First Steps to Value

Process Analysis from Scratch

Sometimes there are set-piece use cases such as which reps are sandbagging or which products cause the greatest customer support burden. Where you don't have an out-of-box analysis project, you have to start from scratch. But fear not, it's easy if you follow the steps. This is also a good tutorial for getting up to speed with Hubbl Process Analytics once you've installed it. 

Before you Open the Tool

When you first open up Hubbl Process Analytics you will see your process flows from start to finish. You may see processes incorporating steps that you did not expect or simply getting stuck. You might see a large number of cases ending unsuccessfully. Or certain flows taking way longer than they should. However, before you open up the tool, in fact this goes for any tool, you should have an idea of what you are going to do when you get in. Begin by developing an analysis hypothesis. 

Following your Hunches

We believe that most people managing or executing business processes have a very good idea of what is wrong with the processes. They know well how customers react when things are taking too long or even fail. They know which processes take up most of their time, and probably have very good ideas on how to reduce cycle time. 

Six-steps to quantifying the cost of inefficiency

What you might not be so aware of is how many cases or opportunities this inefficiency actually affects and what might be the total time spent each month on unnecessary activities. The following six steps will guide your thinking to develop a hypothesis. Spend no more than 15 minutes on this exercise.

Step 1: What’s the process?

• Pick a Process: Decide on the process you want to understand, is it commonly recognized as a business process? 
• Name: What is the process called by everybody
• Start and End: Where does the process begin and where does it end? 

Step 2: Possible outcomes?

• List 3 outcomes: Decide on the process you want to understand, is it commonly recognized as a business process? 
• Positive and negative outcomes: Label the outcomes that you want to happen and also the undesirable outcomes
• Worst outcomes: Sort the list putting the most undesirable outcomes at the top

Step 3: Where are your problems?

• Going too slowly: Identify 1-3 process steps that you know go too slowly
• Causing a lot of work: Identify 1-3 steps that cause the most work for team members
• Most serious: Sort the list problem steps, putting the most problematic steps at the top

Step 4: Where are the bottlenecks?

• Process gets stuck: List 1-3 different situations where your process gets stuck
• Customers in trouble: List 1-3 situations where customers call support or just abandon
• Most painful: Sort the list putting the most painful at the top

Step 5: Prioritize for analysis

• Issue #1: The lists of issues pick the most serious. Perhaps the greatest waste of time.
• Issue #2: Now pick another. Maybe the most painful one for the customer.
• Issue #3: And one more, this could have the greatest actual or opportunity cost.

Step 6: Where to look first

• Hunch #1: For each problem, might it be something to do with the product or service that you are selling?
• Hunch #2: For each problem, could it be because of the type of customer or some exception to the rules?
• Hunch #3: For each problem could there be a lack of information to make optimal decisions?

Opening Up Hubbl Process Analytics

Armed with your hunches as to where to look, you can open up the tool and get to work. Fortunately Hubbl Process Analytics makes it fast and easy, 

Confirm the problems exist

When you open the tool and look at the process, and see which are the most popular variants. Then apply the filters to find the problems you have prioritized.

What’s the impact of the problem?

Once you have identified and confirmed the problems, the next step is to look at the difference between where they occurred and where they didn’t occur. You can then quantify the extent of the problem, how much longer is it taking than it should? 

What are the low-hanging fruit?

How many cases end unsuccessfully? What is the cost of the problem in people’s time or lost customers? Which problems should be prioritized for a resolution based on quantitative evidence?

Pick a Process to Analyze

‍

STEP 1: Select the object (e.g. opportunity)

Hubbl Process Analytics runs on your Salesforce org and accesses data from the Opportunity History and Case History objects 1. 

Give the analysis project a name that you will be able to easily identify from others in the org. 

STEP 2: Select records to visualize

Opportunities and Cases may be used to support multiple processes as denoted by the ‘Type’ field. You can limit the data to just one of the record types. 

For example Opportunities may be ‘New Business’ or ‘Exisiting Clients’. Cases may be ‘email’ or ‘phone’. 

STEP 3: Select fields to filter on

A number of attribute fields are added by default and you can add additional fields when you create the process. 

For example, for Opportunity you may want to add Territory or Expected Revenue if you think they affect performance. Or for Case add Origin or Priority. 

Select the Timeline

‍

STEP 4: Select date-range to show

Select the date range for data to be analyzed. You will be notified if there is no data in the selected range or if the volume of data exceeds the maximum allowed in your subscription plan.

STEP 5: Object history data is imported to Hubbl Process Analytics

The system then uses your parameters to select the data from the org and prepare it to be visualized. This normally takes a few minutes depending on how many records are available.

‍STEP 6: Visualization is automatically generated

When the visualization is prepared, you will be able to examine the process model immediately.

‍

The visualization: What you’re seeing

1. Hubbl Process Analytics draws a process model of each unique sequence of steps taken from start to end: a process VARIANT
2. Each STEP (corresponds to an stage) is shown as a box, containing the step name and the number of cases that pass through it. 
3. The SEGMENTS are the lines between stages, they show the elapsed time between the start of the previous step and the start of the following step
4. The dotted lines from the green start marker show the first step performed. From the last step to the yellow end marker show the final step performed
5. Segment lines are colored green, yellow and red depending on the time taken relative to other steps in the process 
6. The blue button opens the analysis panel

‍

Variants consuming most total time

1. Here you can select which process and view show, displaying the date range
2. The initial view shows the which unique variants take the most total time, many records and long end to end times
3. The variant selector shows each individual process variant (path taken). You can highlight one by mousing over the table row, and select or deselect for display using the checkboxes. The header row in the able allows you to sort by each column
4. The yellow flags indicate variants with a loop or variants that end to end are taking relatively longer than others
5. Statistics at the foot show what from the full set of records are being displayed

‍

Cleaning up the process diagram

1. The green start marker is where each variant begins. Dotted lines go from that to one or more of the process steps. 
2. A process record may have started before the start date. Similarly process records may have ended after the end date. So the model shows incomplete process records.
3. Open the filters tab and select end-points. Leave only the valid first steps and valid last steps checked. Apply the filter.
4. Of all the variants, some will be edge cases having only a single record. For analysis, you can hide these.
5. Back on the visualizations tab, go to the ‘Quick Select Variants’ dropdown and select variants with more than 2 records.

‍

Are processes operating as you expect?

1. Remove all filters and select all variants, look at every step on the diagram. Do you see steps that you did not expect to see? What frequency of records do they have? 
2. On the variants panel, select all variants so the total contributes to around 80% of records (see the counter at the foot). Do these look like the main sequences that are executed?
3. Use the Endpoints filter to retain only the valid start and end steps. Quick select variants with 2 or more records (i.e. no outliers) and look at the average time to complete. Is this elapsed time as expected or is it significantly longer or shorter.

‍

Using the filters to hone in on issues

1. When you created the process you captured historical data to visualize. Each record represents a journey through the process.
2. The filters tab on the right allows to you include or remove records based on:
3. Timeframe (between 2 dates)
4. Field Values (e.g. = owner name)
5. Performance (number of events)
6. Activities (including specific steps)
7. Endpoints (with first and last steps)
8. Follower (events followed by events)

You can apply multiple filters and set logic as to how they are applied.

When you return to the Variants tab you will see a reduced number of records on display after the filters have been applied.

You can also view the source records to inspect the full detail.

‍

Drill downs to get closer to a root cause

1. Sometimes certain segments run slower or faster than others and you want to understand why.
2. Click on the segment line and you will see more details: Path frequency, Record frequency, Max repetitions, Coverage. You will also be able to select all variants with this path. And using the Drill Downs button you can access even more details that break down the step.
3. The records in this segment are shown on a frequency histogram that breaks down the time taken
4. Sliders allow you to focus in on records that are taking longest or working quickly
5. You can select any of the fields available and to show more details for the records selected on the histogram
6. The table displays the values for each attribute. For example which agents owned the service tickets that were running very late?

‍

Compare views side-by-side

1. When you have identified a potential problem or you have found a star performer and want to know what makes her stand out, compare will help you see the differences
2. Create a view where you are isolating some factor by applying various filters (e.g. remove invalid endpoints). Save as “NAME #1”
3. Create a second view that you want to compare it with (e.g. remove invalid endpoints and select a specific owner.) Save as NAME #2”
4. On the Home tab, select the process and then select the two views and press ‘Compare’. The third view combines both views
5. On the combined view color coded you see the different times and process steps taken by each.

‍

Verify a problem, e.g. average handle time

1. Salesforce reports say that the Average Handle Time for Service Tickets is too long. 
2. Create a new process specifying a time period where the problem exists, select all the fields that might have an effect on handle time (e.g. product and escalation) 
3. In the Variants panel, look at the average duration, is the elapsed time representative of the problem
4. Use the quick select: Top 5 variants by longest avg. time. Are the slowest variants causing the problem?
5. Are there segments that are indicative of the delay? Use the drill down function to explore the slower segments, find the records to the right of the histogram and cycle through looking at the field values for them
6. If you find certain fields are consistently appearing in the slow segments, then use the attribute filter to restrict the visualization to those values. Save that as a view (Save-As), then use the compare feature to see the two views side by side to understand the differences.

‍

Getting your arms around the process

1. What are the steps required to complete the process?
2. Apply endpoint filters to only show valid start and end activities. Save the view [All Variants] In the variants view, sort on frequency. Working down the list, select enough variants that make up around 80% of the records. Look at the activities shown—this is likely to be your business process. Save the view [Top 80% Variants]. Use the compare function to see what the outlier steps might be. 
3. How long does each step is likely to take on average
4. Use the Duration view of the process to see how long steps are taking. How should these vary by person, type, product? What are the pain points being experienced by customers or team members?

What are the headlines, what’s the pain?

1. What are the headlines you’d want to report?
2. Are there steps that you know are wasting time? Drilling down into segments that have a high average time reveals the distribution of that step by time. “We kept 30% of our customers waiting over a week before we got back to them” 
3. Which groups felt the pain of a process delay? 
4. a. Use the sliders on the histogram plot to select the records taking the longest time. 
5. b. Then use the field dropdown to select the a field and inspect the values to see where the problem might be. 
6. “Cases for Gift cards and promotions contributed to 70% of the delays” 

‍

Thinking velocity, variety, and visibility

What is the effect of speed?

Can the process be improved by making it faster—or slower? If you can speed up a step or the total time, what does that do, will it allow more cases to be closed or opportunities to be won with the resources available? Study the variants looking at average completion times and number of records

What causes the process to be different? 

Use the filters to see what causes a process to be faster, slower, take more steps, get to a certain result. Filter on owner, deal size, case type, customer size, industry etc. You may need to re-upload the dataset to bring in more attribute fields. 

How does the process affect customers?

Which steps in this process are carried out by customers? Which steps are transparent to customers? Which steps, if delayed, will be painful to customers?