I was selling WebSphere to small and medium sized businesses in southern Ohio and Kentucky. It was a pivotal experience as there were only a small number of businesses that could use the product and they drove a very high percentage (beyond Pareto) of the sales. The first appointment I set up for our outbound WebSphere rep was at a pig farm. The IT director greeted the rep in mud covered hip waders.

WebSphere Application Server was an ephemeral construct. To most people, including us who sold it, it had no reality. Other products were a litte more tangible, like Commerce Server. It was like Shopify before Shopify… but you had to host it and maintain the software yourself. And the software cost $100K per CPU core and you needed at least two cores to run, oh, and you needed a backup server. Oh, and $300K in services to setup.

The key to getting any interest from prospective clients was to present WebSphere as some sort of manna. It could take the taste or shape of whatever the client needed and any of the sub products could be jammed in to fill in the rest. It wasn’t elegant but with clients who believed the adage “nobody got fired for buying IBM”, it could mean a lot of revenue for IBM.

Providing tools is one way of helping customize an offering to clients. The other is creating a singular process that results in a customized solution. For this, the current tools around AI now offer this, at least for software. Every Netflix screen is different, same with our Facebook feeds, every Google search takes into account our history.

Where this will soon take route is in our voice interactions. Today, the cadence and tone of Alexa or Google Assistant is the same regardless of speaker. Yes, you can set accent depending on location but it’s still fairly dumb. What we may find soon is that there might be “generic” responses for non-identified users and then tailored responses when the systems know it’s us. Maybe this is derived from sentiment analysis of our email, emotion in our voice, or other indicators.

The incremental cost of customization is getting closer to zero. The challenge is to ensure that the outcome of the customization process provides value. Part of that might still require setting user expectations upfront.

If all else fails, there’s always the PT Barnum test (providing the same response to everyone).

All Posts

Almost done…

We just sent you an email. Please click the link in the email to confirm your subscription!