What we solve ·What is unaccounted for, and what am I missing?
Vanguard · Guide
What are your customers telling you that nobody has read?
We have years of tickets and complaints. There is something in there and nobody has had time to look.
Over the text your organization has already accumulated, quantified and verified findings that nobody has at hand today.
Hidden value in your textSend this page to whoever decides
This sounds like you if
- Management asked a question your reports do not answer.
- A problem keeps repeating and its cause is buried in text that has never been processed.
- The free text in tickets, surveys and complaints does not make it into any report.
Use this today, without hiring anyone
How to verify a text finding before believing it. It is the part that separates data from an impression, and it gets done by hand with what you already have.
- Pull thirty random cases from the finding and read them. If the pattern is not visible to the eye in thirty, it is not in the ten thousand. It is the cheapest filter there is.
- Count the false positives among those thirty and write the number down. A finding without that number is not a finding, it is a headline.
- Ask what changed in the business over that period. A spike in complaints that coincides with a change of process, vendor or campaign is not explained by the text, it is explained by the calendar.
- Compare against the previous period using the same definition. What looks like a trend is usually a change in how things get classified.
- Put money on it. How many times a month, at what cost each time. A finding without a figure does not compete for budget and will die in the presentation.
And the three questions worth writing down before looking at the corpus, because afterwards you find what you want: what specific decision would change with the answer, who would make it and what they would do if the result is that there is nothing there.
How we solve it
The method, not the promise.
- The questions management wants answered get written down, before anything is looked at. Without that, you find what you want to find.
- The corpus gets chosen for what can answer those questions, not for what is easy to export.
- It gets explored by topic, frequency, concentration and correlation with what happened in the business.
- Every finding gets verified by manual sampling and its false positives get counted. Without that step it does not leave the report.
- A figure gets attached: how many times, how much it costs, where it concentrates. And a recommendation of what to build and what to leave alone.
What you receive
- Three to five quantified findings, each with its manual verification and its false positives counted.
- The samples that support each finding, so you can check it yourself.
- The figure for what each one is worth: frequency times cost.
- The recommendation of what to build, what to measure continuously and what to leave alone.
The proof that applies here
- 250,000 incidents analyzed in a real operation, not in an exercise.
- Six Sigma Black Belt: statistical control applied to service processes.
- 19 years inside service operations, which is what makes it possible to recognize which finding matters and which is noise.
Before you hire
A negative report is also a result: if no relevant finding appears, you receive the evidence of having searched properly. That is agreed in writing before signing.
Every finding comes with the samples that support it and with its false positives counted, so you can verify it on your own. A finding that cannot be audited does not get delivered.
What specific decision would change if you knew what is in that text, and who would make it?
If the problem is a different one