Wednesday, June 14, 2017

Family Tree DNA: My Review

Over the past couple of years, you've gotten ample insight from yours truly on AncestryDNA and GEDmatch. Clearly, this conversation is incomplete without giving Family Tree DNA some long-overdue attention. So today, I'm going to be sharing my thoughts on their Family Finder/autosomal DNA test. Specifically, I'll be sharing my experience with their autosomal transfer service.

As I did in my AncestryDNA review and the follow-up, I'll be breaking down this review into a couple of different categories. I'll grade them individually, and average them together for a final score. Here are the categories:
  • Cousin Matching
  • Admixture
  • Projects
I'm also giving them extra credit right out of the gate for offering the autosomal transfer at all. Their biggest weakness, at this stage of the game, is the size of their database of matches. It's difficult to compete with AncestryDNA's 4,000,000+ tests analyzed, especially with their last million being added between January and April of this year. Having such a common sense incentive for the customer is one of the industry's best bargains. If 23&Me would follow suit, I'd have a full house.

Cousin Matching: C+

There are so many different fronts to approach this from, so let's dive in! Family Tree DNA has an overall great interface. In my opinion, their interface has always been better than AncestryDNA. While the capability of Family Tree DNA's interface isn't as robust as GEDmatch, they have many more analysis tools to offer someone out of the gate than AncestryDNA does. And by using all of the features available on Family Tree DNA, I can see how it would help someone transition much more smoothly over to GEDmatch.

Family Tree DNA's match list is like GEDmatch's One-to-Many match list, on training wheels. It's a sortable table that allows you isolate matches in a variety of ways. You can click the column headers, which allow you to sort X-DNA matches to the top of the list. You can also use the "In Common With" button, to see which matches you share with a given relative. What AncestryDNA customers usually have to find on two different websites, Family Tree DNA has consolidated into one.

Their match list also has an additional field for surnames, which is unique to this site. Those surnames are searchable, as are your cousin matches. This feature is only available with users who fill out their profile, but many of the most active users fit that description. While I do like being able to sort by location on AncestryDNA and feel it is missing here, Family Tree DNA still has a running head start. They could easily add that functionality without a significant disruption to their interface. Which is still more than AncestryDNA can say when it comes to revisiting a cousin match. (Does anyone else realize you can use the Membership Directory to do this? Yeah, no one else does either.)

Family Tree DNA has a chromosome browser, which is also more than AncestryDNA can say. As nice as that is to have, it's actually my least favorite part of working with Family Tree DNA. However, the chromosome browser isn't the problem. It's how inefficiently it operates to do a very simple job.

I greatly dislike that this is the only way to access the full segment data for each match. To do this, I have to systematically input all 2700+ matches into the chromosome browser in batches of 5. And of course, there's literally no way to pull up 5 matches on the same chromosome in the chromosome browser. So they're just 5 random matches from an alphabetical list.

I click on each segment that appears, and copy the data from a pop-up box that has a nasty habit of disappearing if I click anywhere else. This is particularly annoying to do when the segments are small, and there are multiple segments to analyze for the same person. Since I also use Microsoft Access for my data entry, this means I have to move my Access window around my screen based on where the box appears. Many times, I lose the box while trying to situate my windows, and have to start all over again.

First World Rage!
This interface has so much potential, and has so many great features on it. I don't understand why they have it programmed to function only one way, which is the least efficient way possible. They have TWO other options through which they could present the segment data, and THIS is the one they chose. They have the match list and the Excel/CSV download feature they could be using instead.

As an Access user, I can't tell you how valuable it would be to me to export segment data on an Excel spreadsheet. I would be able to do a bulk import of segment data into Access. This would save me hundreds of hours of work, and allow me to spend more time on analysis instead of data entry. And really, what exactly is the purpose of exporting matches into an Excel spreadsheet without the segment data? Like so many things going on with Family Tree DNA, it's a fantastic idea that falls short in the execution.

[NOTE: When I used the Excel/CSV feature in preparation for this post, I used the link available from the Match List page. The matches exported from that page are as I described them: just about useless in every way. It provides the user's name, contact information, relationship estimates, and no full segment data. I didn't know that using the Excel/CSV link from the chromosome browser page would produce a different match list, which DOES include exact segment lengths and locations. The mistake here is mine, and I apologize for the confusion.

However, upon receiving this second match list, I find it equally dissatisfying for the following reasons: it does not include any of the pertinent identifying information from the previous list. It has an entire column dedicated solely to my name, and no contact information or relationship estimates. Having the relevant information I need broken up across two Excel workbooks does not, in fact, allow me to import that information into Access. And given that the segment data match list has matches smaller than 6 cM, I would have to manually remove each of them myself. According to my personal user needs, I'd like to see these two lists combined. And with the knowledge I have of DNA comparison, segments smaller than 5 cM are best removed from that match list.]

I'm also not a fan of Family Tree DNA's matching criteria. They give too much emphasis to smaller segments, with no way to set your own match criteria. With 2700+ matches to process on their website alone, I don't have time to waste with 5 and 6 cM matches. And yes, considering some of my research challenges, those matches are a waste of my time. The closest thing that Family Tree DNA has to custom match criteria is a drop down list of options on the chromosome browser. And since I haven't had anyone else closely related to me tested, most of the options are useless to me.

Family Tree DNA has room for improvement when it comes to cousin matching. But they're in a much better position to address those issues than AncestryDNA is, and I hope they recognize it. They have the potential to be a much more comprehensive service than AncestryDNA probably ever will be. What AncestryDNA has done by being the lowest of all possible fruit, Family Tree DNA can accomplish by superior performance.

Admixture: B-

It's no secret that I'm not the biggest fangirl of admixtures. We've all heard the wacky stories about how ethnicity estimates revealed that babies born a hundred years ago were switched at birth. But really, for most of us, ethnicity estimates don't have that kind of potential unless you're biologically diverse. Nevertheless, I was still curious to get a second opinion on my AncestryDNA ethnicity estimates.

European Ethnicity Estimates from AncestryDNA

myOrigins from Family Tree DNA

You'll note that there are some pretty large discrepancies here, including between ethnic groups that both companies test for. The largest disparity is the 26% Scandinavian from AncestryDNA, versus 0% from Family Tree DNA. With a difference that large, someone here is wrong. I don't have much of sense of who it could be. The more detailed view of the AncestryDNA test seems consistent with what I suspect about the unknown branches of my family. But without more traditional research and confirmation of details to back it up, it's not a question I can resolve myself right now.

I will say that unlike AncestryDNA, Family Tree DNA has a separate, distinct testing region for South America. This was news to me, and has been added since I originally did the autosomal transfer. AncestryDNA treats the Caribbean, North, Central, and South America as one homogeneous region. As someone with Caribbean heritage, with rumors of Native American ancestry I was trying to squash, being able to distinguish the two is useful. Anyone with interests in these regions will receive more detailed results from Family Tree DNA than from AncestryDNA.

Projects: C

AncestryDNA has 4,000,000+ samples in their database. There's no escaping that fact. If Family Tree DNA is going to compete with the statistical probability of making a match, they need to deliver superior service. They can accomplish this by updating features they already have, and maximizing on AncestryDNA's weaknesses.

You can't make money if you don't exist.
From what I can see, a significant portion of Family Tree DNA's attention is still focused on Y DNA and mitochondrial DNA testing. And before anyone challenges me about it, I recognize that these tests have their place. But those tests are niche at best, and stubbornly nostalgic at their worst. There is no sense in encouraging the average person, let alone a new customer, to spend hundreds of dollars on one line in their genome anymore. Commitments to niche tests take resources away from developing autosomal testing offerings. And anyone who isn't committing the lion share of their resources to autosomal testing is making a big mistake. They are going to seem crazy when they try to explain their reasoning in ten years, if they're still around to talk about it.

One example of where I see this happening is with their projects. They are most analogous feature they have to AncestryDNA's newest Genetic Communities. With Genetic Communities, AncestryDNA approximates the geographic regions your ancestors belonged to, based on DNA and family trees. Family Tree DNA has had this feature for years, with user-operated location, lineage, and surname projects. But it's still set up specifically from back in the day when Y and mT DNA tests were all the rage.

Why haven't these been upgraded to function with autosomal testing? They're literally sleeping on one of the biggest advantages they have over AncestryDNA. The might as well stand on a street corner and give away free money, for the amount of good it's doing them.

To those who have taken DNA tests, imagine being able to pull up a map to find collaborative discussions about the communities you care about most. Everyone there is specifically talking about DNA. As you scroll in deeper on the map, more specific projects appear, on state, county, or even the city level. Each one would have surname and lineage projects associated with it, with the ability to add more projects. As much as I enjoy Ancestry's Historic Insight tree hints, these don't create or engage people in conversation. Having projects based on historical events and common experiences would spark those conversations for DNA. And unlike Facebook, where a lot of these community-based conversations are happening right now, Family Tree DNA can integrate their features into these projects and conversations.

One of AncestryDNA's critical weaknesses is not fostering effective communication between it's users. They ignore this in favor of automating as much of their user's experience as possible. Family Tree DNA already has, as part of its offerings, the solution to this problem. But those features are only available for the most expensive, niche components of their offerings. (And that's if you don't want to go as far as I will, and say that Y and mT DNA are already dead.)

As a user experience and a business strategy, it simply makes zero sense. You can't be everything to everyone. It's the same trap that restaurants fall into when they have too many choices on their menu. Instead of trying to provide value through eleven different offerings and options, why not focus on the one thing that matters, and doing it the best way you know how?

Final Grade: B-

Out of the box, is Family Tree DNA a better service than AncestryDNA? In terms of the product they offer and the features that come with it, yes. But when you factor in the size the of their respective user bases and the company's ongoing leverage, the story changes. If you're going to compete with a much larger company, and a growing playing field with increasing amounts of competition, you have to focus on innovation.  You can't rely solely on carrying obsolete products your competitors don't have, and charging more for them. That is not a strategy for growth. That's a memoir for Radio Shack.

The Eighties called.
They want their DNA test back.

But from where I'm sitting, there's no reason for Family Tree DNA to be in the position they're in. They have everything they need to give AncestryDNA a serious run for their money. They just aren't using any of it to their advantage. And as a customer who is approaching the decision about purchasing secondary kits for family members, I'm finding it difficult to decide who deserves my money more. The company that doesn't listen, or the company that is moving precariously close to the wrong meaning of "throwback"?

Some might say that "only time will tell," but I disagree. It isn't time that will tell. It will be innovation and common sense that determine the future of Family Tree DNA and their Family Finder test. And the silver lining here is that innovation and common sense are 100% within their control.