How To Use Crush Errors Online & Excel Add-in
Workflow Documentation & Tutorial for For AI-Automated Analysis, Deduplication, Comparison, Fuzzy Matching, Reconciliation & Reporting
- Import (Drag-drop or Upload) One or More Excel Workbooks or CSV Files (or use Excel Add-in)
- Auto De-dupe (Remove Duplicates, even Partial Ones)
- Confirm Auto-Match (Auto Consolidate, Map & Correct Near Matches with Fuzzy Matching)
- Auto Group / Combo Matching (Plan vs. Actual, Budget vs. Payments, Items vs. Groups, Activity vs. Subtotals)
- Auto Eliminate (Move to Report) Perfect Matches
- AI-Guided Matching of Remaining
- Optional Rating & Commentary for Matches/Corrections
- AI Generated Reports
- Export to Reports or Corrected/Commented Data to Excel/CSV
- Optional In-place Corrections & Report Updates with Excel Add-in
Steps in Workflow Diagram
- Export or gather the data you want to analyze, compare, fix, de-duplicate, reconcile or report on with Crush Errors and identify which of those or other supported activities you wish to utilize Crush Errors AI to automate.
- The user should look in the Apps & Templates Gallery for a solution that precisely suits their needs. For many cases, a solution has already been built that will guide the user through the process step-by-step. More Apps are being added all of the time.
- If there is no App, then the user should try the Setup Wizard. Or, the user can learn some of the basic tasks to figure out how to configure a Crush session. For many cases, it is easy. For some, it can be a learning curve. Most of the work is to A) filter the rows that really needed and B) add math calculations, like instead of debit and credit columns, you'll need to make one Amount column that's equal to debit - credit. These row filters and new math columns can almost always be done in Excel. So it is recommended that you filter the rows in Excel and add the appropriate columns in Excel. At some point, when the user understands Crush better, everything can be done in Crush.
- Users typically need some trusted data. This needs to be cleaned up so that it's set up in structured rows and columns. If you only have PDF's, you'll need to convert them to CSV's. Contact Crush if you need help with this because we have a great partner, PerfectAudit.com, that does this work. It's amazing and unlike any other PDV to CSV solution. If you're deduping data, then you may only have one table. That's fine. This diagram is setup for a typical bank reconciliation.
- In the typical bank reconciliation, as represented by this diagram, you'll need your accounting system's your check register to compare to the trusted data.
- The data in #4 and #5 can be brought into CrushErrors.com. The data is never sent to the cloud. It only lives in the browser or the user's hard drive. It can be dragged from the file system. It can be uploaded with a button. Or, it can be pasted from Excel or Google Sheets.
- This box is where the Crushing happens. The general flow is .... find perfect matches. Send them to the report and get them off your desk. Find less than perfect matches. Notate why they're less than perfect. Send them to the report. Keep the cycle of finding, resolving, notating, until nothing is left. Then save and generate a report. You can save frequently to resume work.
- If you're searching for dupes, there are also fuzzy matches, like #10, but there's no math involved so it's a separate branch. You'll select which fields you want to check dupes on. Unlike other products, you can select more than one field. Not just dupes for Names, but dupes for Names and Zip Codes. So you can find dupe names in same zip.
- The goal with Crush can be to find mismatches. What's a match? A match is when the numbers land where you expect them to land. So, for example, a $100 check, number 532 in the bank, has a corresponding $100 check, number 532, in the accounting system. So a mismatch is when there is any deviation. The amount or the number of the check. There can also be a wire transfer in the bank for $14,000. But the accounting system shows it as two disbursements. One for $9,000 and another for $5,000. That's less than perfect. Another problem could be that the accounting system shows a deposit from Dan's Bar and Grill, but the bank shows a deposit from Dans Bar and Grill. That's less than perfect. Those imperfections can be flagged and notated. Crush uses a starring system. So 5 stars would be a perfect match.
- See #9. A little more information. These are called Fuzzy matches. Every fuzzy match is scored. So a really close match would be 99%. But often a 40% match could be just perfect as well. For example, sometimes there are extra characters in one side of the match, like "0000000XXX Dans Bar and Grill" and "Dan's Bar & Grill". Because of the extra text, the score may be 40%. But they could be perfect matches. Human participation is required to investigate. Crush is a tool to make it easier for people to do their job well. Another issue is when the score could be very high but the two results are NOT a match. For example, "Bo Didley" may score close to "Bob Didley" but they could be two very different people. Again, Crush is a tool, but it needs to be used wisely.
- Combinations are an important breakthrough for Crush. Let's say a vendor sent one 10 bills and your business made 3 payments. The vendor says you owe $1500 but you insist you only owe $100. So you need to figure out how the payments were applied. Maybe your business is missing a vendor bill? Maybe your vendor is missing a payment? This is really hard to resolve even when the numbers are 10 and 3. What if they were 100 and 30? The complexity grows dramatically as the numbers increase. Crush can solve this problem. Even with Crush this can get tricky with big numbers because the number of combinations can be enormous. There are Apps to handle this specific problem, but contact us if you need help.
- In many cases, users will want to compare only one number on the right with one number on the left. For example, maybe the user only needs to compare account numbers in a trial balance. In that case, there should only be one account number on each side. But what if the user wants to compare all of activity for one account number with its corresponding balance. So the activity could have many many rows and the balance only one row. So the summation feature is essential. In this little example, the user will designate the Account Number as a Key Field. The user can manually do this with the Control Box. To get to the control box, you can use the s,c shortcut. It stands for show columns. You can manually navigate by click on the gear box on the upper right. Alternatively, mouse over the big green bar at the left. Then it should be easy to click on the radio button for intersection of KEY and "Account Number". You'll need to do this on both sides. Now you'll need to do same for the intersection of SUM and "Amount" on both sides. Then press the Crush button. Now you'll be able to compare the SUM of the activity with the balance for every account. If there's a difference, you may be able to use the Combination feature to find the difference.
- After the Crush button is pressed the first time, go to the Matches button. The results should be perfect matches. Then press the control button and click on the button that looks like |_|. All of these buttons will turn to P, which stands for Pending. These can all be moved quickly to the final report. To do that, you'll need to see # 14 below.
- To trigger the box represented by A, you can mouse over the green bar on the far right. Or, you can use the p,o keyboard shortcut. That's p, followed by o. It stands for pop out. You would click the number of stars you want to designate for the quality of the match. If perfect, you'll designate 5 stars. Then notate it that it's a perfect match. In the final report you can filter for the number of stars. See # 18 below.
- When you click the Differences button, you will see the rows that do not match perfectly.
- This will is essentially your desktop or your work space. You will try to resolve each mismatch manually with the Crush tools. You will click the |_| button on each side to designate it as Pending. Sometimes 2 items on the left will equal one item on the right. These have to marked pending individually.
- See # 14.
- Users can see the Final Report. This will show all the items that were marked pending. All the notes. All the stars will be there. The report can be exported to Excel. Everything is date stamped. Users can add a header with their own information. Users can filter the report based on the numbers of stars and other parameters.
- All of the activity can be saved. The file has all of the original data, the Apps, and all the stars and notes and configurations. It can be stored just like any spreadsheet. It can be emailed to anyone. It can be updated and saved later. Just like any spreadsheet. A highly proficient user can set everything up and then send to a less proficient user to actually identify the problems. Essentially, the crushing errors task can be delegated to users of different error crushing proficiency, until the task is completed.