Read Part 1 and Part 2 before reading this post.
By now you can understand how your intelligent shopping agent will get its data. You can hopefully also imagine its predictive technology, anticipating what you want, when you want it, and where you should buy it, well before you even need it.
OK, there are many other features of this intelligent agent, but I just want to focus on one more, the purchase process, before tying this together with some ideas of the business model.
Your intelligent agent will have your payment information and permission to automatically make purchases. If you’re not comfortable with that, there will be an option to opt-out of this automation, but it’ll be an opt-out; the general user of the agent will trust it because of how well it knows you.
Your agent will know your buying habits and be hooked up to your bank and credit card accounts, so it will know what you can spend on a purchase. But this doesn’t mean it will willy-nilly spend whatever to get a product for you. That’s unacceptable. Your agent will haggle on your behalf. CUC International (now Cendant Corporation) owns a patent for Hagglezone and once had a site which featured the haggling technology. Read this NYTimes article (which even mentions Mercata.com – oh, the good old days of ecommerce!) from 2000 on how this worked. Merchants aren’t currently setup to haggle, but merchants could be part of the intelligent agent preferred marketplace powered by the haggling technology. While your intelligent shopping agent will always scour the web for the best deal and could have hooks into distributors and manufacturers, it could feature a marketplace (just like the old www.hagglezone.com) with select merchants who want to be part of this bazaar and get first crack at the consumers. In the Hagglezone, your agent could always just set an offer price and wait – as it will know the urgency of your need, it doesn’t need to transact immediately. And if the agent is working for thousands or tens of thousands of consumers, it could haggle or negotiate a great group buying price (back to the Mercata model).
So what does the intelligent shopping agent cost? That’s the great part, it’s free to engage with the agent. There’s no $999/yr price tag. While there could be premium features you could add to the basic model, that’s not going to cut it. There are two possibilities for the revenue stream.
1. Your will pay your intelligent agent a concierge fee for making your life easier. Basically, it will take a very small % on top of each transaction. And I really do mean a small % because if we’re talking about everyone engaging an agent at some point, those pennies can add up very quickly. The % fee would probably be on a sliding scale, so if you spend a lot through the agent, the effective commission you pay will go down.
2. Your intelligent agent will earn a commission from the retailers it engages with.
I think I prefer option 1. I think we’ll get to option 1. But option 2 is already a proven entity. It’s just a typical marketplace. Consumers (through the agent) will get what they want and retailers are willing to pay for that targeted business. It’s lead generation 101.
And if we go with option 2, you can also start to see that this is all easily extensible past the retail space. The intelligent shopping agent should be shopping for your auto/home/health insurance policy, mortgage, student loan, online probability course, phone/internet/cable service, lasik eye surgery professional, personal trainer, SAT prep course, and much more. Anything that the lead generation industry has gone after is a perfect opportunity for your agent. When [insert the service] compete, you win, and so will your agent because it will get paid to make the introduction.
I’m sure a lot of what I’ve written sounds vaguely familiar. There’s a lot of basic online marketing/lead gen in these posts. There’s a bit of science fiction. And there was even a company called Root Markets way back when which I thought was hoping to be your intelligent agent, but it seems that I warped the idea a lot. At its core, Root offered a secure locker of your personal information and you choose to share as much or as little of that with marketers. If you shared a lot, you’d get lots of offers because Root will be able to facilitate a lot of ‘introductions’ to various companies competing for your extremely targeted business. If you shared a little, you’d still get offers, but they probably wouldn’t be as relevant/targeted, which means they’re less valuable to you, to Root, and to the advertiser, a lose, lose, lose situation which will push you out of the market. Or at least that’s what I remember.
There are a ton of holes in my intelligent agent. And as my first real blog posts in a half a year, I know this writing is not my best. Don’t worry, I’ll get back into a flow.
Now, it’s time for you to comment. It’s time for you to tweet. It’s time for you post a link on your wall. (Yes, I’ll bake those buttons in soon.). Thanks.
[...] Read Part 3. [...]
Thanks for the great thoughts here Brian. It seems plausible that some day an agent of this complexity could exist but I suspect it’s a long time before it would look like such an integrated service.
There is a tendency to see certain aspects of technology and assume ubiquity quickly. I’m thinking about “The Road Ahead” by Bill Gates that depicted lives where doors opened on command and home thermostats were based on your current whereabouts, etc. Housework would become a thing of the past, fully automated. Self ordering refrigerators to keep milk and eggs stocked. Those technologies were very real and possible as he was writing those words but are still out of reach for 99.999% of people (meaning everyone but Bill Gates).
I also remember a “Nature of Cities” class in college where my professor proudly touted a picture of him next to a flying car after he’d taken it for a spin. He was certain in the next 15 to 20 years we’d all have flying cars. It’s been at least 10 years and I don’t think we’re anywhere close.
What you’re describing is awesome and I have no doubt it will come together but I’d bet it will be varied and in a different and slower way than we might expect.
Just like Rhoomba’s and the affordability of air travel. And so many other shopping tools we’re already seeing like AuctionSniper, rapidly improving search, Groupon, etc. etc.
Anyway, great post. Thanks for bringing it all together (and nice reference to Mercata.com!)
Nice to see a meaty topic and hopefully, we’ll get a nice discussion going. I guess my main question is what problem is this intelligent agent supposed to solve that’s not getting solved today?
If it’s about product discovery, I suppose it could be interesting but there are a bunch of sites like Kaboodle, ThisNext, Boutiques, long tail of afficianado sites, etc. that cater to that need. I also see a few flaws with the idea that you could triangulate disparate pieces of data in order to “learn” what you like: 1) mathematically, it’s an almost impossible problem to solve because the signal to noise ratio is simply too low. How does the system know what you’re interested in vs. what you’re buying for your niece vs. what’s on sale vs. what your friends recommended vs. what you need because it broke, etc. etc? 2) Accordingly, the results will either be so obvious like the retargeting guys (interested in a blender, we’ll show you blender ads) or so non-relevant that you’ll turn off the experiment. 3) even if it worked, discovery tools are so far up the purchasing funnel that it’s hard to monetize. Suggestions are great but they don’t convert as well as a product one click away from the checkout.
If it’s about a personal shopper type tool, there are already services like Shopittome where you go through a wizard telling what you want and it does the shopping for deals for you. Works well enough. Not sure it’s a huge idea.
If it’s about saving money, not sure how well this tool will be able to generate savings or coupons or offers better than what’s already out there. You can only generate discounts if you drive volume. This tool, by definition, is a narrow-cast agent that drives a volume of one.
I do think there’s some merit to the location-based value proposition. For too long, online shopping has been relegated to online stores and free shipping, etc. It’s time to connect the best of online shopping with real world stores. Proving conversion has always been the problem here, though.
In summary, I ask the core question about what problem it solves because I see time and time again people using Google to shop for things. When I tell them there are so many better ways to shop (CSEs, comparison shopping apps, coupon sites, etc.), they nod politely and continue using Google. In other words, to get any traction, you need a tool that’s orders of magnitude better than what’s available today. Sites like Groupon seem to work because they’re offering deals you can’t get or find anywhere else. That’s a defensible USP. Not convinced yet that a shopping agent is that killer app.
Hi Brian,
Any thoughts on what shopping engine would win out, out of all the major engines, such as shopping.com, shopzilla.com, nextag.com, etc?
Also, doesn’t look like you are updating with new posts these days, hope to read more… Thanks,