1

How does the best match filter work on 3D Hubs? I would like to make my hub better for this!

Thanks!

8 Likes
2

Yes I would like to know this as well.

2 Likes
3

It would be interesting to hear what goes on behind the curtain…

1 Like
4

Fair question, as I can understand it feels a bit ‘vague’ right now. The purpose of the Best Match is to present the best Hubs to the customer based on their files uploaded, location and needs.

We do this by rewarding good performing Hubs on the platform, both experienced and new Hubs (based on their Marvin review). The following points increase your ranking on Best Match:

  • Good customer reviews
  • High ratio of orders received to succesfully completed
  • Quick customer communication
  • Returning customers
  • Complete Hub profile (e.g. pictures of prints, reviews, description)

Another variable that is used in Best Match is location. We increase the rank of Local Hubs (to a certain extend) than those that are far away from customers. The exact algorithm I cannot disclose, as that would open it up to manipulation. In that way it’s a bit like Google ranking algorithm.

I can give you a tip, so what you can do right away to improve your score is uploading a header pic and adding images to your profile, and remaining active on the platform.

Let me know if you have further questions or ideas to make it better.

Brian

ps. we’re doing a few more improvements in the coming 2 weeks that will improve the location factor so far away Hubs show up less often in local results

9 Likes
5

Two weeks to change the default setting from 250 miles to 5 miles? Might need to hire some new developers!

1 Like
6

When someone searches for Hubs that deliver only is the distance bias discounted? Stuff will get around the UK in the same time regardless of distance, so it’s nonsensical to put a bias on those local. It will also make prices more competitive, which is a good thing because the prices on this site are pretty ridiculous. Lower prices and more volume is certainly the way to go here.

There is already manipulation of distance you mentioned going on. Try searching for hubs in London. There are several at distance 0. I’m just as much in London and as accessible as others, but mine is a long way down because my distance is further from what you define as the centre of London.

Why are there no filters for price or time?

Do these factor into best match? If not, why not?

It also seems silly for a prominent feature to be to search by printer. These people want a 3D print and aren’t experts on models of printer.

7

Thanks for the feedback! Indeed when users search and indicate they want it delivered the distance factor is weighted much less, ideally it would search by country, something we’ll likely add.

We’ll start countering the manipulation of distance soon by verifying Hubs locations, it already improved a lot by showing just 1 listing rather than showing an entry for each printer listed by the Hub.

The reason that we’re not yet sorting by price is that each of the 25.000 Hubs have different pricing setup and sorting that requires quite a heavy calculation and lots of currency translations, we’ve got it planned on the roadmap and know that it is a valuable feature. We do offer a filter by time, using the ‘deadline’ selector we show Hubs that match the desired delivery time.

The search by printer filter is hidden in the advanced menu, do you find it still too prominent in there??

8

That wouldn’t be a solution as we have users all over the planet and they are often more than 5 miles away from the nearest printer. Showing no results is not a great experience.

9

I’ll admit, I live in a crowded city with a lot of hub options so I didnt think of that, sorry.

So how about this:

Set it so it shows 10 hubs minimum automatically starting at 5 miles and scaling out until it hits at least 10 hubs. That way Hubs in a more crowded market are better highlighted and customers in areas with fewer hubs still can see a range of best options.

10

Cool idea, Josh. Our current thinking is to apply the distance factor as a weight in this way the closest Hubs will be ranked higher, but still better Hubs rise to the top and inactive Hubs drop in the ranking.

As soon as we implement this the distance filter should not be needed as much anymore and it should show the best hubs nearby first both in crowded and spread out areas.

2 Likes
11

Still a little confused as to how you can rank hubs based on activity.

What do you count as activity? Sales, reviews, photos, messages? Because its hard to get any of those when your lower on the list. Seems like a downward spiral for anyone who isn’t already at the top.

For example if you type in my exact address I am 2nd on the list and my zip code I am 5th. I know the company above me for my address does more in sales than I do because I’ve met with them before. But they don’t post any new photos and I try to do one or two every week or two. They are also now an HD hub (which kind of annoys me but for other reasons) so I guess that gives them priority?

As for the zip code a few are listed as closer to the zip, not sure exactly how you can be closer to a range of area’s but ok. Most of the ones above me have more reviews and seem to have more in sales. Besides that not much else is different from my hub. Again though its a little hard to improve reviews and messages with people when you are getting none to start with. I already have just about the lowest pricing in the area and I cant really go any lower as its already almost not worth the time and effort for the money. Luckily I want to do this for more than just the money.

12

How soon will you verify location? You should also do the distance weighted on travel time. E.g. prefer a hub close to a major station that is the same distance as another that isn’t. You appear to be using google maps or something, so that shouldn’t be too difficult.

Currently your only non-advanced filters are location, material and deadline. Most people will open up the advanced settings to look for some of the things I’ve mentioned. Although I don’t think it’s a big issue, I certainly think it’s odd to have that and not a pricing feature. It’s worth noting that most people don’t know what material they want. Someone wanted to print a large bracket on my hub at 0.05mm layer height on Monday for example. The fact that you’re giving customers access to complex filtering that is likely to be arbitrary to them and better for the hub to decide instead of things they understand (like cash), is a little strange.

The price range is given when you search with a model, and otherwise should be shown with a standard (the keyring, perhaps), so I don’t see why you can’t implement this feature soon.

Look at search engine advanced features. Have a bunch of filters to narrow down the numbers and then have a bunch of different filters.

Just some thoughts.

13

Thanks for the feedback Daniel, much appreciated. The location verification is on the roadmap for second half of November, so it’s just a few weeks out. I agree that resolution and price filtering would be great additions and me and our UX’er @Gregor_Fifer are working to get these out. Using an example file to calculate the price is one of the options indeed.

14

Sounds like you’re doing everything right to be honest. You’re active on the website (that includes Talk) and you have your profile completely set up and have competitive pricing as well as some good reviews you should be all set for receiving more orders.

15

Whilst we appreciate this is how it should work, it doesn’t. That’s why the confusion arose in the first place and I decided to start this thread.

1 Like
16

I should be but I’ve only been getting less orders over the past few months. And I get ultimately it comes down to the customer and their is only so much you and I can do to get orders but somethings changed in the past few weeks and its killing orders for a lot of us. The worst part is you have made so many changes all at once its hard to pin point the exact problem.

Its like you had a working car but decided you wanted to upgrade the engine. So what did you do? Instead of upgrading one thing at a time and testing it, you upgraded every part of the engine in one go. Sure you got a faster, cooler car but now the car wont start. Good luck trouble shooting that one.

One thing that would be nice is this; You guys like to post a lot of statistics in the monthly trends. None of them really help Hubs try to improve their offerings and gain customers. You currently share things like how many printers are listed, and what has the most popular for owners. How about including the most popular printer for orders on 3DHubs? And what the average order cost for that printer is. Because sure the Makerbot is the most popular printer but anyone who knows 3D printing well knows it is by no means the best printer as far as quality and reliability. And the info of most popular orders will probably be slightly skewed by the most popular listed printer but maybe we will be able to see a trend in something like duel vs single extruder, build volume, possible materials, etc. Never know till we see it.

That way someone looking to expand their 3D printers can see what be the most cost effective printer might be to buy. Next is you mention the most used colors. That’s good, it helps me know what to purchase to have in stock. What about what material type is most commonly used? There is so many options out there now that I could not possibly have every material in every color. This interests me as a Zortrax owner because my material offerings are limited at the moment. I am looking to get another printer soon and I am trying to decide if i should get another Zortrax or get a more open source printer so I can offer more exotic materials.

There is so many more helpful statistics you could easily share with Hub owners that could help us improve our offerings and answer so many question about where this additional “20% increase” in sales is actually going. I also think more transparency on average user statistics. Things like how much business successful hubs actually bring in, how many people actually look at our pages. I know we have a listing rank or something but that’s not helpful when we don’t know how to improve it. Maybe show stats for hubs in our area. Would be nice to see if other hubs near by are having more success than us. Would allow us to see if maybe its our material offering, our build volume, our pricing, etc. or maybe that just there isn’t business in your area (which I know isn’t true because two of the people above me have over 50 reviews each). This would also allow some of us users who have been waving the pitchfork in talk lately to see if 3Dhubs is actually killing sales or if its just been a slow few months for everyone and get off your backs a little.

TL;DR: Slow down the updates. Give us statistics that actually mean something to improving our business.

1 Like
17

I really like this idea Josh, a kind of ‘Hub Insights’ to help you better run your Hub. Of course this data could be somewhat sensitive to competitors, but perhaps we can think of a way in which we share just the relevant info with the right people and not have it as public as the trend reports. In this way we could probably share more detailed information.

I’ll have a talk with @Paul_71 who organises our Trend reports and is currently working on the 3D Printer Buyers Guide.

2 Likes
18

“High ratio of orders received to succesfully completed”
Does this mean if someone starts an order and drops off the face of the earth it counts against me?

“Complete Hub profile (e.g. pictures of prints, reviews, description)”

If you guys think this is important than why did you remove the link for customers to see it?

19

Of course it can happen to any Hub that a customer loses interest, we don’t count that against you. Of course if you don’t respond or keep declining orders we will count it against you as it’s a poor customer experience, and the customer would be better off at another Hub.

Regarding the Hub profile all that information is used in the /3dprint checkout, and like we’ve mentioned already a few times, we’re doing research by running this A/B test in order to learn how we can improve the service more. It’s essential to test many different things quickly and learn which work and which don’t. This is how we got to where we are today, and how we can keep making the service better as we move forward.

1 Like
20

Ya, I get some of the info would be a bit sensitive. I think maybe offering a monthly private “Hub Insight” that each user could view that showed just local data would help with that. That way even if a competitor made an account to collect data they would only be able to see a little bit of it. Main thing I would love to know is, in order:

1. How many people actually go to my hub?

  1. What material types are most popular for orders?

  2. Why someone didn’t order from me?

4. What printers are the most ordered from?

5. How many orders happen around me?

First two I think would be the easiest and least sensitive, at least in a private “Insight” page.

And I think I’ve seen this mentioned elsewhere but I definitely think either a overhaul of the Trends report or a whole new section like this is needed. The current trends info, well they look nice and might be fun to look at for new users, are not at all interesting to experienced users. I remember looking at it a bunch when I first started and was all excited. After a while I realized it was all kind of fluff, for lack of a better term at the moment.

A quick breakdown of what I mean as an experienced, well versed user (To clarify, I get these trends are usefully new hubs/customers but I am going to ignore those points to show the other side of things):

1. World Breakdown, now apparently World HD Breakdown- So I just noticed this, didn’t it use to show every hub? Now it just shows HD? Really, that’s a lot upsetting but moving on.

2. Highest Rated 3D printers- Ratings are for the most part biased. There is no way the Form 1 can actually rate higher than the Form 1+. Its just a fact. Last time I filled out a rating I gave my Zortrax on average I think a 4.8. Was it biased? Of course, I love this machine. And these don’t help hubs expand business because like I said they are partially biased and just because the Form 1+ is a good printer doesn’t mean people are ordering from it. Sorry Formlabs for calling you out twice, nothing personal just the first example I saw, I love your printers. Also maybe just because I gave the Zortrax high marks doesn’t mean those same reasons as the operator are as highly valued to the customer. Does the customer care that their slicer software is so easy my 5 year old brother already can operate it all on his own? Nope. Does the customer care that the average print set up is about 5 minutes from downloading the print to pressing start on the machine? Nope.

3. Trending Printers- Ok, so more people bought them this month. Does that mean they are better printers for selling prints? Maybe they just were on sale. Maybe they are a new launch and had some good marketing. Again, not to call out Formlabs but you list the Form 1+ as jumping 30 spots, but we know why that was. They dropped in price. Has it helped boost any hubs sales that bought one though?

4. Top Cities- Well I would sell a kidney or two to move to Amsterdam, Paris, Milan etc. I wont be doing that anytime soon (Though I have been to all those cities plus a few more and Amsterdam is by far the best, I do plan on moving back soon, looking to hire? haha) so that wont help me boost sales. Plus I already live in #8 city, Boston. Also this is just based of printers listed. What about sales trends in that city? Sure one city might have the most printers but that also might mean since everyone owns one no one is ordering.

5. Printer Distribution- See points #2 and #3 as same issues apply.

6. Manufacturer Distribution- See point #5.

7. Manufacturer By Region- Huh, who would have thought American printers would sell best in America and European printers best in Europe? Again though, see point #6.

8. Printer Listing- Well I love to see that 3DHubs as a company is growing in listings, it doesn’t really show anything important. Maybe this is actually a sign that 3DHubs is failing. With too many printers listed there is now to much overhead with people hubs and photos and information. Too many people own printers and no one actually has to order prints since they already own the machine. How about how many sales you have total? Because just because people are listing things doesn’t mean people are buying, and lets be honest, that’s whats important.

9. Print Categories and average value. Kind of good to know I guess. But there is nothing we can do with this info. So scale models is the most expensive average order. What can I do with that info. Maybe post a few photos of small models to show off our printer? But its not like I can tailor my page to bring that in. Maybe I could get a Form 2 (again with Formlabs, almost like I got them on my mind) because they can generally do tiny models a little better than FDM. That would also account for why the average order is more expensive. But things that I wish I knew related to this are: Whats the most common printer chosen for “Scale Models” and how frequently are their “Scale Model” orders? Maybe its the most expensive because that is the fewest orders and the few that do happen are just expensive. If buying the Form to get those orders would help would it even be worth it if I only get one scale model print a month?

10. Popular Print Categories- Well I guess I should have really reviewed all the graphs before writing them as I partially answered point #9 with this graph. So as I brought up in point #9, Scale models is the 3rd least popular category meaning if I wanted to get into that for the expensive orders it doesn’t seem worth it since prototyping is over 4 times more popular and the average order is 17% less. But again I just had to kind of piece that together. I might be completely wrong. Maybe adding a Form 2 to my hub will bring in more customers, even if in the end they only actually needed an FDM print.

11. Color Distribution- The one useful thing as a Hub operator that is here. Because of this I didn’t have to go out and buy every color right away to do process of elimination to know what to stock. I already had white and black to start, I figured blue and red would be next, but I never would have though transparent would be next, or that grey would be so little. To add to this though, like I’ve mentioned, would be what type of material is popular? Because my machine currently cant print PLA or things like Ninjaflex and I feel that might be an issue but I really have no evidence to justify purchasing another machine or modifying my machine to do that.

12. 3D Printing World Map- Oh, wait, there it is. Well that makes me a little more upset. Why does all reliant data get grouped together but this? People already feel like you are pushing the small hub out of the way to make room for the big, bad HD hubs. Now you have literally pushed the entire world that made 3D hubs what it is to the very bottom of a very long page. Hmm, if I were to overthink this I would take it as some sort of subliminal message about 3Dhubs plan for us.

So take away: Almost half the trends are user biased. The remaining don’t promote any useful knowledge that anyone who is really into 3D printing didn’t already know. Some of them are fun to look at to start but once you really break it down there isn’t anything there but a good UX designer.

2 Likes