Working With Automated Testing Tools

[article]
A Pragmatic Approach
Member Submitted
Summary:

Unfortunately many corporations make blunders with their automation strategy even before a single test script has been automated with a testing tool. In this article, Jose Fajardo will discuss some considerations that should be examined before purchasing expensive automated testing tools and working with them. The author also suggests some robust automated testing approaches and dispels some common testing misconceptions based on his actual hands-on automated testing experience.

Automated testing tool selection and acquisition
Many companies first learn about automated testing tools through conferences, trade shows, internal contractors, consulting companies, or from employees that have previous experience with one or more automated testing tool. Usually after companies learn about automated testing tools and decide to acquire an automated software solution they get product demos from the vendors, then the vendors provide the companies a copy of their software copy with an evaluation license. After a company's evaluation period of the software expires the company may wind up spending hundreds of thousands of dollars on automated testing software solutions that may not even meet a partial listing of the company’s testing needs.

The need to select and purchase automated software solutions should be compelling and critical while meeting a vast majority of the company's testing efforts. The reader should note that it might not be possible to find an automated testing software solution from a single vendor to meet all of the company's automated testing needs. Companies should have a test manager or a testing champion on-site who is thoroughly familiar with automated testing tools software in general and can provide substantial feedback to the company researching test tools for possible acquisition during the evaluation period. The test manager should articulate clearly and have a vision as to how the automated software solution will be used, how compatible the automated software is with the existing IT environment, who will use the software, and develop a realistic schedule of how many test scripts can actually be developed with the automated testing tool given the company’s already established deadlines for a given software release or deployment.

Test managers should also recognize that the mere fact that a software vendor provides a satisfactory product demonstration for its automated testing tool is not sufficient to guarantee that the automated testing tool will be compatible with the project's application or with the testing team's on-going testing objectives. The test manager should perform due diligence in asking the automated software vendor for a demonstration of its automated software against the company's IT environment. Furthermore, the test manager or the project's testing champion with previous knowledge of an automated testing tool should compose a "wish-list of automated test scripts" from the project's current test scenarios ranging from low complexity to medium complexity that they would like to have the vendor fully automate in their environment before acquiring a specific automated tool from a vendor. The onus and burden should fall squarely on the vendors to demonstrate how their automated testing tools will fit the needs and be compatible with the company’s software application.

Test managers should be cognizant of the fact that many automated testing vendors will make many promises about their tools during their product demonstration or “dog and pony show” to make a sale for their tools when in fact these promises may not be consistent with the company’s testing goals or may go unfulfilled. The test manager should seek software demonstrations from at least 3 different vendors during the tool selection phase and should rely on various readily available automated tool selection matrices before recommending making a purchase for an automated testing tool. Armed with the knowledge of another automated testing tool the appointed test manager or test champion should inquire and verify that the product that the vendor is demonstrating has features and capabilities that exceed or are similar to automated testing products that they have worked with at previous projects.

Minimum Criteria Factors to consider before tools acquisition
At the very least the acquired automated testing software should meet the criteria specified below. The reader should note that it might be necessary to purchase more than one software solution from a vendor or vendors to meet the various testing needs of the project. The criteria is enumerated below.

  1. Automated testing tools should have version control capability
  2. Automated testing tools should have workflow capability for the reporting and closing of defects.
  3. The recording testing tool should recognize the custom controls, objects, GUI, and generic controls (i.e Active-x controls) for the application under test to allow playback of the recorded scripts.
  4. Automated testing tools should allow for the sequencing of test sets with dependencies for the execution of the automated test scripts. For instance execute test script B only after test script C has successfully completed.
  5. The recording testing tool should have a scripting language that is widely accepted, robust and recognized (i.e. Visual Basic).
  6. The recording testing tool should produce reports to verify the execution of the scripts and provide a means to store the execution reports.
  7. The automated testing tool should be compatible and integrated with standard word processors and spreadsheets
  8. The vendor of the automated testing tool should offer online support to allow the customer to report identified problems or bugs with the automated testing software.
  9. The recording testing tool should work with external/internal data sheets, to allow for the creation of parameterized data driven scripts.
  10. The automated testing tool should have the capability for report generation to track and collect metrics for the number of scripts that passed, failed, number of defects open/closes, number of test cases that have been developed, etc.
  11. The automated testing tool should have email notification for the reporting and closing of defects.
  12. The automated testing tool should have an open architecture that makes it flexible enough to modify it to extend the tool’s functionality.
  13. The automated testing tool should have the capability to allow storage of automated and manual scripts and to serve as a repository for test artifacts.

Again the reader should be aware that the delineated criteria 1-14 above is by no means an all exhaustive list of attributes that the automated testing tools should possess before acquisition but rather the aforementioned criteria should serve as a baseline for the minimum features that the automated testing tools should possess.

1. Straightjacket effect:
Companies that fall under this category purchase all their automated testing tools from a single vendor and make a significant commitment of time, money and resources to the automated testing solutions from a single vendor. The company is highly dependent on a single vendor for all their automated testing needs. Although, this approach may be cost effective if a single vendor is capable of meeting all of a company’s testing needs, in actuality this is highly unlikely due to a company's bevy of heterogeneous IT applications. The test manager and the company may instead opt for a hybrid solution of automated testing tools where automated solutions are procured from 2 vendors or more to avoid the straightjacket effect of working with a single vendor and its potential limitations. Below is an example from an actual project where the straightjacket effect hampered one of my client's ability to automate test cases.

Example: My client, a consumer products company for which I did consulting for had spent over 500,000 dollars in automated testing software from a single vendor. The company bought a bolt on to their existing ERP system running in production and wanted to test the ERP bolt-on with their automated testing tools. The vendor offered an add-in program at a cost of an extra $10,000 dollars that supposedly could recognize the objects on the ERP bolt on for record and playback. As it turns out my client company could not record or script at all the newly acquired ERP bolt-on despite the vendor’s promise. Having stretched the budget for automated testing software on a single vendor my client had to abandon its plans to automate the testing of the ERP bolt-on even though there were other vendors in the market that had demonstrated the compatibility between their automated testing software and the ERP bolt-on in question. My client’s only vendor for automated testing software made promises that within 6 months they would deliver a solution to test the ERP bolt-on but in the meantime my client had to test by hand and manually the newly acquired ERP bolt-on and thus rendering the automated tools useless for this effort.

2. Future of the application
Test managers in selecting automated testing software should not only consider the current state of the architecture of their software application but future releases of their software. For instance an automated testing tool may be compatible with a currently installed client server software application that is invoked via a GUI on a desktop, but what if within a future release of the software the new architecture for the software is to become a completely web-enabled solution that is not compatible with the automated testing tools that were recently purchased? If the purchased automated testing software will not be compatible with a future release of the software that will be tested then the test manager may want to postpone obtaining the automated testing software.

3. Vendor Support
Another guideline for a test manager to follow in selecting automated testing software is the quality of the vendor's software support. I was once in a project where the existing automated software did not recognize newly introduced active-x controls within the application under test. I worked closely with the vendor to get a beta version of their software to support my testing needs at the project for the application’s active-x controls. The vendor sent me within a workweek 3 different beta versions of their software until I was able to record and playback successfully against the active-x controls within my application. During the evaluation period for the automated testing software the company purchasing the software should have its test engineers call the vendor for support and evaluate the quality and responsiveness of the vendor’s support when problems arise with the recording of test scripts during the evaluation period.

Automated Tool Administration
After a company has selected and purchased automated testing software from a vendor that meets a vast majority of the company’s automated testing needs the next step is to assign roles and responsibilities for administering the software. The test manager should appoint primary and back-up administrators for managing the administration of the automated testing software. The roles of the tool administrator would be to install the software, install software patches, report bugs and defects with the software to the vendor, configure and customize the software as needed, provide log-on user access to the software as needed, maintain the software's documentation and user's guides.

The test manager is hereby reminded that in many corporations the automated testing tools are not part of the help-desk standard software image and therefore the company's help desk does not support the automated test tools at all when end users report problems with the automated test tools.

Should end users of the automated test tools encounter problems with the automated test tools within the project that are not quickly resolved then the end users may lose faith in the testing tool and resist using it. I have seen this happen in particularly with test management tools that serve as the repository for test cases, and test sets, and defect reporting test tools and have large population of end users. The tool administrator should be able to answer and respond to the questions that the end users present when they have problems using the test tools.

Automated tool’s process owner
In addition to appointing the automated testing tool administrator the test manager should also appoint the testers or group that own the processes for using and working with the automated testing tools. The automated software tool's process owners develop the specifications as to how they want to see the testing software customized. A paradigm of a process that could be customized is the workflow process for reporting defects and the fields that need to be populated for creating a defect based on a company's defect reporting procedures. Another example of a process that could be customized are the fields that need to be populated or created from scratch, within a test management tool in order to create a test case.

Another task assigned to the automated tool's process owners is the creation of customized internal training materials for the end users of the automated testing tools within the project.

The roles of the tool administrator and the tool process owner should be complementary. For example the process owner may specify how the particular test management tool will be customized and in turn the tool administrator would perform the associated tasks to customize the test management tool.

Who will perform the automation
Developing and creating automated test scripts requires programming skills. Many organizations erroneously assume that any tester within the testing team, or QA team has the ability to write code for test scripts because they believe that creating script is a simple matter of just "record/and play back". This is a fallacy and it behooves test managers to avoid this fallacy.

Writing and developing scripts with an automated testing tool requires knowledge of the test script language, the ability to embed exception handling logic within the script, insertion of logic operators, while loops, for loops, parameterization of test scripts, correlation of test scripts, adding verification points to the script, adding logic to recognize objects within an application, etc. Merely recording a script does not assure that a script will playback successfully specially when the script is data-driven. After the test script is recorded the automation test engineer needs to "massage" the script and tailor the script until it successfully playbacks back in a manner consistent in which an end user would execute the transactional steps of a test script.

Just in the same way that writing code and developing code to produce an application requires technical skills so too is developing automated test scripts. The test manager should recognize which members from his/her team will be the core automators with the automated testing tools for the test scripts that need automation and which testers will be responsible for the creation and documentation of the test design and test scripts that will be automated. In some instances a test manager will need to create a test team structure in which some of the testers are divided into subject matter experts with functional and in-depth knowledge of the application under test and core test automators that have knowledge of the automated testing tool but limited functional knowledge of the application under test.

The reader should note that for some applications specially ERP applications the functional testers have much expertise as to how the ERP application was configured and what data sets are appropriate to define test cases and test sets to test the ERP application, however these functional testers have little knowledge if any as to how to properly develop an automated test script that successfully plays back with an automated testing tool.

The test manager should identify which testers within his/her team have the ability to develop automated test scripts and identify a tool champion for script automation within his/her team. In the event that the test manager does not have the in-house expertise to assign an automated test tool champion he/she should seek to hire an automation expert or contractor to impart tool knowledge to the other core test automators. The reader should note that the technical tester might rely heavily for support on the functional tester for functional expertise and subject matter expertise associated with the application under test.

Automated tool training
The test manager will in all likelihood get some days of training for the testing tools from the vendor after the purchase of the automated tools. The test manager should decide which testers within his/her team would be the core test automators and should therefore provide training for the testing tools to the core automators and users in general of the automated testing tool such as the test management tool, and the defect reporting test tool.

Some test managers after choosing test automators to receive training for a test tool, sent the testers training with the concept of "train-the-trainers". The purpose of this concept is to have the "trained-trainers" share their knowledge with other testers within the project that may work with the automated test tools in the future. I have seen this concept worked with mixed results since at the beginning the newly trained core automators and testers do not have sufficient knowledge or know-how of the automated tools to teach other testers right away, and these core automators and testers only gain enough knowledge of the automated test tool after having worked with the automated tool extensively for a period of at least 1 year.

The test manager should also be cautious that often times the vendor of the automated test tool only has generic training with generic training materials on a generic environment which might be considerably different from the test environment in which the automated tools will actually be used against for the project needs. In such a scenario the test manager may request customized training from the vendor for the project specific applications and testing needs.

In any event the test manager should ascertain that the recently trained core automators have a tool champion with them for at least one release of the software in case they encounter difficulties automating test cases and test scenarios.

What to automate
In this document under the section of misconceptions I provide some scenarios under which test cases should not be automated. In general one should automate processes: 1. That will be repeated, 2. That require testing with multiple values of data, 3. That are stable and constant, 4. That require verification of multiple business rules, 5. That are resource intensive, 5. Critical to the application, 6. That will be part of regression testing, and 7. That require integration testing. This is not an all-inclusive list of all the general rules for automating a test script but a set of criteria guidelines for reviewing which test cases are good candidates for automation. The reader should note that in order to automate a test script, a stable or non-changing environment is needed to gain the most benefit from an automated test script. For example if a script is automated and signed-off for a given process on a Monday but the script will be executed on Thursday of the same week but between Monday and Thursday of the same week the underlying recorded business process has changed then the automated test script quite possible may not playback successfully on Thursday. A better approach would be to wait until the process that requires automation is stable and will not be changing until a future release and thus the process can be automated and successfully played back multiple times within the same release or future releases with minor test script modifications if the process has changed between releases.

The reader should note that an automated test script should trace back to either: a requirement, business process procedure, or test scenario.

Hybrid Solution (mixture of tools)
As previously mentioned it might not be possible to acquire a one size fits all automated testing software solution from a single vendor. A company might have to buy automated testing solutions from more than one vendor to meet all the project's testing needs since an IT project may be composed of several heterogeneous IT solutions.

The concept of hybrid testing solutions is not unique to automated testing some companies for instance might purchase a human resource solution from one ERP vendor, yet the company's supply chain solution might come from a second ERP vendor, and the company's help desk solution might come from a third CRM vendor. The company goes through the challenge of integration solutions from competing vendors because they want the best possible solutions for its IT infrastructure since it is possible that a single vendor will not have a world class solution for every critical process such as Human Resources, Supply Chain, and Helpdesk. The challenge now is to obtain the necessary automated testing tools that will enable the company to test how all the pieces of its integrated IT solutions work as a single component and as individual stand along components.

Avoid Redundancies
The test manager should strive to have the project wisely and consistently utilize the purchased testing tools. I was in a project once where various teams were implementing several modules from an ERP vendor. The project had purchased a test management solution that contained defect-reporting capabilities that everyone in the project could have utilized effectively. However one of the teams implementing the ERP solution was not aware of the test management tool’s capability to report defect and this one team had plans to create from scratch a database for reporting defects. This was an unnecessary waste of resources and man-hours. I brought this issue to the attention of the project director and the plans to create a customized a database to report defects from scratch were abandoned in favor of all the teams implementing the ERP solution using the defect-reporting capabilities from the purchased test management tool.

Production rates for testers
A skilled technical tester with a few years of hands on experience should produce 3-4 maintainable, and robust automated test scripts per working day. A fully automated test script means that the script has been recorded and modified with all the necessary programming logic (i.e while loops, data driven, contains error handling logic, is correlated, contains functional calls, contains verification points, etc) and the automated test script has been tested and has functional tester sign-off.

These rates presume that the automated test scripts are recorded for test scenarios that range from medium to complex difficulty and require more than just record and playback.

ROI
The return on investment on automated testing tools is best qualified as a series of intangible benefits since demonstrating how a particular automated testing tool saves money over manual testing might be an intractable challenge. Many vendors drum up their automated testing tools with a plethora of benefits that few projects ever actually manage to gain.

The main benefits of purchasing automated testing tools and using them to automate test scripts are: 1. Consistency, 2.Under certain conditions (i.e. regression testing, and benchmark testing) a previously recorded test script is repeatable, 3. Automated test scripts may help save time during the test execution phase, 4. Electronic storage of test results after test execution, 5. Some automated test tools produce graphs, results that would be inconvenient to produce by hand, 6. Execution of test scenarios with multiple data sets for parameterized data-driven test scripts, 7. Integration testing of various business processes for correlated test scripts, 8. Automatic sequencing of test sets with dependencies, 9. Automatic validation and verification of business objects for the application under test and 10. Batch scheduling of test scripts for execution. There are various other benefits gained from using automated testing tools such as: the workflow activities during the reporting and closing of defects, the automatic collection of test metrics from test management tools, the mapping of requirements to test cases, a central repository to store all test scripts and automated test scripts, etc. Again this is not an all-inclusive list of benefits of working with automated testing tools but just some of the salient benefits of working with automated testing tools.

A company may save money by not hiring as many testers to perform test execution during the test execution phase after the company purchased an automated testing solution and thereby lowering the overheard costs associated with test execution and this may be perceived as a return on investment from the purchased test tools. But what about all the costs associated with maintaining the automated test tools, and the costs of hiring technical testers to produce and maintain automated test scripts? These costs may easily outweigh any savings incurred by not hiring as many testers to execute test scripts by hands and thus the return on investment from the automated test tool may prove difficult to quantify.

However I do think that over time although the automated test tools might not generate any tangible ROI the test tools do bring the project a level of convenience and assurance that may not otherwise be available in the absence of an automated testing tool.

As an example of the convenience that the automated tools may bring to a project I will relate my experience from a previous project. I was at a project where a tester executed test scripts by hand during the middle of the night and the tester had to verify the test results before the morning shift started. I created an automated test script for the same process that this tester executed by hand and embedded programming error handling logic within the script to send an alarm to a pager if the application under test encountered a given set of errors during testing. This tester no longer had to work the night shift to manually test the application since I had automated the process for him and I embedded test script logic to alert him if anything fails with the application during the batch execution of the test. Needless to say the automated testing tool brought much convenience to this tester.

When built correctly automated test scripts that are data driven can cover more test scenarios and permutations of business processes than a tester can cover through manual testing. For some companies covering more test scenarios with different data values may mean the identification of more defects and bugs before an application under test is released commercially. Some companies may prevent lawsuits, loss of contracts, financial losses and damage to their reputation by building more robust software applications that meet all the intended requirements. These are areas where automated test tools may maximize their ROI. For example the automated testing tools may prevent a company from losing millions of dollars though a lawsuit due to their faulty software product or may help a company generate more revenue by keeping satisfied customers.

Tool location
The test manager should decide where the test tools would be installed. For instance some projects that I have worked with have dedicated test labs with hardware equipment for the testing team, where the tool administrator can install the automated testing tools. Other projects that I have been involved with do not have a dedicated test lab and the automated testing tools have to be installed on individual testers' desktops. I have also seen examples where the installation of the testing tools whether at a test lab or individual desktop is irrelevant since the automated testing software is web based and can be invoked from a web browser without requiring any installation at the client machines.

Misconceptions of automated testing
1. Repeatability
Often vendors trump their automated testing products as the panacea for having consistent and repeatable test scripts. In fact a selling point for a software vendor is the perceived notion that test scripts once automated can be re-used over and over and thus saving testing time. Although I would agree that this argument from the vendor has merits, this argument is shortsighted and suffers from the tunnel vision effect. In order to have a repeatable test script that is valid for re-use for various software releases the underlying test conditions under which the test script was recorded, and the underlying application under test's data and environment would have to be static and constant every time the automated test script is executed. This is seldom if ever the case. Most projects have their software undergo some sort of change from release to release, after bugs are discovered, after customer complaints, or after undergoing a gap analysis. Once a test script is recorded and it is underlying conditions on which it was recorded have changed the script needs to be modified and maintained for correct playback on an application that undergoes changes.

We can safely say that test scripts are repeatable when conditions for the application under test stay constant after the script was recorded. And this concept works during a single release when the testers conduct regression testing and in essence re-execute the previously executed test script without making any modifications at all to the script or possibly to the data that the test script is accessing if the test script has been parameterized.

However, for the most part given the nature of ever changing IT applications one may assume that test scripts will require modifications after the underlying conditions in which the test script have changed and are repeatable only to the extent in which the underlying test conditions under which the test script were recorded remain constant or static.

2. Automating a script only involves Record/Playback
As previously mentioned automating a test script is more than merely recording and playing back the test script.

The challenge of automation is to identify skilled personnel with a sufficiently proficient technical background to create scripts, develop or revise the code to automate the script, inserting loop statements, insert verification points, add error-handling logic, etc. Merely recording a user action or keystroke against a business process does not by itself create a fully robust automated script or even a valid reusable script for that matter. A functional tester may have some mild success in creating scripts for simple tasks or business processes such as viewing a report or viewing a business process in read-only mode, but this does not necessarily mean that the functional tester can create data driven, correlated, maintainable and fully robust scripts.

I hereby place a special emphasis in separating the roles of the functional expert and the technical expert while these two types of testing roles need not be completely independent they can certainly specialize in different areas. For the functional tester an are of specialization might be drafting test cycles, drafting test scenarios, sequencing of test cases, drafting test sets, providing navigation support and transactional steps within the test script, identifying test data for the test script, ensuring that requirements are tested via RTVMs, etc. For the technical tester (core automator) on the other hand an area of specialization might be to create fully robust, data driven, and maintainable automated test scripts and analyzing the results from the test script execution.

3. Everything under the sun can be automated
A common misconception is to expect that every single requirement, test scenario and test case can be automated. According to industry best practices one does not automate testing cases for all business processes and test scenarios. The following criteria for instance identify business processes and test scenarios that would not be good candidates for automation: business processes/test scenarios that will only be executed once; business processes/test scenarios that require intuitive testing, test scenarios that require negative testing, test scenarios that rely extensively or solely on analog and bitmap testing which could be impractical since recoding a test script based on pixels and screen coordinates may be unpredictable, and recording of test scripts on an unstable or frequently changing environment. Under these aforementioned conditions the test manager should employ manual testing techniques.

4. The automated tool is the means to document the test design, and test script
This is a misconception that I have seen at some projects in which test leads incorrectly believe that recorded automated scripts serve as the basis for a documented test script. There is no substitute for a well-documented test script that is written in coherent English. The test script should include at the very minimum a description of the steps to be captured with the automation tool, the expected results of the step, the data values that need to be populated within the application under test to record the test script, the output that needs to be captured, any pre-work needed to record the test script, and a description of the warning and information messages that the system may display during the recording of the testing script.

A technical tester (core test script automator) should have the ability to work with a self-explanatory, thoroughly documented test script before proceeding to automate it. An approach that requires a technical tester to divine what the testing steps ought to be from a poorly recorded script is not a suitable approach for script automation and is counterproductive as well as inefficient. To augment this argument in the absence of a well written test script the technical tester would not know if the steps that he/she is recording are malfunctioning because the application under test has a problem or is malfunctioning. In order to have fully maintainable and properly recorded automated test scripts the technical tester or core test automator should have access to a well-constructed and fully documented test script that serves as the basis for recording automated test scripts.

5. Automated testing speeds up the testing effort
This misconception is pointed out and that is that many test managers and project directors mistakenly believe that creating automated scripts speeds up corresponding testing activities. Creating automated scripts is a daunting, time-consuming and monumental task in particular when one is doing so for the very first time on an entirely new release of a software system. Creating automated scripts requires having valid documented test cycles, test scenarios and test scripts that can be translated into an automated script. Planning for the creation of a test case and test script with valid data is in itself an initial input for the automated script (the output), since the automated script is a by-product of the testing cycle. Identifying the requirements and scenarios that need to be tested, and creating the testing cycles and automated scripts are resource intensive and time-consuming activities that pay-off for subsequent releases of the software where the previously created automated scripts can be re-executed with minor modifications or "as is".

The initial hurdle and obstacle is to create the first batch of automated scripts and then leverage off these automated scripts for later software versions or releases with minor modifications or no modifications at all. Realistically, a testing team that can successfully automate 40-50% of the test cases that are applicable for automated testing when test case automation has been attempted for the first time has indeed met a major testing challenge./p>

6. Maintaining scripts a previously recorded test script is always easy
This assumption is wrong! Maintaining some automated test scripts may prove to be a logistical nightmare. Sometimes a core automator may have to modify an automated test script that another core automator created but has left the company already. If information is missing as to what the automated test script is supposed to do, or if the previous core automator that left the company did not follow best practices in developing the test script the new core automator might struggle to modify the previously recorded the script or worst yet might have to create the new test script from scratch. Another obstacle that makes test script maintenance difficult is that some parameterized scripts require multiple data sets, which may create multiple permutations for executing the automated test script since each data set in the parameterized test script might require error handling or if/else programming logic.

The test manager should ascertain that best practices are followed in creating the automated test script along with appropriate documentation for executing the script and comments embedded within the test script. Furthermore, to the extent possible automated test scripts should be modularized.

7. Functional testers record scripts and the technical testers revise the poorly recorded scripts
A final misconception that I would like to bring to everyone’s attention is that some test managers actually have functional testers record test scripts even though they know that the functional testers have little or no knowledge of the automated testing tool. The rationale for this misconception is that test managers want to have skilled technical testers (core automators) worked off a previously recorded script and then revise the test scripts that were poorly recorded by the functional tester as a means to speed up the automation testing activities. This practice is highly inefficient and I would advise against doing this.

Functional testers often introduce recorded steps into the test script that were recorded by accident or erroneously then the test managers expect that the core automators can "fix" any poorly recorded test script. The test manager should separate the roles of the functional tester and the technical tester ("core automator") and only have the technical tester recording automated test scripts.

An analogous situation would be with software development where you have a team of functional analysts gathering requirements and creating the documentation for the project's specifications from which the development team would work off the functional specs to write the programs for the application. The functional analysts would NOT initially start coding in an object oriented language or some other programming language and then the development team would go and "fix" the programs that the functional analysts coded. The same principle applies to the automation of test scripts. The functional testers perform the initial test planning activities and the creation of the test scripts, test scenarios, test cycles, test sets, etc then the technical tester would work off the fully documented test scripts and test scenarios and proceed to automate the test script with an automated testing tool.

The functional tester and the technical tester can complement each other's work for instance after the technical tester has created a robust test script the functional tester can verify that the script correctly plays back against the application and provide sign-off on the automated test script.

Conclusions
This article illustrates some commonly held misconceptions about automated testing and articulates some insights that are optimal for working with automated testing tools such as guidelines for automating test scripts. The reader also has is informed on techniques for selecting ad acquiring an automated testing tool, how to implement the test tool within a QA team, and what are the expected benefits of working with automated testing tools.

About the author

AgileConnection is a TechWell community.

Through conferences, training, consulting, and online resources, TechWell helps you develop and deliver great software every day.