The document provides guidance on when to use web, apps, bots, and agents from a marketing perspective. It recommends using the web when ubiquity, speed, information delivery, or existing infrastructure can be leveraged. It suggests using apps when context, engagement, accessibility from a single source, or rich interactivity are important. Bots should be used when existing customers or large audiences need service or attention. Agents are best utilized when integrating with other platforms, differentiating on those platforms, voice is sufficient, existing payment solutions apply, or simplicity is key.
6. My goal
Provide some clarity about when you should use:
- Web
- Apps
- Bots
- Agents or assistants (but not bots)
I’m doing this from a marketing perspective:
- Partially because this technology is often utilized by marketers
- Partially because “marketing” is now part of everything a company does
7. - Write for Forbes, VentureBeat, Business Insider
- Mobile Economist for TUNE
- Marketing attribution, tech in 3.5B devices globally
- Founded the research division of VentureBeat
- Focused on martech
- CEO of Sparkplug9
@johnkoetsier
johnkoetsier.com
John Koetsier
8. Use web when ...
- You want ubiquity
- Base layer
- You need speed
- Friction
- Info dump
- You are bundling multiple media in one experience
- You can capitalize on existing infrastructure
- Payments
9. Use apps when ...
- You want context
- Location, time, activity
- You need engagement
- E.g, games, media, social, content creation
- You need accessibility
- One layer from HERE
- You are providing rich, fast interaction
- Full interactivity: image/video editing,
10. Use bots when ...
- You have existing connected customers
- Fans/followers and customers; need service
- You can get attention
- Robot Lawyer
- Christopher bot
- You need one-to-one at scale
- You have a simple, bounded problem
- Most bots are stupid, so ...
11. Use agents/assistants when ...
- You can integrate into someone else’s platform
- You can differentiate on someone else's’ platform
- Voice is enough
- You can use an existing payment solution
- You need simplicity
- Base layer
- Payments
- Connectability with simple links
One of the biggest challenges with new technologies: where do they fit; where do they work, where do they not work…
Websites … brochures
Apps … websites
Bots … apps
We’re going to go a little bit fast -- not much time -- but I’ll post this on my website so you can get the full presentation.
March 2016: Satya Nadella
Image: https://www.slideshare.net/coreindustries/bots-are-loving-you
Satya Nadella is a very smart man, and he’s doing amazing things at Microsoft … but …
That is bullshit
Bots are many things, but they are not the new apps
In fact, they’re very different than the new apps
Image from Wikipidia: https://upload.wikimedia.org/wikipedia/commons/a/a1/Bullshit_46346.jpg
Marketers always jump on the latest tech
But they’re also always very confused about the latest tech
Probably because there’s so bloody much of it … 3,500 to 4K martech companies
I’ve talked to Scott Brinker … new one has 5,000
When bots first came out, there was a ton of hype
Now, we are in the trough of disillusionment on bots
Both extremes are silly
This is from Botanalytics … notice the large 57% retention rate for “commonly used entertainment bots”
Actually … kinda cooking the books
Check the smaller print: retention rates CAN be as high as 10%
Average is 4%
Image via Botanalytics: http://venturebeat.com/2017/03/20/these-are-the-most-important-chatbot-metrics-to-track/
Added agents
Wasn’t in my program description
Felt it was important to add
If you don’t like it or disagree … sue me :-)
And … btw … I differentiate:
Bots: simple, text, constrained, verticalized …
Agents: complex, platform, text & speech, broad, connect multiple functions
Feel free to agree/disagree later in question period :-)
OK, let’s start with the web
Still a very viable choice for development and delivery!
Ubiquity: any device, any platform, any message, any location … the web is the base layer of technology
Speed: remove friction (apps: install; bots: platform), info dump … web is good at longform
Multiple media: apps can do this too, although integration of disparate sources of data/media not as easy to accomplish. But bots struggle at image galleries, videos, comparisons, complex layouts, interactive charts/graphics
Ease: two examples … payments: Apple Pay, Android Pay; connectability … simple links … getting that in apps and bots: deeplinks and universal links, but need an additional technology layer
And now … apps
A bit out of favor … we’re past the gold rush stage … have major competition with incumbents with entrenched positions
But again, still VERY viable
Context: location, time, activity
Engagement: deep, intense immersion over time (games … media … social … content creation … etc.)
Accessibility: we live in a mobile-first world, so phones are the default device. Apps are the fastest way to X: open phone, tap icon … one layer from immediate. Bots are 3 layers: phone, messaging app, conversation
Rich interaction: full interactivity
Distribution: app stores are crowded and busy and we all use 5-8 apps mostly … but there’s still no other distribution method on the planet that can get content & functionality into the hands of tens of millions as quickly (possible exceptions: web and social platforms). App installs have slowed, but not stopped, and not everywhere.
Finally … bots
Huge opportunity … but not the way many have built them
Existing customers: incentive to work with you, need help from you, will seek you out. Important b/c discoverability is still a major problem for bots (e.g., KLM)
Attention: Discoverability is a challenge, but if you have something awesome, people will seek you out (e.g., the world’s first robot lawyer … maybe the most successful bot to date. important note: that bot is built on the web.) Other example: the homework bot … Christopher Bot … lots of media attention, but under 1,000 likes
One-to-one: Chris Messina, inventor of the hashtag … request his schedule, where he’s speaking, etc. Big brands even more so of course.
Simple, bounded problem: this is the most critical one right now. Most bots are really stupid, and the huge proliferation of bot-building tools out there is ensuring that this will continue to be the case for some time yet. Avoid the comprehension problem by offering a bot for a very specific purpose. Example: MobileMonkey, which does nothing but book appointments for professionals like dentists and doctors. Simple, defined problem with limited capacity to screw up, but also an important challenge.
And … agents/assistants
Also a huge opportunity … also just beginning to see what’s possible here
Integrate: You’re not going to build your own agent, b/c you’re not Amazon, Google, Apple, Microsoft, IBM, Facebook, Baidu, etc. But if you can integrate into Alexa, Siri, Google Assistant … you can win. Kayak: vacation discovery via Alexa … “where can I go for $500?”
Differentiate: Agents are going to drive a huge percentage of commerce … how will you be set up for success? Why will they recommend you? Relationship with platform owner, engagement and retention in skill/action/integration, kickback or retail markup, volume of $$$ you drive. This is early days still … but start sooner rather than later, learn faster, become #1.
Voice: Turning up the temperature, adding two numbers, or getting the capital of Kazakhstan are good voice-first problems. Comparing two or more cars that you’re considering buying is not. Pick agents/bots/apps/web for problems that they can solve.
Payments: if you’re re-ordering washing detergent, you can re-order via Alexa and Amazon knows how to take your payment. Can’t say the same for most bots. Payments built into Messenger, sure, but how many people have given FB their credit card?
Customers: true for all of the previous slides … are your customers early adopters of one or more technologies? You should know … and you should get where they are.