I Need Someone to Help! A Taxonomy of Helper-Finding Activities in the Enterprise

Svetlana Yarosh, AT&T Research Labs
Tara Matthews, IBM Research — Almaden
Michelle Zhou, IBM Research — Almaden
Kate Ehrlich, IBM Research — Cambridge

We conducted an empirical study of helper-finding needs in the workplace by asking 36 workers in a large company to keep two weeks of diaries of situations where they were looking for a colleague to help with a task. Unlike previous studies, we were able to examine and classify situations where the worker did not turn to any tool or system to find the helper. Based on our analysis of the diaries and the in-depth interviews with participants that followed the diary study, we developed a three-part taxonomy of helper-finding needs that highlights the ways that these needs are broader than current systems support and points to specific directions for future design and research in this space.

Systems are not always particularly good at finding somebody who can help in the workplace. (Comic by Svetlana Yarosh, in the style of xkcd)

We found that for the vast majority of diary entries (76%), participants asked colleagues to suggest an appropriate helper rather than turning to any sort of a system. Participants had a number of frustrations with the current approach of finding a person to help. To better understand the types of scenarios that are not currently well-supported, we analyzed the diary entries and created a three-part taxonomy of helper-finding needs. The first dimension, tasks, concerns the reason why a participant is seeking help, such as in order to learn a particular task. The second dimension, topics, concerns the subject of the helper-finding need, such as some bit of domain knowledge. The last dimension, helper selection criteria, concerns the requirements that target helper must meet in order to be relevant, such as being part of a certain group within the company. Each of these dimensions, their composing elements, and the observed frequencies of these elements are expanded in the handy cheat sheet image below.

The taxonomy helped us clarify the following frustrations with the current process:

  1. People frequently look for help not just to answer a quick question, but rather to have a longer interaction such as a collaboration or to learn a complex task. Systems designed for giving a quick answer did not meet these needs of forging a relationship with the potential helper.
  2. Most systems focus on providing answers to questions of domain expertise, but the majority of helper-finding needs were actually logistical in nature (experienced by 89% of the participants). This knowledge of how to get something done within the company was valuable but not well represented or searchable in any system.
  3. Participants frequently had specific criteria in mind for a helper, but specifics about potential helpers’ past experiences and roles within multiple groups were not well represented within current systems. This made it hard to understand whether somebody could be of help or not. Past experience was particularly important (to 94% of the participants) and yet hard to nail down at a useful granularity.
  4. Participants think about helper-finding in a way that is easy to express to a colleague but hard to specify to any current system in the company. They frequently looked for a person similar to somebody they already knew (X), such as somebody who performs the same function as X but in a different group, location, or in a different field of expertise.  It is easy to ask a colleague “Do you know who is X’s equivalent in the California group?” but hard to specify the same idea in a system-based search.

While our study highlights that helper-finding in the enterprise is a difficult task and current tools rarely help, our findings also point to specific directions for research and design that might be able to improve the effectiveness of future systems, but for these you will have to turn to the paper.

For more, see our full paper, “I Need Someone to Help! A Taxonomy of Helper-Finding Activities in the Enterprise.” 

(Click for larger image) The taxonomy of help-finding in the workplace includes the three dimension of help-finding and the observed frequencies of each element of each dimension. I’m sure that you’ll want to print this out and carry it in your wallet.

About the author

Lana Yarosh

Lana Yarosh is a researcher at AT&T Research Labs and a recent graduate of the Ph.D. program in Human-Centered Computing at the Georgia Institute of Technology. Her research interests fall primarily in the area of Human-Computer Interaction, particularly Ubiquitous and Social Computing. She has a passion for empirically investigating real-world needs that may be addressed through computing applications. After identifying these opportunities, she designs and develops technological interventions, evaluating them using a balance of qualitative and quantitative methods. Most recently, she has designed a media space system supporting synchronous remote communication between children and parents to help address the needs of divorced families.

View all posts


  • I find it fascinating that in Figure 1 (Tasks Needed reported by interviews and diaries) the ratio between how much people think they want to “give info” and how much they actually do is comparatively low. (The ratio is 2:1, compared to the other ratios which seem more like 5:1)

    Giving information is important. It’s a good way to let people know how you can be helpful. It seems like maybe not everybody has realized that. We often think “I need X. Who can help me with X?” Maybe we need to think more “I have X. Who else can benefit from X?”

    • Thanks for the comment, Lydia. That’s an interesting note about the ratios for diaries vs. interviews on the info giving element. I think that some jobs involve more giving information than others. While it would be nice if people thought that way more often, it rarely happens unless it’s an actual part of their jobs.

      • That makes a lot of sense to me. I have background in Economics and I have learned that people pretty much respond only to incentives, such as something being part of their job. One of my visions for crowdsourcing is to be able to visualize the small contributions people make, and get credit for them. Letting people visualize their question asking and answering along specific dimensions like you did in this study would be helpful I think. At CSCW can I ask you about how much work it was for people to keep these logs?

  • Interesting that institutional knowledge like “how to do X” was so common. Were people continually engaging in new activities that had support paths they were unfamiliar with?

    • This makes me wonder about socialization practices. In the paper they mention that they recruited from a wide variety of experience levels (something like 1 month in to many years). I wonder whether the results might be a bit skewed by really asymmetric needs by some of the workers. For instance, I can imagine a newcomer having a drastic need for simple institutional knowledge and likely low needs for collaboration until they have adapted. Experienced workers may more often be bridging disciplines or institutional boundaries, changing their needs compared to less senior people. I’d love to see a future study adopt this kind of methodology and take a larger cross-section of an institution to see if these kinds of differences show up.

      • @bernstein I can believe that people have a lot of “how to do X questions” I’m constantly asking payroll people how to do thing that I only have to do once a year. It might not be that the activities are new, just that you’ve forgotten how to do them. These always make easy questions because I know who to ask – the person I asked last time 🙂

        • Yes, both old-timers and newcomers had “how to” questions (both logistical and domain knowledge related). Several participants mentioned that institutional processes frequently change. So, I might have known how to requisition some supply last year, but I have to relearn the process again this year. But also, as part of their job, people constantly need to learn new domain knowledge skills such as “how do I do pattern matching in Scala?” and may turn to colleagues for that help.

  • Sounds like a really worthwhile study, and I look forward to reading that paper and checking out the presentation.

    In your interviews or diary data, did you identify any patterns or themes around the reasons people are sometimes hesitant to ask for help? In some of my research (and everyday observations, personal experience, etc.), there are all sorts of reasons people don’t ask for help even when they need it: they don’t feel they have the proper terminology, they don’t know how to identify SMEs, or don’t even know if such people exist, they don’t know the proper forum or venue, they are afraid of looking “stupid”, etc.

    Curious to hear if you found anything related to that.


    • We didn’t explicitly examine it, since in our study we only had people talk about when they DID look for help. But, there’s lots of social theories that might be relevant here such as Goffman’s presentation of self (e.g., need to appear competent) and social capital theory (e.g., not wanting to call in the favor until you REALLY need it, since you’ll have to return it).

  • I LOVE this! ** Helper-Finding and existing sites / systems are not always good at finding somebody who can help in the workplace ** EXACTLY! I can’t wait to share this with the folks I colleagues … really excellent work! Thank you!