On Thursday, November 30, 2023, Md. Habibur Rahman, Senior Secretary of the Ministry of Power, Energy and Mineral Resources, presided over a presentation on the final report of the "Customer Satisfaction Survey with Recommendation of Quality Power Supply including SAIDI/SAIFI" study, implemented by. How to ask the right first hundred days questions in the right order
Photo by Press Information Department on Wikimedia Commons, Public domain

Maintenance

Part of Why your own first hundred days rarely go in a straight line

How to ask the right first hundred days questions in the right order

First 100 days questions grouped by who you ask: the customer who stayed, the one who left, the record, and yourself, with how to score each answer.

Records Desk · · 7 min read

Questions in the first hundred days are cheap and answers are not. The difficulty is that the four groups you can ask know completely different things, and most teams ask all four the same question and average the results.

So these are sorted by who is being asked. Each comes with what a useful answer looks like and what a worthless one looks like, because at this stage a bad answer is more dangerous than no answer.

What to take away

  • The people who stopped hold the information about your ceiling. They are available for about a week.
  • Every question here asks about what happened. None asks what someone would do, because a prediction from a person with no stake is not evidence.
  • Score answers against your written expectation, not against how the conversation felt.

Ask the customer who stayed

They are the friendliest group and the easiest to learn nothing from. Keep the questions about behavior.

Five questions for staying customers

  • What were you doing before?
  • What nearly stopped you paying?
  • When last used, and what for?
  • Who else has seen this?
  • What would you stop doing?
What were you doing about this before?
A useful answer names a method, a tool, a person, or an hour of the week. A worthless answer is "nothing" without detail, which usually means the problem was not being acted on and therefore was not urgent.
What nearly stopped you from paying?
Useful answers name a specific step, a specific doubt, or a specific comparison. If the honest answer is nothing, you may be talking to someone who bought out of goodwill, and it is worth asking how they came to you.
When was the last time you used it, and what for?
The date and the task, in their words. A vague answer about general usefulness usually means it has not been used recently.
Who else in your organization has seen this?
The answer tells you whether the thing has spread past one person, which is the difference between a customer and an account.
What would you have to stop doing if you kept paying for this?
This is the honest version of asking about value. It puts the cost against something real in their world.

Ask the customer who left

The hardest group to contact and the only one that can explain the limit. Ask within a week, ask once, and make it short.

Exit interview in three questions

  1. Ask within a week, once, short
  2. What did you hope it would do?
  3. Back to old method or something else?
  4. Was there a moment you decided?
  5. Make canceling easy
What were you hoping this would do that it did not do?
One question, open, no options. Options collapse three findings into one label.
Did you go back to what you were doing before, or find something else?
Back to the old method means the problem tolerates the old method. Something else means you lost on comparison and you should find out to what.
Was there a moment you decided?
Often there is, and it is nearly always a specific event: a bill, a delivery that arrived differently than expected, a colleague's question.

Canceling has to be easy for any of this to happen. Where billing recurs, that is also what the rule on negative option offers is about, and a person who gives up and disputes the charge answers nothing.

Score these by whether the reasons cluster. Three people saying the same unexpected thing is worth investigating. Three people saying three different things means you do not have a pattern yet, and saying so is a legitimate result. Separating a price reason from a value reason takes one follow-up, and the method is in the pricing questions.

Ask the record

The question log, the arrival record, and the cancellation file will answer questions nobody has asked them.

QuestionWhat you are looking forThe trap
What is the largest group of questions?The place the product or page is unclearAnswering the loudest question instead
Is the question rate per customer falling?Comprehension improving over timeReading total questions, which rise with customers
Which route produced customers who stayed?A repeatable channel worth more effortRanking routes by volume rather than by who remained
Do cohort one and cohort three have the same shape?Whether anything actually changedReading the whole population, which the newest cohort dominates
How many leavers were asked why?Your own discipline, which you controlReporting how many answered instead

The last row is the only figure on this page you fully control, and it is the best single indicator of whether the hundred days will produce anything. The counting rules underneath these are in the market metrics.

Ask yourself

Written answers, dated, in the same file as the weekly notes.

Four written questions for yourself

  • What did I expect, what happened?
  • What am I avoiding?
  • What would make me stop?
  • Which claims can I support?
What did I expect this week, and what happened?
If there was no expectation, the week produced activity rather than evidence.
What am I avoiding?
In this period it is nearly always the same two things: talking to people who left, and finding a route to strangers. Both are unpleasant and both are where the answers are.
What would I have to see to stop?
Written before you need it, in counts and with an action attached. A stopping condition invented after a bad month is a rationalization.
Which of my claims can I still support?
Pages accumulate claims fast in the first weeks. The standard is that a reasonable basis exists before a claim runs, and the review takes ten minutes with the page open beside the list.
Am I fixing or flinching?
A fix follows a grouped pattern. A flinch follows the most recent conversation. The difference is visible in whether the change was on a list before someone complained.

Questions that sound rigorous and are not

Would you recommend this to a colleague? A prediction, given free, by someone with no stake. Ask instead whether they already have, and to whom.

Rigorous-sounding questions to replace

Don't ask

Recommendation
Would you recommend?
Renewal
Likely to keep paying?
Features
What features do you want?
Price
Is the price fair?
Yourself
How is it going?

Ask instead

Recommendation
Have you, and to whom?
Renewal
The renewal date
Features
What did you do last time?
Price
A real decision with a number
Yourself
The four written questions
How likely are you to keep paying?
Same problem. The renewal date answers it for free, and honestly.
What features would you like?
Produces a list of things nobody will pay extra for. Ask what they did last time the problem occurred instead.
Is the price fair?
Everyone says a lower price is fairer. The only price question with an answer is one attached to a real decision, which is why pricing validation puts a number in front of a person who can refuse it.
How is it going?
Asked of yourself, it is the question that produces a hundred days of narration. Replace it with the four written ones above, and with the weekly discipline set out in the first hundred days.

Common questions

How many people should I ask?

Everyone who left, and five or six of those who stayed, per cohort. The completeness matters more for leavers because they are the smaller and more informative group.

By email or on a call?

Leavers by email, one question, because a call is a large ask from someone who just stopped paying. Current customers on a call, because follow-ups are where the useful part lives.

What if nobody replies?

Record the ask rate anyway. A file showing thirty asked and four answered is still a record, and the four are still the best evidence you have.

Can I use a survey?

For the record questions, yes. For the leaver question, no: a form with options gets you the option list back, and the wording of the options will do most of the work. Open text, short, one question.

Should I offer something for answering?

It changes what people say and, if you then quote them, it is a connection that has to be disclosed. Better to ask smaller and accept fewer answers.

Filed underfirst 100 days questions

More in Maintenance

Latest from Records Desk