Small-business outsourcing research

How often do price-increase questions reach an untrained assistant?

A protocol for measuring whether price-increase questions arrive before the support team has an approved explanation and a clear escalation path.

Research question and scope

This protocol asks how often price-increase questions reach an assistant who has not been trained on the approved explanation. The unit is one customer conversation about a price increase, identified by the customer, the date, and the channel. The study measures whether the assistant had the approved explanation, whether the answer matched it, and whether the conversation was escalated. It does not evaluate whether the price increase was fair.

Set the scope to one small business, one support channel, and one price-change announcement. Record the announcement date, the training date, and the escalation rules at the start. If the training changes during the window, keep the earlier definition for the full sample or analyze the change as a separate period. A new script can look like better readiness when it is only a new document.

Population and unit of analysis

Define the population as every customer conversation that mentioned the price increase during the window. The unit is the conversation, not the message, because one conversation can span several messages. A conversation that is reopened after a customer reply remains one unit with an additional interval. A separate conversation about a different product is a new unit.

Record the channel, the customer tenure, the plan, and whether the customer asked for an exception. These strata matter because a long-term customer and a new customer may receive different treatment. Include conversations that were resolved without escalation, because the base rate is part of the finding. Test the unit definition on two ordinary conversations before collection begins.

Definitions that make the measure reproducible

Define the approved explanation as the written message the owner authorized before the price change. Define a trained assistant as one who completed the training and has access to the current explanation. Define a matching answer as one that states the same effective date, amount, and reason without adding an unauthorized offer. Define an untrained conversation as one handled before the training date or without access to the explanation.

Do not count a link to a generic pricing page as the approved explanation. Do not count an answer that gives the right amount but invents a reason. A conversation that was escalated before an answer still counts as a price-increase conversation. Each definition should include one positive example and one boundary example.

A price increase can be announced to customers before the support team is ready, especially when sales sends the message. Record the announcement channel and the first support question. A gap between the two is a readiness problem that the owner can fix with sequencing. The protocol should show it clearly.

Evidence standard and data dictionary

For each conversation, capture the customer, the channel, the date, the plan, and whether the price increase was mentioned. Add whether the assistant had completed training, whether the answer matched the explanation, and whether the conversation was escalated. Record the customer's response and the outcome. Keep the packet with the conversation record.

Also capture the time between the price announcement and the first customer question. That gap shows whether the team had time to prepare. If the answer was given verbally, record the note and the assistant. A second reviewer should be able to reconstruct the conversation from the record.

A second view is the share of questions that arrived before the training date. That number shows whether the announcement and the training are sequenced correctly. Record the announcement date, the training date, and the first question. The owner can then decide whether the training needs to start earlier.

Sampling and collection

Draw the sample from the support system rather than from a list of complaints. Record the export date, the filters, the row count, and the selection rule. Include conversations from every channel and every plan so the result is not dominated by one group. If the population is ordered by date, a fixed interval spreads the sample across the window.

Collect the conversation as it existed at the time of the answer, not as it looks after later edits. If the customer followed up after the training, record the dates and analyze the first answer separately. Do not infer a missing explanation from the eventual resolution. Note the collection date separately from the conversation date.

The review should also separate a factual question from a request for an exception. The first can be answered from the approved explanation, while the second needs the owner. Show the counts separately and describe the pattern. A single escalation rate hides the difference between a knowledge gap and an authority gap.

Analysis and reporting

Report the number of conversations, the share handled by a trained assistant, the share with a matching answer, and the share escalated. Show the counts by channel and plan. Present the median time from the price announcement to the first question. Use the same denominator for every rate.

Inspect the untrained conversations and describe the pattern without naming the assistant. A cluster before the training date may point to a scheduling problem, while a cluster in one channel may point to a missing reference. Compare the result with the customer tenure, because a long-term customer may ask a more complex question. Report the exception requests separately.

A useful additional view is the share of answers that added an unauthorized offer. That is a boundary failure even when the customer was happy. Show the offer type and describe the consequence. That view helps the owner decide whether the training or the authority limit needs the change.

A useful check is whether the assistant had access to the current explanation at the time of the answer. A trained assistant without the current document is still unprepared. Record the document version and the answer date. That evidence helps the owner decide whether the problem is training or publishing.

Limitations and sources of error

The record cannot show whether the assistant understood the explanation or whether the customer accepted it. It can show only what was written and when. A matching answer is not proof of a good conversation, and an untrained answer is not proof of a bad one. The protocol also cannot capture a question asked in a meeting or a phone call that left no note.

Support systems can omit calls, direct messages, and conversations handled by the owner. A price-increase question that was answered outside the system may never appear in the sample. Record those cases and their treatment instead of dropping them silently. If the assistant also wrote the approved explanation, note the independence limitation and analyze that subgroup separately.

The final view is the share of answers that added an unauthorized offer or promise. That is a boundary failure even when the customer accepted it. Show the offer type and describe the consequence. That view helps the owner decide whether the authority limit or the training needs the change.

Review cadence and next action

Give the result to the owner and the support lead and ask them to choose one observable change. The change might be a training date before the announcement, a required reference link, or a rule that exception requests go to the owner. Record the owner, the effective date, and the measure that will show whether the change worked. Avoid changing the training and the offer at the same time.

Repeat the protocol with the same definitions after the change and compare the same fields. If matching improved but escalations rose, the assistant may be routing too many routine questions. If escalations fell but unauthorized offers rose, the boundary may be too loose. Close the review when the owner accepts the finding and the next sample is scheduled.

Keep the conversation and the approved explanation together so the next review can compare them. A finding without the actual answer is hard to verify. Record the reviewer and the date. The packet should let a new assistant learn from the case. Record the review date with the sample.

Sources

Frequently asked questions

Does a favorable sample prove the routine caused the result?

No. The design is observational and supports a bounded operating decision, not a universal causal claim.

What should an owner review first?

Start with high-consequence defects, missing denominators, exceptions, and cases that required the owner to reconstruct the evidence.

Related research