What are the points to be aware of when selling new products called AI?

Talla is an AI for the customer support team, which allows students to learn by reading human-oriented documents without preparing special scripts, and answers various questions through chat. Rob May, who is the CEO of Talla, has learned what he learned from experience of selling corporate AI for two years.

14 Lessons Learned From 2 Years of Marketing Enterprise AI Bots and Knowledge Bases For Support Teams

As the question of "How much does it cost to update?" At the beginning when the purchase type package software replaces the monthly system SaaS , understanding of a new form of business does not progress easily.

Talla is AI, which means that it must be understood from a different viewpoint from the conventional SaaS. For AI to demonstrate its ability it is necessary to learn properly. So far SaaS has many free trial versions available, so you can purchase after checking what kind of tool you actually used. As with new employees, AI will take weeks to get to work well. Also, you should be aware that AI produces stochastic output like "... with a probability of 94% ...".

Based on the experience of Talla actually marketing AI, I'm summarizing 14 points on what I should pay attention to in AI's marketing.

◆ 01: Make content for educating people about AI
When an AI company interviewed potential customers, even if they asked potential customers for "a need to investigate about AI", they sometimes did not know about "what they can do with AI" at all is. However, it is a waste of time to tell all potential customers "What you can do with AI" every time. Therefore, Mr. May argues that instead of using content such as blogs, we should show what AI is like. Instead of using content such as blogs to show what AI is like. For example, Mr. May wrote a blog post called " The PAC Framework ".

◆ 02: Although it is not good to focus on "new and innovative" in advertisements, it is effective on the site <br> assume that the company's AI should be expressed on an extension of the conventional concept, or is totally new It is a difficult decision. Based on the conventional concept of "knowledge base" at Talla, I advertised with the message "It is a new kind of knowledge base" but it did not work. However, this message seems to have been good on its site. It is better to use "new" or "innovative" words on their own site and use another message in advertisements.

◆ 03: The advertisement "by AI ~" was not so good performance for advertisements using the word mediocrity "by AI".

◆ 04: "Automation" seems to have been very demanding requests from advertisements using the word "automation " which is very effective in advertisement.

◆ 05: Advertisement focusing on a specific use case is the worst "shorten the time to complete support response", "replicate the best reply so far", "slack it" ○ "efficient knowledge We focused on specific use cases, such as "shared to" advertising performance was worst.

◆ 06: Published in Venturebeat and MIT Tech Review <br> The amount of traffic from the above two sites is subtle, but it is said that those who actually introduce AI make it the source of information.

◆ 07: Retargeting advertisement is also good - In the performance of demo request, it seems that it is second in number following "Automation".

◆ 08: The free trial is not good <br> I tried a free trial for months, but I did not connect much to the actual purchase.

◆ 09: Bot app store is not that good
Slack and Microsoft Teams have a Bot app store that makes it easy to add bots. Through such a store there are many installations, but mostly it is fake, the fact that the number actually going to contract is not very high.

◆ 10: Events can be explained more more <br> Events seem to be more expensive than CPC advertisements at first glance, but events that can describe products for 30 minutes in front of potential customers rather than ads that disappear from the screen in 6 seconds It is said that it is more effective.

◆ 11: AI companies say "AI's marketing" and "AI's what to do" such as "support bot" and "corporate chatbot" that have a dual challenge such as AI Has not yet been unified. If you make a mistake in showing in marketing, you will have a layer that does not lead to purchase.

◆ 12: AI solves a large number of imaginary problems <br> It is rare for a customer to ask "What to look for in a new technology" what the customer is going to be doing well, and customers may "have such a problem" It is necessary to keep distance from the imagination challenge that I believe. It is necessary to let customers know that it is necessary to change the workflow in order to make AI work well.

◆ 13: Use marketing to update potential customers' ideas <br> Many people have false ideas about AI. For potential customers expecting AI to work like magic, let knowledge that it is better to structure the data to make AI work better.

◆ 14: Find key to trigger purchase <br> Users of traditional accounting software never purchase AI accounting software because of AI prediction function etc. "Better predictions" are not required. Instead of saying "1000 companies save 100 million yen a year" will be more effective points.

in Note,   Web Service, Posted by log1d_ts