Hello Again, 2006: The Economics of Reference Management Software

The tiny and insular world where academia, technology, and business converge buzzed all day yesterday (and continues to do so today) about publishing giant Elsevier’s rumored bid to purchase Mendeley for $100M. TechCrunch’s dependably credulous reporter duly transcribed the leaker’s claims that publishing is “the world that Mendeley is disrupting.” But this story has really nothing to do with a “disruption” in academic publishing, and if anything what we’re seeing is a reversion to 2006 or so. When Zotero launched then, the major players were Endnote (Thomson Reuters) and RefWorks (ProQuest), each owned by a major content provider. And now in 2013 we can add to that stable of publisher-owned reference managers Papers (Springer) and, apparently, Mendeley (Elsevier).

As someone who has led a successful and sustainable project in this space for over six years, I’d like to put this rumor in perspective, because it speaks volumes about the space of academic research software, even if the Elsevier purchase never materializes.

1. Research software doesn’t make (enough) money
There’s just not a lot of money to be had in this space. Thanks to certain events, I have an exceptionally clear idea of EndNote’s revenue and overhead. Trust me, it’s not the thing of VC dreams. I also know exactly what good developers and sound infrastructure cost. And as for Mendeley’s “business model,” well, that’s what Zotero has actually been doing since 2009. With that in mind, I can assure you that there is simply no way to pay 3 FTEs, let alone Mendeley’s claimed 45, on “tens of thousands of dollars” per month. And cover the infrastructure costs. And the rent. We’re talking about an annual burn rate that’s solidly in the millions, no matter how you slice it. And now to turn some massive, VC-satisfying, profit? The “business model” could never continue on its own, let alone withstand the scrutiny of a public offering. Acquisition was always the exit strategy. Now there’s nothing wrong with this strategy, or even with the need to be coy about it. But based on the market it was the only remunerative outcome. This gloomy outlook doesn’t mean that independent projects can’t thrive. It just means that they won’t be gold strikes on the merits.

2. Users don’t pay for research software, just like they don’t pay for content
It’s misleading to blame the need to be acquired by a deep-pocketed publisher on the Web 2.0 generation’s purported aversion to paying for software. Individual researchers, in the aggregate, never paid for the software. EndNote’s bread and butter isn’t the poor sap who shells out $300 for a boxed copy. It’s the research institution (e.g. university library) that signs multi-year site licenses for tens of thousands of dollars each. RefWorks relies — though it feels odd to continue to use the present tense — nearly exclusively on these deals. So if Elsevier does acquire Mendeley, researchers won’t be getting it for “free” any more than university students get RefWorks or EndNote “free” at subscribing institutions. The software will be bought effectively via library subscriptions to Elsevier content. It’s just more wealth transfer from research institutions to for-profit publishers.

3. Publishers think they need to own a reference manager
All available evidence shows that it’s certainly not about money directly generated by reference managers, and yet publishers still want them. Elsevier’s ill-fated 2collab didn’t teach them that they should own a reference manager. It just taught them that they needed to own a different one. Springer’s purchase of Papers likewise makes sense only from the perspective of this mentality. Then why do it? From a sales and marketing perspective, there’s a perceivable benefit to being able to bundle content with research software. Not only does it help to establish a loyal community of users at an institution, it’s simultaneously inevitably a competitive play against another publisher, which has its own competing software to offer the purchasing organization. If Mendeley is acquired by Elsevier, it marks something closer to a reversion to the status quo from a business perspective, not a sign of a dramatic shift in the market. Indeed, for Mendeley to be purchased it needs not to be disruptive.

4. The data is worthless
Had Mendeley been able to monetize the data piecemeal, they would have. They’ve admirably devoted a ton of effort on this front, and still have little to show for it. This is not at all to say that Mendeley users should not be concerned about privacy or data-mining, but rather that there was no palatable way to sell the aggregated data.

In sum, if Mendeley does get acquired by Elsevier, its hyperbolic stated mission to disrupt academic publishing will of course finally be laid to rest. But this is not to say that there’s no synergy, to use some TechCrunch-speak, between research software and publishing. Hold on to your hats, but did you know Zotero’s parent company also runs some journals?

  • Harald

    Very nice and convincing analysis. Thanks!

  • And now in 2013 we can add to that sta­ble of publisher-owned ref­er­ence man­agers Papers (Springer) and, appar­ently, Mende­ley (Elsevier).” — add Readcube (Macmillan)


  • allison stelling

    this is why i use LaTeX when i write my manuscripts (i am currently doing research in the intra-operative diagnosis of brain tumors); it has a much easier to use referencing and citation system than EndNote. (i usually cite 30 to 40 papers per manuscript, and EndNote just does not cut it.) also JabRef (my citation manager) is free, if you want a GUI.

  • adam.smith

    Allison — one problem for bibtex is the frontend — gathering resources is a drag (even with jabref) compared to Zotero. That’s why many people use Zotero or Mendeley as BibTeX frontends.
    (The other problem with bibtex is the lack of standards — which @types exist and what labels they should have is all over the place, which is bit of a disaster. BibLaTeX is an interesting attempt to deal with that, but it’s far for universally accepted).

  • allison stelling

    i don’t know about the lack of standards— it’s really just copy and paste for a given journal’s template, so it’s pretty easy to re-work a draft into whatever format you like. journals these days all have different templates and styles for how to format the reference section, so it’s nice for me at least to be able to switch between different styles with just the style command.

    my only main issue right now is that i can’t search my own database of citations (can you do that with Mende­ley, or this Zotero one? right now i have to skim through about 80-odd citations just to see the one i want to cite for the particular manuscript!). i’m going to switch over to Askemos (http://www.askemos.org/index.html?_v=about&_id=1854&_o=2) so i can keep a searchable, personalized “digital notes” record for my entire online bibliography. (also i know the guy that made it, he’s a comp sci guy who’s pretty keen on keeping the virtual library from burning, if you see what i mean. some info should be private, but science paid for by taxes should be public and thoroughly backed up.)

  • Pingback: How to export, delete and move your Mendeley account and library « O'Really?()

  • adam.smith

    of course Zotero and Mendeley allow you to search and organize citations in various ways, including full texts of PDFs that you attach etc. They allow you to annotate citations, tag them etc. That’s why you should use a bib manager — even if you’re not using one of those, at least use something like jabref, just having your bibtex in one file is a big waste of time. They also, rather importantly, prevent you from ever having to enter citations manually (Zotero more reliably than Mendeley).

    As for standards: BibTeX doesn’t have data entry standardized. How do you input a webpage in bibtex? some flavors use @webpage some @electronic, some @online, standard bibtex uses @misc… the access date is sometimes saved as urldate, sometimes as lastchecked, sometimes even as accessed-date. People disagree whether bibtex entries should have keywords, and if they do there is no convention on whether they’re delimited by space, comma, or semicolon — and so on. That’s a huge problem if you’re publishing beyond a narrow set of journals.

  • Pingback: Elsevier buying Mendeley? › Hybrid Publishing Lab Notepad()

  • Pingback: Unruhe auf dem Literaturverwaltungsmarkt – Connotea vor dem Aus « Literaturverwaltung & Bibliotheken()

  • Nathan

    Allison and Adam: I just made a comment on the older post ‘Zotero Versus’ about how I have found ways to fit Zotero very nicely into a TeX–based workflow. (I call this perspective ‘Zotero et’ rather than ‘Zotero versus’.) I agree with Adam that the BibTeX entry types and fields are excessively flexible. But I disagree with Adam that this is a ‘huge problem’, because any researcher who is using TeX is probably smart enough to work with that flexibility: for example, since BibTeX files are plain text it is easy to make batch changes using grep and any of various scripting languages. More challenging to me is the emergence of unique identifiers for scholarly authors such as ORCID, ResearcherID, ISNI, etc. I don’t currently have any way to integrate those into my workflow, but I also don’t have any need for them yet.

  • Pingback: Mendeley, Connotea, and the Perils of Free Services | The Scholarly Kitchen()

  • Pingback: “Mendelsevier” – “Mendelete” …? Erklärungen, Reaktionen, Konsequenzen in Sachen Mendeley und Elsevier | Literaturverwaltung & Bibliotheken()

  • Pingback: Special Interest Group Science 2.0()