Quick Counting - Listknife
James Lawson reviews a simple but very effective fast counting tool.
Time was, all that any database marketer wanted was to have swift access to their data. Running counts and pulling out selections in minutes, no messing about with the IT department - what a dream scenario. Much more may be possible in automated direct channel marketing now, but a surprising number of volume mailers are still in the same position. Listknife might just be the tool they are looking for.
No frills or fancies
Forget about automated multi-wave, multichannel campaigns, triggered communications and clever SOA integration. Listknife is a batch tool in the classic mode, and is all about exploring large datasets quickly.
There's no disputing that it's easy to use. Drag-and-drop Windows functionality plus a selection of wizards make this a code-free experience - no SQL here. And it certainly is fast. The demo system was able to run a complex multivariable count totalling 600,000 records from a 35.5m list in under a second.
The graphical drag-and-drop query builder is the outstanding feature of Listknife
The stand-out feature of the package, and one that must have taken some considerable development, is the graphical query builder. Extremely simple to use, it is one of the more intuitive examples of the breed and something like the AnswerSets tool (last heard of around 2002) that employed interactive Venn diagrams to achieve much the same results.
Users simply choose a variable from the listing down the left-hand side and drop it into the main selection builder ("Count Logic") window to have it included within the selection. You can even drag and drop data straight from the Windows desktop. Clicking on the resulting variable object brings up a dialogue box within which you can refine the details of your count for that variable. For example, if you have included all UK cars in your selection, do you only want Mercedes?
Boolean operators are dragged and dropped in the same way to form boxes within the Logic window (the main window itself effectively acts as an AND function, combining everything within it), and to have that operator act on a variable or another operator, you simply drag the operand inside the operator box. So to exclude all people with CCJs or suppress records from a previous count, you would place the CCJ variable within a NOT box. To build complex multivariable queries, you simply nest blocks of objects to combine AND, OR and NOT operators.
Tabular profiling gives list managers what they need to know
It sounds tortured but is actually very simple in practice as the screenshots on this page should hopefully show. Lots of nested statements might get a little messy but probably less so than any other graphic selection tool.
Clicking on the "i" button on the top row then runs the count, which can easily be saved as either data or the query itself. A useful small display in the bottom corner updates the count as you then make changes to it. Other nice tweaks include flagging individual records as members of the same household during the database build. This means you can immediately get a count of the number of households within an individual-level selection.
With a selection live, clicking on the Profile tab at the top of the main window opens up a tabular profile report on the attributes within the current count: the percentage that any variable makes up of universe volume, the percentage of selection volume, and the index and z-score built from those two figures.
In same way, you can explore profiles of variables within the whole database by simply dropping them onto the Profile window and it's also possible to compare two profiles side-by-side. There's a simple wizard for this but it's barely necessary.
The third and final function of the main window besides profiling and selection building is a cross-tab function that can handle two or even three variables. Like the rest of the package, this is simple but elegantly implemented and very fast. Again, you simply drag and drop the variables you want onto the window, while the cross-tab values can be dragged straight into Excel.
Two- or three-dimensional crosstabs are the work of seconds.
There's a good data browser that can handle up to a million records on its own. Showing its list management roots, the package also offers decent tools for splitting out lists, for example, into males and females or to create test cells or 1 in N selections. Another good idea is to be able to flag a record as part of a test cell on the database.
Reporting is minimal. Clicking the Audit function on the bottom window gives you a tabulated report of the current count process, but further charting and manipulation would have to be done within Excel. For this, you can simply drag and drop selections into Excel with a single click rather than having to set up an export. Further connectivity employs ODBC to access databases, though no native connectors for the likes of Oracle appear to be available. Text files for import and export are another option.
Another notable feature of Listknife lies in how it is built around the use of URNs rather than holding customer names and addresses. In this age of corporate concern over data security, it means that losing a laptop full of customer data doesn't necessarily put your company on the front page of next morning's papers.
It also gives a way for the package to easily match lists without having to get into full-blown name and address processing. For example, if you have the URNs of a list of respondents from a campaign, you can easily import them and match them to the original target list by using URNs.
When it comes to production, one restriction could be its lack of remote access capability. Though the vendors say that remote access is possible, there are speed implications and no current customers use it in this way. Listknife does have an up-to-date .NET platform so there should be a way around this, but it's an isolated batch tool as it stands, and as such, a tad lacking in today's connected world.
Dedicated functionality makes splits and control cells easy to set up.
Like almost all fast counting tools, Listknife has a proprietary, highly-indexed database structure and requires a pre-load import stage to convert data into this format. Both of these may well deter a company's IT department, but it's par for the course in this area. A standalone utility is used for the import, with five million records claimed to take an hour's processing. No "trickle feed" updating is possible; the whole database has to be rebuilt to incorporate any changes.
Focused on lists
For end user marketers who run a few big monthly campaigns around a monthly database update cycle, Listknife would do most jobs perfectly well. Where one person has only, say, half their time allocated to handle mailings, they could easily jump in front of Listknife and sort out that month's selections. The main competition comes from other straightforward fast-counting tools like Minotaur or a low-cost standard application like SQL Server running on fast hardware. Many tools today can do much more, often as part of a wider suite that takes in stats, web-published reports and campaign management functionality. Something like FastStats Discoverer is a much more capable package.
But if running lots of counts and generating lists is where it's at, then Listknife could well be for you. The name is something of an indicator: it's perfect for list owners and managers. In fact, consumer data specialist The Trading Floor liked the tool so much, it bought the company.
Costs and specification
Listknife pricing is by annual fee with a nominal set-up charge. The fee tends to be from £14k to £20k depending on the number of users and database size/complexity. Listknife only runs on Microsoft operating systems. www.thetradingfloor.co.uk
(Database Marketing: Software Review – March 2008)
Time was, all that any database marketer wanted was to have swift access to their data. Running counts and pulling out selections in minutes, no messing about with the IT department - what a dream scenario. Much more may be possible in automated direct channel marketing now, but a surprising number of volume mailers are still in the same position. Listknife might just be the tool they are looking for.
No frills or fancies
Forget about automated multi-wave, multichannel campaigns, triggered communications and clever SOA integration. Listknife is a batch tool in the classic mode, and is all about exploring large datasets quickly.
There's no disputing that it's easy to use. Drag-and-drop Windows functionality plus a selection of wizards make this a code-free experience - no SQL here. And it certainly is fast. The demo system was able to run a complex multivariable count totalling 600,000 records from a 35.5m list in under a second.
The graphical drag-and-drop query builder is the outstanding feature of ListknifeThe stand-out feature of the package, and one that must have taken some considerable development, is the graphical query builder. Extremely simple to use, it is one of the more intuitive examples of the breed and something like the AnswerSets tool (last heard of around 2002) that employed interactive Venn diagrams to achieve much the same results.
Users simply choose a variable from the listing down the left-hand side and drop it into the main selection builder ("Count Logic") window to have it included within the selection. You can even drag and drop data straight from the Windows desktop. Clicking on the resulting variable object brings up a dialogue box within which you can refine the details of your count for that variable. For example, if you have included all UK cars in your selection, do you only want Mercedes?
Boolean operators are dragged and dropped in the same way to form boxes within the Logic window (the main window itself effectively acts as an AND function, combining everything within it), and to have that operator act on a variable or another operator, you simply drag the operand inside the operator box. So to exclude all people with CCJs or suppress records from a previous count, you would place the CCJ variable within a NOT box. To build complex multivariable queries, you simply nest blocks of objects to combine AND, OR and NOT operators.
Tabular profiling gives list managers what they need to knowIt sounds tortured but is actually very simple in practice as the screenshots on this page should hopefully show. Lots of nested statements might get a little messy but probably less so than any other graphic selection tool.
Clicking on the "i" button on the top row then runs the count, which can easily be saved as either data or the query itself. A useful small display in the bottom corner updates the count as you then make changes to it. Other nice tweaks include flagging individual records as members of the same household during the database build. This means you can immediately get a count of the number of households within an individual-level selection.
With a selection live, clicking on the Profile tab at the top of the main window opens up a tabular profile report on the attributes within the current count: the percentage that any variable makes up of universe volume, the percentage of selection volume, and the index and z-score built from those two figures.
In same way, you can explore profiles of variables within the whole database by simply dropping them onto the Profile window and it's also possible to compare two profiles side-by-side. There's a simple wizard for this but it's barely necessary.
The third and final function of the main window besides profiling and selection building is a cross-tab function that can handle two or even three variables. Like the rest of the package, this is simple but elegantly implemented and very fast. Again, you simply drag and drop the variables you want onto the window, while the cross-tab values can be dragged straight into Excel.
Two- or three-dimensional crosstabs are the work of seconds.There's a good data browser that can handle up to a million records on its own. Showing its list management roots, the package also offers decent tools for splitting out lists, for example, into males and females or to create test cells or 1 in N selections. Another good idea is to be able to flag a record as part of a test cell on the database.
Reporting is minimal. Clicking the Audit function on the bottom window gives you a tabulated report of the current count process, but further charting and manipulation would have to be done within Excel. For this, you can simply drag and drop selections into Excel with a single click rather than having to set up an export. Further connectivity employs ODBC to access databases, though no native connectors for the likes of Oracle appear to be available. Text files for import and export are another option.
Another notable feature of Listknife lies in how it is built around the use of URNs rather than holding customer names and addresses. In this age of corporate concern over data security, it means that losing a laptop full of customer data doesn't necessarily put your company on the front page of next morning's papers.
It also gives a way for the package to easily match lists without having to get into full-blown name and address processing. For example, if you have the URNs of a list of respondents from a campaign, you can easily import them and match them to the original target list by using URNs.
When it comes to production, one restriction could be its lack of remote access capability. Though the vendors say that remote access is possible, there are speed implications and no current customers use it in this way. Listknife does have an up-to-date .NET platform so there should be a way around this, but it's an isolated batch tool as it stands, and as such, a tad lacking in today's connected world.
Dedicated functionality makes splits and control cells easy to set up.Like almost all fast counting tools, Listknife has a proprietary, highly-indexed database structure and requires a pre-load import stage to convert data into this format. Both of these may well deter a company's IT department, but it's par for the course in this area. A standalone utility is used for the import, with five million records claimed to take an hour's processing. No "trickle feed" updating is possible; the whole database has to be rebuilt to incorporate any changes.
Focused on lists
For end user marketers who run a few big monthly campaigns around a monthly database update cycle, Listknife would do most jobs perfectly well. Where one person has only, say, half their time allocated to handle mailings, they could easily jump in front of Listknife and sort out that month's selections. The main competition comes from other straightforward fast-counting tools like Minotaur or a low-cost standard application like SQL Server running on fast hardware. Many tools today can do much more, often as part of a wider suite that takes in stats, web-published reports and campaign management functionality. Something like FastStats Discoverer is a much more capable package.
But if running lots of counts and generating lists is where it's at, then Listknife could well be for you. The name is something of an indicator: it's perfect for list owners and managers. In fact, consumer data specialist The Trading Floor liked the tool so much, it bought the company.
Costs and specification
Listknife pricing is by annual fee with a nominal set-up charge. The fee tends to be from £14k to £20k depending on the number of users and database size/complexity. Listknife only runs on Microsoft operating systems. www.thetradingfloor.co.uk
(Database Marketing: Software Review – March 2008)
0 Comments:
Post a Comment
Links to this post:
Create a Link
<< Home