Reality Defender · Marketing operations

Turning sales questions into better marketing

Every week, buyer questions from Gong and Sales shaped what Marketing worked on next.

The problem

A seller needs an answer for a call this afternoon. Somewhere, someone has already answered the question. But which version is current, and can we stand behind the claim? Answering faster only solves half the problem. Marketing also needs to know when the next five sellers are likely to ask the same thing.

Two agents with different jobs

I built the intelligence system around the people using it. We tightened the workflow to two agents, Gabe and Ardy. Gabe was the first stop for Sales. It looked for approved answers and existing materials, and tracked recurring questions and trends. We also pulled all Gong calls into a weekly voice-of-customer review. Together, those inputs let Marketing see what buyers were asking and where our materials fell short.

Ardy supported Marketing. It could find and summarize approved material, draft from it, and route work to the person responsible. People owned prioritization and approval. That distinction mattered for a company making technical claims about detection.

Sales asks
Gabe checks the buyer context, current claims, and material a seller can use.
Marketing listens
Recurring requests and the weekly Gong review show where buyers need a clearer answer.
The team responds
Ardy helps draft from approved sources. A person reviews the claim and owns the update.

What changed

The work continued after a seller got an answer. A recurring question could become a marketing priority and a reusable asset for the next conversation.

From a buyer question to a better first call

Illustrative walkthrough. Imagine a contact center buyer asks, “If the caller is an AI assistant booking an appointment for a customer, why would I block it?” It’s a fair question. A slide that jumps straight from synthetic speech to fraud leaves the seller explaining an exception before they’ve explained the product.

Here’s how I’d use the system. Gabe would check what Sales could already say. In the weekly Gong review, Marketing could look for other calls where buyers made the same distinction. If it kept coming up, I’d change the first-call deck. The new slide would open with two calls: an assistant booking an appointment and a caller asking to move money. Both voices could be synthetic. The request, account context, and company policy determine what should happen next. Ardy could draft the speaker notes from approved sources, and the product owner could check the wording before Sales used it.

Illustrative first-call slide

What is the caller asking to do?

Book an appointment

An assistant calls on a customer’s behalf. What information does the business need to schedule a visit?

Move money

The voice could sound the same. What verification does the business require before it authorizes a transfer?

Detecting synthetic speech gives the operator a signal. The request and the business’s policy determine the next step.

Giving writers a voice they could use

A first-call deck has to sound natural when a seller says the words out loud. An executive byline has to sound like the person whose name is on it. Both can be recognizably Reality Defender and still read very differently.

I built our editorial guide in Notion, starting with the brand archetypes and translating them into voice, tone, and writing guidance by deliverable. I also documented executive voice so a writer could preserve a leader’s point of view and cadence. The same guide informed AI-assisted drafts alongside our approved claims. A writer could change the opening, emphasis, and level of detail for the reader while keeping the underlying product claim accurate.

One idea in three formats. These illustrative lines show how I’d approach the same subject in different settings.

  • Website heading · Identify AI callers while the call is live.
  • First-call deck · When an AI assistant calls on a customer’s behalf, how do you decide what it can do?
  • Executive byline opening · Contact centers used to be able to assume a person was on the other end of the line. Now they need a policy for what happens when that assumption breaks.

Putting the brand to work

Some requests were much smaller than a sales deck. Someone needed an email or a Zoom background. Those were good places to put the brand rules directly into the tool people used to make the thing. I built a ChatGPT maker for email and Zoom backgrounds, worked the design system into Claude, and built the Duo photo grader for the photography workflow.

The design decision was where to spend a person’s attention. A repeatable format is a good candidate for self-service. Deciding what to say to a new buyer, which image tells the story, or whether we can stand behind a claim still deserves someone’s judgment.

Jordan Key’s brand expansion shows the meeting backgrounds, document templates, and webinar graphics our visual system supported. See the design work in Jordan’s portfolio.

Frequently asked questions

How did you use Gong for voice-of-customer research?

We reviewed all Gong calls on a weekly cadence. That gave Marketing a broader view than the individual requests sellers thought to submit and surfaced the language buyers used themselves.

What happened when Gabe could not answer a question?

An uncovered need could become a marketing request with the deal context attached. Marketing assessed priority and ownership. Gabe could then report the status back to Sales.

Who approved AI-assisted marketing claims?

Ardy could retrieve and draft from approved sources. The accountable person reviewed the wording and decided what could go into an asset.

Previous
Previous

UX Writing & Wireframes