Hacker News Reader

Comment by pg | Hacker News Reader

The general implications of this are quite interesting. It means you can compete with Google at anything where really good customer service is an advantage, because the concept of customer service (except at such low granularity as generating good, fast search results) is so alien to their culture.

View on HN

That's how we kicked their (and Yahoo's, and MSN's) ass in Latin America for search advertising. We spoke the languages, answered the phone, answered emails, dealt with icky non-US forms of payment, walked people through their first ad campaigns, etc.

Who is 'we'?

Terespondo.com. Comically tiny team compared to Overture. Yahoo/Overture bought it in 2005. At our high point we handled search advertising for 7 out of the top 10 sites in Latin America: UOL, Buscape, etc. Even MSN became a customer.

Even the name implies good customer service: "I respond to you".

Yeah. :) That was by accident though. Tere started out as a metasearch engine and answers service. (I was an early contractor. I left for a few years then came back after they switched to ads and landed the MSN contract.)

The technology was very lean and cheap -- Lackner, Leo, Drago, Paterlini, and Freitas are real hackers. But it wasn't sophisticated compared to Google then or Panama now. Our advantage was that when the phone rang a real human said "Bom Dia".

Amy's problem is not rare. It's one of Google's weak points.

I think you can generalize this to say "you can compete with companies that focus on scaling to large markets of X by providing services for X which don't scale".

Maybe that's the reason why they're not doing it. Cost of satisfied customer vs. cost to make him happy. Don't be evil left google and is now at Friendfeed anyways :)

Even, for a small company, something like adwords. The problem I have with AW is that if I change the wording of an ad (say, to add the price, or change the price or other "trivial" changes), I have to wait DAYS for the ads to run. I know others who say their changes take effect within a few minutes. I get no direct answer other than "we try to treat all of our customers equally."

I think I draw a flag because I sell educational software, and they want to make sure that I'm not selling homework answers or term papers (or whatever they are worried about). I would benefit from some sort of flag that says they've checked me out N times, I'm probably good for the N+1st time (with maybe random checks afterwords). Or let the ads run and check concurrently.

For example, I shop sometimes in a supermarket that lets me scan the items with a handheld scanner and pay at a self-checkout lane. When I first started doing it, I was randomly (?--I assume) audited fairly frequently. But after very many (over 8 months maybe?) honest check-outs, I haven't been audited in at least two months. It seems I've earned some trust, and I find myself shopping there more because of the convenience.

Putting some sort of trust-factor into Adwords would be useful for me. Alternatively, somebody that could get me the returns that Adwords can (big hurdle) but puts in a trust factor would get my attention.

There is some trade-off between efficiency and friendliness, as long as friendliness requires some real human services (which is likely the case for the foreseeable future). Of course, there is definitely room to provide more of both, but trade-off exists nonetheless.

When a typical merchant signs up for a service, she'd look first for the expenses involved. If a non-free service springs up even with a great customer service, there is little chance they will get enough scale to compete successfully in the long run. As long as most merchants don't have big problems with mostly automated (and less friendly) services like this, Web-based payment services will stick to this free model.

That's an issue with this 'free' economy. Nothing is really free. We've collectively given up human customer service to get these 'free' services.

I think it might be useful to generalize that even more. Google are good with aggregate. When problems are granular they are not.

I don't get it -

aggregate (n): a sum total of many heterogenous things taken together

granular (adj): composed of or covered with relatively large particles

You mean that Google provides services that scale - like a fully automated checkout system, and doesn't provide services that don't scale - like personal support. I don't know if it'll end up being a wise one, but it's certainly an interesting business decision they've made...

there are already lots of credit card processors that compete on service. they cost more to use (both in fees and your labor to setup). people buy on cost then wonder why their cheaper solution isn't as good as the pricier one. like people who buy the cheapest airline ticket then gripe that the food wasn't good. you get what you pay for.

IMO, this is a problem with many things - it's very easy to focus on that which is quantifiable, and ignore the fuzzier stuff. For instance, I did a comparison of Linode and Slicehost which has proved quite popular, but one of the things I stated up front was that while it's possible to measure costs and things like memory and disk, it's not really possible to easily compare service, especially as for the most part, hosting is invisible, and service only becomes critical in an emergency.

Economics in general suffers from this problem too. It's easy to quantify things related to money, but much more difficult to deal with the benefits from more abstract things.

Service is a feature that's much harder to quantify, but there _are_ ways of demonstrating it. One of the best recent examples I've seen is http://www.engineyard.com/, which lists its sales & support phone numbers on the header of each page. In this day and age, where phone numbers are being increasingly hidden, a company that stands out like that really makes a good impact.

That's right in general, but a payment processor cutting off the service without notice or explanation isn't comparable to bad food on a cheapo flight. It's more like offloading passengers during a stopover in Mogadishu.

It sounds like they make decisions when confronted with heaps of data. A billion searches? Sure, we can figure out what worked & what didn't. Versus a few dissatisfied customers, orders of magnitude too small to train models on.

Also, given that engineers have to design their services to scale well, perhaps the answer is that we just don't know how to build cheap customer service yet? Perhaps they need a Mechanical Turk like interface for CSRs.

It's not just "dissatisfied customers." It's any customers at all.

I dare you to try to get an answer from Google about how your site is indexed, or about how to accomplish something in analytics. There are several Google products where the only support option available is to post in a Google Support Group and pray that an engineer responds.

It's pretty frustrating.

Replace Google with most large Co's out there and the argument is still valid.

Name me a traditional payment processor (e.g. a bank) that can just walk away with your money and you have no means of recourse.

WTF? I say "Not only Google, but pretty much any other large co has crappy customer service. Let's compete on good customer service!".

Was your reply supposed to be an objection of some sort that I missed? What are you disagreeing with? Yeah, and what's up with 17 people who upmodded it?!

Reply on news.ycombinator.com