Recommended Hubs - How they are sorted

[Georgei]

I was looking at the hubs that 3DHUBS recommended in London and I was wondering how they are sorted out? The first recommended HUB is not the fastest, Is not the closest, it hasn’t got the best reviews but it is the more expensive (attached image). Can 3DHUBS give us some information about how it works because it looks like that there is push to promote the most expensive HUBS. If that is the case then I might have to increase my prices to promote my HUB.

[3DFlying]

3D Hubs will not reveal exactly how, they keep it a secret so people will not game the system. Price though is not a factor, there is a hub near me that charges $100,000+ for prints and they are usually 2nd or 3rd on the list.

Here is what we do know affects rank:

-Reviews and overall score

-Shipping & completion time - shorter is better

-Declined orders

-Your past history of completing jobs on time

-Response time to customer inquires

Things I suspect might affect it:

-Being online

-Number of materials available

-Forum participation (whenever I post like 5x msgs I usually get an order)

They have said in the past that the number of printers you have makes no bearing on ranking. Though looking at the number of times my hub turns up in daily searches, having 2 printers seems to make a big differences. Just an offhand observation though and might be coincidental.

[cobnut]

Indeed, it’s common to keep such algorithms secret to avoid gaming, so don’t expect a detailed answer any time soon. However, as @3DFlying said, it’s certainly not sorted only by most expensive and I’m sure you could do any number of other searches for other locations to quickly prove this yourself. It really wouldn’t make sense to promote the most expensive Hubs since that might deter potential customers; if I was designing such an algorithm I’d try to promote the Hubs that were most likely to result in a sale - any sale - and most likely to satisfy the customer so they return and order again and/or tell their friends about it.

[Georgei]

I am sure that there is an algorithm and I am not asking for that. On the image I have posted though it is clear that the first recommended HUB:

It is not the fastest

It hasn’t got the best reviews

It hasn’t got the cheapest shipping

It hasn’t got the best price

It is not the nearest.

It hasn’t got the largest number of reviews and probably the largest number of completed orders

It hasn’t got the higher number of printers

So there must be a secret factor that is very much weighted to overweight all these factors.

And that is my question.

Just to make it clear I have nothing against the specific HUB or any HUB. It is just a random screenshot from the search results.

[cobnut]

@Georgei it may not be one single thing, it could be a combination of factors and I’d even suggest it may be a random element - when I do the same search, I don’t get the same order as in your screen shot so I’m pretty sure there’s a random selection or a rotation of some kind going on. I’ve done this in the past when coding “featured” elements of a website.

It’s also, I think, not a good test if you’ve not uploaded a model and chosen a material; 3DHubs are going to try and match your order to the most appropriate Hubs, if you’ve not even chosen what material to print in, and they have no idea how big your model is, the list suggested is obviously not going to be as closely tailored to your order.

[Perry_1]

Whenever ranking comes up, I always interject something. I love 3dhubs, think they do a fine job, and think they deserve a ton of credit. HOWEVER, this has been the one thing that kills me. The ranking is not logical. I have been watching it for years.

I have continuously watched ranking in my area. I have seen hubs on their 3rd print rank higher than hubs on their 100th print.

I have seen lower reviews, less material, less completed hubs, lower delivery times, etc. rank above solid good hubs.

3dhubs support CANNOT help with this. They do not know why the hubs rank the way they do.

It is not that it is secret, it is that they DO NOT KNOW how it works.

In this thread Talk Manufacturing | Hubs

it is revealed that:

"our Hub scoring algorithm is quite complicated. Reviews and number of orders are not the only metrics we look at. We are currently building tools that will help us quickly see why some Hubs rank higher etc. We will keep you updated. Adam "

which means 3dhubs employees can’t even tell why. And this:

“Currently it is very hard to tell what exactly is the differentiation between two Hubs. It requires one of our engineers to look into the data. We are building tools for customer support to have all of this on one click. adamhayek”

which means that its probably not even a real, solid algorithm, if it cant be explained to their own employees in plain english, and requires an engineer to look at.

I’m not convinced it is a secret. I think the secret here is that it does not work correctly.
I actually do this for my real job- applications-- and I can tell you, there are really not that many metrics involved here that the ranking can be based on, and the ones they say rank you, do not work as they should, period.

But be very sure that if you do the best prints, offer fair pricing, have the most materials, get the best reviews, answer customers quickly and completely, ship quickly, and customers love you, you may still never be number one in your own town, even if you obviously should be.

That being said, I come off as critical of 3dhubs with this as my primary issue, and the recent tolerances fiasco, but they have a hell of a system for putting customers and hubs together. I do love it. But it definitely is NOT working exactly right, and their own employees couldn’t tell by looking why that is the case.

[cobnut]

@Perry_1 the problem with basing ranking upon absolute metrics is that it only works for one Hub in any one area. New Hubs have to have the opportunity to grow; if ranking is based purely on absolutes (number of materials, number of prints, user rating, etc.) then a new Hub will never be #1 because they’ll never be able to overtake older Hubs that are still working, still increasing their number of prints and their number of reviews, etc. There must be some randomising element to allow Hubs of comparable performance to spend some time at #1. Note that I stress comparable, I’m not suggesting a Hub with one print and a customer review of 2 stars should appear above a Hub with 100 reviews at 5.0, but if there are 10 Hubs all with more than 20 prints, all with 4.8 or higher rating (these are just random example numbers, no thought applied) then I believe they should “rotate” or be randomised to some degree. A clever algorithm could give the very best Hub more #1 appearances than the “worst” of that 10, but it should still give a chance to newer Hubs to compete, otherwise we’ll end up with a dozen or so Hubs in every country getting 90% of the work.

[Perry_1]

All things being equal within a hub ranking, number or materials, etc., I agree with you completely.

Even giving a new hub a little push when that is not the case, for awhile, seems like a good idea.

This is not the case.

Not being argumentative, but your post proceeds from a false premise based on my research.

OF COURSE, a hub that has done over 100 prints, has a 5.0 rating rave reviews, offers the most materials, and has comparable pricing should at some point get to number one in a search, in a general market area. This is not the case. I have watched it happen in several key markets in the US. Some hubs way outperform other hubs, and NEVER get to #1. It just does not happen.

If I could ask for just one clue on ranking, it would be on pricing. How does that affect ranking?

But the fact is, if you read my first post, 3dhubs support can’t really even look at the results of any rankings and say “oh, thats why…” which leads me to believe it doesn’t really work at all.

Number of prints is a bad factor for ranking, if the spread is not big enough, because it becomes a self fullfilling prophecy. Pricing is a bad factor for ranking, because it probably breaks US law. So what metrics are left.

Weighted location, reviews, etc.

IN fact, number of cancellations is a bad ranking item. I go out and attract extra attention to 3dhubs. I advertise my hub in google. I put posters up for the discounts in 2 universities. I get more unprintables as a result. I am doing a good job for 3dhubs, thing many other hubs will not do. This attracts a higher number of prints that wont work out, people who ask for bigger discounts, etc. I also have a good rep. So I get more complicated prints. I push ALL of my business through 3dhubs, its on my business cards, etc. I communicate strongly with customers about what to expect from their prints, what will look bad, etc. Because of all of this, of course I get more cancellations than the average hub. This should not hurt my ranking.

[Perry_1]

I wish they would just admit they don’t have a handle on ranking, and it does not work as well as they would like. Saying its a trade secret is a dodge. Saying they wont say how it works is fair. But listing things like print times, response times, number of materials, etc, and leading a quality hub to believe they can get to number one by killing themselves, when it might never happen comes off as very misleading.

For example, they recently said that they do location ranking by straight lines, instead of using the google mapping API, this was a smart move. They explained the resources were just not there, and it takes too much time. Traveling Salesman Problem, very computing intensive. This was a good, honest answer, and I felt good about them for saying it. They obviously cant come straight out and say their ranking does not work, because it is core to their marketing. But it does not work well, and they know it.

Adding again, I think 3dhubs does a heck of a good job. I have sent several support requests to them, they answer immediately. They do a pretty good job of handling a double market business. (2 sets of customer bases). That’s a hard job, and they do a good job at it. They are a pretty amazing company, and their software over all is pretty impressive. I am critical of the ranking, that is my only complaint. So I always jump into the ranking threads with a fury, especially when their answer is always the same- “trade secret”. I come off as more critical of 3dhubs than I really am, because these threads pop up about once a week.

[Georgei]

What is even more interesting is the following:

You can see that all these hubs they offer postage.

If you click apart from the recommended, also the shipping button then the list changes dramatically. And no the first one hasn’t got the cheapest shipping either.

Regarding the comments that 3DHUBS they keep secret the “algorithm” about how the recommended hubs are sorted because hubs can manipulate that I have to say: no you cannot manipulate it. You cannot manipulate how many orders have you done, you cannot manipulate your feedback, you cannot manipulate if you have done them in time, you cannot manipulate how fast you reply to the customers. The only things that you can change are your prices and your materials. So I do not understand the secrecy.

[adamhayek]

Hey guys,

we certainly keep the details of our ranking system secret to avoid gaming (yes, there are metrics that could be intentionally inflated), but I can give you some more insights here.

We do not promote more expensive Hubs and we do not look at your activity on Talk. What is true is that we give a minor boost to new Hubs to give them a chance to get their first orders.

There is a ranking algorithm behind the scenes which takes in many inputs. Although we know exactly how it works, it is hard to tell you a reason why you are not ranked at the top. For that we would need to see all the metrics for you and your competition and their recent movement. We don’t have these tools yet, but we are currently working on that.

Hopefully I will have an exciting update related to this in the upcoming weeks. Building tools for you to be able to see in which areas you can improve to rank higher is now our top priority.

Adam
3D Hubs

[FabD]

Hello,

in the last 3 months my hub ranking fall down in the recommended hubs list from the first or second to the nineteenth rank. I dont know why, but that`s frustrating. Maybe you can help me?

Best

Fabian

[Zaka]

Just add a lowest price section , I mean, did you really think a new customer will loose time to give a chance to a farest, newest, with less material and less reviews(quality) Hub ?? I think customer will just have his object fast and at a good price with the material he wants. If I try to find my Hub , even if I type my own country I’m only on the third page !!

Your “let a hub a chance” is just stoping Hubs that just begin having orders because “recommended” is the default section, the customer thinks it is better for him to choose recomended Hubs, but it isn’t , it is only good for the Hub .

Some Hubs (not me ^^ ) try to get their profile as atractive as possible , spending time in photos and explenation, all this work for nothing, cause every day a new Hub opens, and I will never hit the 1.

You should change the name from “recomended” to “new hub” cause that’s the truth.

Bye.