Customize Source of Business Survey

Handbook for Custom SourceOfBusiness Survey

This feature was first added in mid 2008, and had a major upgrade in April 2016. Its
purpose is to accommodate servicers for whom the standard survey does not provide
an adequate set of options.

It allows you to design the survey in its entirety (subject only to moderate constraints).

The first step is to create a SurveySourceFile.

You’ll do this in Excel.

For an example file, see

The simple idea is, this file (or spreadsheet, if you prefer to call it that) has three columns. Column A is for the questions your call-takers will be prompted to provide answers to. Column B is for the answers from which they’ll be permitted to pick. Column C is for links to Next questions.

If you look at the example file, the above principles should be obvious. You’ll see that I’ve drawn in arrows, from particular answers to particular follow-up questions. These have no effect on operation. I placed them there solely to help me mentally keep track of my own logic in structuring the flow. You may elect to do the same as you structure your file (or, if you like, you may of course start with mine and modify).

The third column simply contains the line-numbers for the questions that should follow the respective answers (they are what actually make the link that my arrows signify). You’ll note that many answers do not have these “nextquestion” line numbers. This signifies that, for such answers, the survey is complete (i.e., no more questions).

I recommend that you refrain from filling in the third column numbers until after everything else in your layout is perfected, and lines drawn showing your logic. Then it’s very simple to look at each line/answer that’s got an arrow, see where the arrow leads, and type in that line number.

You’ll also notice that I’ve left many lines with no text at all. These no-text lines have no effect on operation, but help me keep logical track of my thinking as I’m working out the scheme, and also leave potential spaces for adding new items when I think of them. One precaution in their regard: a grouping of answers (i.e., the set that pertains to a particular question) should all be together, with no empty lines between (also, the first answer must always be on the same line as its question).

You’ll further notice that I’ve numerically numbered each answer set. This is not essential, but helps the operator when she’s considering each set of answer options. Also, ServiceDesk is configured so, if the operator prefers, she can strike the number on her keyboard that corresponds to an answer’s ordinal position (i.e., in lieu of using mouse or cursor keys to select the answer). So, if you provide such numbers, they serve as labels for the operational shortcut.

I suggest you structure your query series to leave a sensible “out” if the call-taker does not otherwise have a basis to provide a knowingly accurate answer. Suppose, for example, the consumer hangs up before your CSR can inquire as to how a caller chose your company. Provide an answer that can sensibly be picked (maybe “Unknown”) in that circumstance. Try to assure, with every avenue that you design into the query structure, none of your operators is ever forced to guess at an answer that may not be accurate (“Unknown” is better than false information).

After you’ve perfected your spreadsheet to preference, save it under whatever name you like (and as an Excel file with .xls extention), but be sure to pick a location that is accessible to all ServiceDesk stations. The \sd\netdata folder on your server is, by far, the most sensible choice, especially because each station will auto-interpret that particular path, even if its absolute description varies from one station to another.

Be sure you pick the Excel (.xls) version of the file. After this is done, save your Net-Wide values. It’s a very good idea to additionally save a secondary copy of the file in comma-delimited format (thesesave with a .csv extension). Just assure it has the same name (excepting the difference in extension)and location. ServiceDesk is configured to look for the presence of this alternate, and, if finding thesame, to use it (in alternative to the .xls instance) for simple loading of the survey structure as neededfor survey-taking purposes. This has two benefits. First, you may have stations that are not Excelequipped.Such stations would not be able to open from the .xls source, but can open from the .csvsource just fine. Second, even stations that are Excel-equipped will open and pull from the .csv sourcemuch more quickly than from the Excel source (the latter may produce a two or three second pause asExcel opens and reads; the former will not produce any distinguishable pause at all). There is animportant caveat. Please assure, if you do create a .csv alternate, it is always the same in substantivecontent as is the .xls primary. Suppose you design your structure, save as both .xls and .csv, then realizeyou want to make some substantive changes before actually implementing. You do that work within the.xls, but forget to correspondingly again save as .csv. Now the two do not match. You go to work infollowing days and weeks conducting surveys. But it’s messed up, because the ultimate Survey-Tallyreport will depend on the .xls version, while the results as created each day match the .csv version.Don’t let this happen to you.

The second step is to setup ServiceDesk for use of your custom survey.

Just go to the Settings form (the quickkey combo is Ctrl-F1). Click on the “Enable SourceOfBusiness Survey” checkbox (toward the top-right corner of purple section). This now displays a new window. In it, you can indicate that your Type of Survey is “Full Custom.” You must further inform ServiceDesk (in the space provided) of the filename and location that you’ve setup under Step 1.

The third step is to test.

Just like the normal survey, your custom survey is auto-invoked whenever an operator completes the input of appointment information in the appointment box of a Callsheet. However, you may also deliberately invoke by doing a right-click in any Callsheet’s Criteria button (in its bottom-right corner).

We recommend that you fully test your SurveySourceFile by doing repeated faux surveys from a test Callsheet, assuring via a varying set of answers that the navigation in each instance takes you down the path that you design-intended. You can re-initiate repeatedly from the same Callsheet by using the right-click method, above described (the system will ask if you wish to replace prior survey results, and you may simply indicate yes).

There is no need to worry about these faux surveys going into your real results. The system itself will maintain a file that contains survey results, 2 and this file is not written to until you do the JobSale transition from a Callsheet in which a survey was conducted.

Fully behind the scenes, the system names this file for you. It begins by starting with the same namerootas you’ve picked for your SurveySourceFile, then appends with the expression “_DataTally.DAT”.Thus, if you’d named your SurveySourceFile as “MySurvey2016.xls”, your results file would end up beingnamed for you as “MySurvey2016_DataTally.DAT”.

The fourth step is to operate.

With the above-described prep work completed, the system will immediately begin collecting data.

Please assure your operators know, if they are going through the question series and realize they flubbed on a prior answer, they can hit Esc on their keyboard to re-start from the beginning. Please assure they also know that, even if they have completed the survey, they can right-click on the Callsheet’s Criteria button to re-do (at least up until the point they have done the Job-Sale transition).

It is important that you refrain from making any structural changes to your source file during the course of any period. ServiceDesk is compiling data based on line-numbers as exist in that file. If you change the line-positions of answers (or the essential substance of answers), it will essentially corrupt the data (because there will no longer be a proper correspondence between the particular line positions that ServiceDesk has scored and what’s now in your file).

If you must make structural changes, you should end the prior period and begin a new survey period. Do this, simply by making a new source file (i.e., with a different name as compared to your old one, and setting it as the source within the ServiceDesk Settings form. Automatically, the system will begin compiling new results matching to that new file.

Changes that do not modify line positions (or ultimate meanings in lines) do not require that a new survey sequence be commenced.

The fifth step is to obtain results.

To do so, go back to that same window (as described in Step 1) in the Settings form. There’s a button in it labeled “export Results.”3  Click on it and follow the prompts.

Essentially, you’ll be asking ServiceDesk to create a new Excel spreadsheet. At its core this spreadsheet will contain the same layout as present in your source file, but now it will contain a nice header at top, plus added columns and results information. You should find that it’s a wealth of material to enjoy and digest.

You can create this analytical export as often as you wish, and for whatever spread of dates that you wish. Please do keep in mind that small sample periods can quite easily be skewed (if you toss a coin just four times, it’s not uncommon that you’ll get 75 or 100 percent results in one direction; if you toss it a thousand times; your results are going to be consistently close to 50 percent on each side). Larger sample periods will generally provide more robust results.

Constant Survey, or Intermittent?

Even if you design your survey so that its query structure is relatively brief and simple, its execution still takes a little CSR time. National polling agencies do not eed to ask every single voter how they intend to vote, in order to create a
viable metric regarding voter sentiments.

Nor do you, likely, need to survey during the full course of each year, year after year, to have viable information regarding what is working in your marketing. Rather, a sampling period each year (perhaps as long as two or three months) is likely sufficient. Or, if you have a particular advertising campaign running over a particular period, you might make that a particular period for which you survey.

The general idea is there may be no real need to survey continuously, and to do so continuously may represent somewhat of a wasted effort. At least for many businesses, periodic sample survey periods may make better sense.

End One Survey Campaign, and Begin Anew

Even if you choose to survey 365 days out of every year, you will need to periodically change your survey structure. Among other reasons, you’ll be involved in new marketing schemes that were not formerly included in the structure, and need to be included now. It’s important that each survey sequence be different and separated from its predecessor. It doesn’t mean a particular survey sequence cannot last a long time. If you had no need to change the structure and did prefer to survey continuously, you might potentially have a particular sequence run for years.

Regardless, if either you must change the core structure or you decide to have a hiatus period where no surveying occurs, it is important that any resumed surveying is done under a differently-named SurveySourceFile. In general, it is likely a good idea to pick a name that directly expresses the intended period or its inception (e.g., “BusinessSourceSurvey Fall 2016.xls”).

There are two reasons this demarcation-by-differently-named-source is important:

  1. If you had a hiatus within a single survey-setup (e.g., you survey for two months, turn it off for two months, then resume under the same file name for another two months), your results will be diluted in regard to the expression of rate at which you are getting hits.
  2. If have a structure change in the source (but do not change to a different source filename), the system will end up writing to the results file according to the new structure, on top of results that are already written according to the old structure. It will be impossible for any output tally to ever make sense of these mixed-structure results.

So, please adopt whatever scheme you want in regard to whatever is your preferred succession of surveys as years go by — while simply realizing the need for this kind of separation between instances.