Title: Voice of customer report template from calls | Spellit
Description: A copyable voice of customer report template built from sales calls: five blocks, one filled example, and the base every number needs.
Page: https://spellit.ai/blog/playbooks/voice-of-customer-report-template/

---

[Blog](https://spellit.ai/blog/) [Playbooks](https://spellit.ai/blog/playbooks/)

# Voice of customer report template: five blocks built from sales calls

A voice of customer report built from sales calls needs five blocks: a scope line that names the calls, the period and the base; objections ranked by family; the buyer's own wording; requests nobody answered; and one decision per owner. Every number carries the calls behind it, and every claim carries a quote. Copy the template below, fill it from last quarter's recordings and keep it to two pages. [Spellit Client Truth](https://spellit.ai/products/clienttruth/) counts the material for blocks two to four.

Oleg Kulakov CEO, Spellit 3 November 2026 7 min read

On this page

- The whole report fits on two pages when each block has a fixed shape
- The scope line decides whether anyone believes the rest
- Rank objections by family, because the wording does not travel
- Buyer wording belongs in the report as a ranked list of phrases
- An unanswered request is a decision for product, so it gets its own block
- Every finding should end in one owner and one date
- What to do next

A survey-shaped report starts with a score. A call-shaped report starts with a count of what was said and the calls behind it, which changes what the page has to contain. The headline stops being a rating and becomes a sentence the reader can check against a recording. Surveys also reach fewer people than they once did: in 2017 Pew Research Center reported that response rates to telephone surveys had stabilized at around 9%. The rest of this piece is the page itself and the reasoning for each block, so you can keep what fits and change what does not.

## The whole report fits on two pages when each block has a fixed shape

Paste this into a document and fill the brackets. Delete a block only if the calls cannot support it, and say so in the scope line.

### VOICE OF CUSTOMER REPORT: [team or product], [period]

### 1. Scope

[N] calls, [period], [teams or segments]. Each share below is divided by [objections tagged / calls where the topic came up]. Not included: [what was left out and why].

### 2. Objections, by family

Family Mentions Share of [base] Teams where it appears [family name] [n] [%] [x of y]

Top three together: [%]. Biggest change since [previous period]: [family, +/- points].

### 3. Buyer wording

- Phrase heard most: "[buyer phrase]" ([n] calls)
- Phrase that is new this period: "[buyer phrase]" ([n] calls)
- Phrase we use that no buyer used: "[our phrase]"

### 4. Requests without an answer

- [request in the buyer's words]: asked in [n] calls, answered in [n].
- Quote: "[line]" (call [id], [time])

### 5. Decisions

- Marketing: [one change], owner [name], by [date].
- Product: [one change], owner [name], by [date].
- Sales: [one change], owner [name], by [date].

The sections below explain how to fill each block honestly.

## The scope line decides whether anyone believes the rest

The scope line is block one because every later number is a fraction, and a fraction with no stated base can be argued to mean anything. Write the count of calls, the dates and the teams in one sentence. Then name the denominator for each share, because "objections as a share of all calls" and "objections as a share of tagged objections" are different figures that look alike on a slide.

The denominator is also where a report most often misleads without anyone intending it. If one team's tag list has a "no objection raised" option and another team's does not, the share of clean calls measures the form, not the market. Our own [count of objections across 3,091 calls](https://spellit.ai/blog/call-data/sales-objections-data-424-labels/) found exactly that split between teams. Put one line in block one about how each team tags, and the question never reaches the meeting.

Add the exclusions in the same block. A team that was onboarded halfway through the quarter, or a product line with too few calls to count, belongs in "Not included", with the reason. A reader who finds that gap alone will assume you hid it.

How many calls are enough has no fixed answer. Nielsen Norman Group advises continuing until new interviews stop changing the themes, and it cites a 1993 estimate that 20 to 30 interviews uncovered 90 to 95% of customer needs for one product. Calls are not interviews, so take that as a rough scale and not a rule, and state the count next to every table.

## Rank objections by family, because the wording does not travel

Block two ranks objections by what they mean, not by the label a team happened to choose. In the 3,091 calls above, the 4,050 tagged objection mentions fell into 424 different label strings, and 11 of those strings sat inside the price family alone. A report that ranks strings splits one concern into eleven small ones and understates it.

Here is block two filled from that slice: 3,091 calls at eight sales teams in five companies, analyzed in Spellit over the 90 days to October 4, 2026. Families were grouped by hand, and the shares are of 4,050 tagged mentions.

Family Mentions Share of tagged mentions Companies where it appears (of 5) Price and budget, needs installments 805 19.9% 4 Needs to think, not now 627 15.5% 4 Distrust, too good to be true 552 13.6% 4

The three together carry 49%, which is 1,984 of 4,050. The last column does the useful work in a board meeting. A family that shows up at four of five companies is a feature of the market. A family that shows up at one is a feature of that company's offer or its tag list, and the owner of the decision is different. If you want a sharper definition of the second family before you write it, the glossary entry on the [timing objection](https://spellit.ai/blog/glossary/timing-objection/) separates a real delay from a polite no.

Keep the "biggest change" line even when nothing moved. "No family changed by more than two points" is a finding.

## Buyer wording belongs in the report as a ranked list of phrases

Block three is the part marketing will actually use, so it has to contain sentences, not themes. List the phrase buyers used most, the phrase that appeared this period and not before, and one phrase from your own site or script that no buyer said. The third line is uncomfortable and usually the most productive. Spellit Client Truth ranks buyer wording by how often it comes up, and each phrase links to the call it came from, so the list can be checked rather than trusted.

Wording is also what turns a persona from a workshop artifact into something with evidence. If you maintain a [buyer persona](https://spellit.ai/blog/glossary/buyer-persona/), block three is where its claims are tested: each worry written on the persona should have a phrase and a call count in this block, or it comes off the persona.

One filled line shows the standard. Weak: "Customers are price sensitive." Usable: "Price and budget: 805 of 4,050 tagged mentions." Better: the usable line plus one phrase a buyer said, and the call it came from.

## An unanswered request is a decision for product, so it gets its own block

Block four exists because a request left open on a call does not show up in an objection count. The buyer asked for something, the rep said what reps say, and the call moved on. Nobody logs it, since logging it is nobody's job. Spellit Client Truth counts those requests across every call and keeps the sentence each count came from.

Write each one in the buyer's words, then give two numbers: calls where it was asked and calls where it was answered. A request asked often and answered rarely is the item product should see first. One asked often and answered well is a training win, and it belongs to sales. If the count is zero, widen the window before you conclude anything. Rare events show zero in a short window and the zero describes the window, not the market.

## Every finding should end in one owner and one date

Block five is the only one with names in it. Marketing gets one change, product gets one, sales gets one. The limit is deliberate: three owners each holding one dated item will report back at the next meeting, while three owners each holding six will report back on none. If a block above produced more than three findings, the leftover ones stay in the appendix and nobody is assigned to them yet.

Date the decisions in the report itself, then open the next report with a single line on what happened to the previous block five. That one line is what separates a recurring report from a recurring document.

## What to do next

Pick the quarter you already have recorded and fill block one before anything else. If you cannot state the denominator in a single sentence, stop there; the rest will be built on it. Then fill block two from your top three families, and write the "Not included" line before anyone asks for it.

The [voice of customer use case](https://spellit.ai/use-cases/voice-of-customer/) shows what blocks two to four look like when the counts come from your own recordings, and you can [book a demo](https://spellit.ai/request-demo/) to see it on your calls.

Key points

- Block one is a sentence, not a chart: how many calls, which period, which teams, what each share is divided by.
- Objections go in by meaning family. The wording of a family differs from team to team, so ranking labels compares nothing.
- Buyer wording is a ranked list of phrases someone said, with a call attached to each.
- A request the team left open is a product input, so it gets its own block and its own owner.
- The last block holds at most one decision per owner. Five findings without an owner are a reading list.

See the demo now. Then run it on yours.

A live demo of our products and a real conversation about the growth problem you need to solve.

[Book a demo](https://spellit.ai/request-demo/)

See it on your own calls

A live demo and a real conversation about your growth problem.

[Book a demo](https://spellit.ai/request-demo/)

45 minutes · nothing to prepare

## More in Playbooks

[All Playbooks](https://spellit.ai/blog/playbooks/)

- [Playbooks Steering committee notes template for consultants Steering committee notes should fit on one page and hold four things: decisions made, actions with a named owner and a date, decisions still needed wi… 3 November 2026 6 min read](https://spellit.ai/blog/playbooks/steering-committee-notes-template/)
- [Playbooks How to turn lost calls into AI roleplay scenarios You turn a lost call into a roleplay scenario in five steps: find the moment the call turned, group those moments by what the buyer said, write one bu… 3 November 2026 8 min read](https://spellit.ai/blog/playbooks/lost-calls-ai-roleplay-scenarios/)
- [Playbooks A call scoring rubric for insurance renewal calls Score an insurance renewal call in the order it runs: verification, what changed, the premium, the cover, the decision, the close. Keep about a dozen … 3 November 2026 9 min read](https://spellit.ai/blog/playbooks/insurance-renewal-call-scoring-rubric/)
- [Playbooks Do you need conversation intelligence? A decision guide for teams under 20 reps Not yet, if you can still listen to every call yourself, if nobody has an hour on Monday to act on what turns up, or if your calls are not recorded. Y… 3 November 2026 7 min read](https://spellit.ai/blog/playbooks/conversation-intelligence-small-sales-teams/)

## FAQ

### What should a voice of customer report include?

It should include a scope line, a ranking of what customers raised, the customers' exact wording, the requests that went unanswered, and a short list of decisions with owners and dates. The scope line states how many conversations were read, over which period and which base each percentage uses. The ranking groups similar concerns together so the same worry in different words is not split into small pieces. The decisions block is what makes the report operational. Without owners and dates the document gets read and then filed, and the next one repeats the same findings.

### How often should you produce it?

A quarter is a default worth starting from, because most families of objections are frequent enough to show a stable ranking over that window and rare events need the longer span. A monthly cut works for the top three families and the new-phrase line. Weekly views are useful for a sales leader watching a single objection after a script change, but they are too thin for requests, where a week often returns zero. Keep one fixed template across periods so a change in the ranking is a change in the market and not in the layout.

### How many calls do you need before the report is worth sending?

There is no single threshold. Count calls per segment, not in total, because a share computed on a dozen calls moves by several points when one call changes. A practical rule is to state the call count next to every table and flag any segment you would not trust on that count. If a team has too few calls, put it under "Not included" and say so. A smaller report that is honest about its base is more useful to a board than a larger one where every row rests on a different and unstated number of conversations.

### Can you build the report without special software?

Yes, with limits. A person can read a sample of calls, tag them and fill the template in a spreadsheet, and for a first pass that is a reasonable way to learn which blocks matter. The limit is coverage. A hand-read sample covers a fraction of the quarter, and what ends up in the report depends on which calls were read. Spellit Client Truth reads every recorded call and keeps the calls behind each number, so the report is checked against the whole quarter rather than a sample.
