Pipedrive and Gmail: which deals are waiting on a reply from your side

It is 9:40 on a Monday. The sales manager has the Pipedrive pipeline open, filtered to Proposal sent and Negotiation, sorted by value. Twenty-three deals. Two of the four reps are on the call already, one is joining from a car.

She goes down the list and asks the same question about each one: where is this, who has it. And for maybe a third of the deals, the honest answer from the rep is a version of "I think they came back on something, let me check." Then there is a pause while someone searches their own inbox on a phone, scrolls a thread, and reads out a sentence from six days ago: "Can you confirm whether the SSO line is included before we send this to legal?"

Nobody answered it. It was not deliberate. The message arrived on a Thursday afternoon in a thread with nine other people on it, the rep read it on the way out, and the thread sank.

That is the question the pipeline review actually wants answered, and neither screen answers it: which open deals have a message from the customer that our side has not responded to.

What Pipedrive shows on its own

Pipedrive is good at the record half of this. Connect Pipedrive and Gmail with two-way email sync and the threads land on the deal, attached to the right person, visible in the deal detail view without anyone forwarding anything. Pipedrive email tracking, on the plans that include it, tells you when a message was opened and when a link in it was clicked. Activities give you a scheduled next step. Rotting settings flag deals that have sat in a stage past the idle limit you set. Filters can find every deal with no activity scheduled, every deal over a value threshold, every deal owned by one rep.

If email sync is on, the deal even carries date fields for the last email sent and the last email received. So the raw ingredients are there.

Where it stops is specific and worth naming precisely. Pipedrive filters compare a field against a value or against a relative date range. There is no condition that reads "last email received is later than last email sent." You cannot compare two fields to each other, which is exactly the comparison this question needs.

And even if you could, that comparison would be a weak proxy. The last inbound message is often an out of office, a calendar accept, a "great, thanks" that closes the loop rather than opening one, or a bounce from a distribution list. The last outbound message is often a nudge that did not address what was asked three messages earlier. Direction is a fact about metadata. "Waiting on us" is a fact about content.

Tracking has the same ceiling. Knowing the buyer opened your proposal twice is genuinely useful, and it is not the same as knowing they replied with a question that is still sitting there.

What Gmail shows on its own

Gmail has the other half, in full fidelity. The sentence is right there, in the words the buyer used, with the timestamps and the participants and the quoted history underneath. Search operators get you a long way inside one mailbox: from:, to:, newer_than:7d, in:anywhere. Gmail's nudges will even push a thread back to the top of the inbox with "Received 4 days ago. Reply?"

Three things stop it from answering the pipeline question.

Gmail is organised by person and by thread. It has no concept of a deal, a stage, a value or a close date. It cannot tell the difference between an unanswered message on a 90,000 renewal and an unanswered message from a vendor selling office chairs, and it cannot sort by which one costs more to ignore.

Gmail nudges live in the rep's own inbox and are ranked by Gmail's guess. They are a personal prompt, not a team view. So is the unread flag, which is unreliable for a different reason: reps mark things read while triaging on a phone. Read is not answered.

And the manager cannot see any of it. A manager cannot search across the team's mailboxes without delegation or Google Vault, and Vault is a compliance and export tool, not something you open during a Monday review.

Why the gap is structural, not a missing feature

Half of the evidence for this question is a record. The other half is a sentence somebody wrote.

The record half has a schema. Stage is an enum, value is a number, owner is a foreign key, close date is a date. Records can be filtered, grouped, counted and charted, which is what Pipedrive's reporting layer does.

The sentence half has no schema. "Can you confirm whether the SSO line is included before we send this to legal" is not a field. Whether it counts as an open ask depends on what came before it, who said it, and whether the reply two days later actually addressed it or talked about scheduling. Answering that requires reading, not matching. Gmail does not attempt it, because Gmail is a mail client with a search box, not a reporting layer.

So this is not a report Pipedrive forgot to ship, and not a filter Gmail is missing. It sits between two products that are each doing their own job correctly.

It also sits outside the two obvious categories of tool you would reach for next, and both of those are good products:

An automation platform such as Zapier or Make will connect Pipedrive and Gmail without complaint. Their shape is trigger based: something happens, then steps run. That is the right shape for moving a record forward as events arrive. It is the wrong shape for a question about a set of things that already happened and are now sitting still, because there was no trigger at the moment nobody replied.

A BI tool such as Power BI Copilot or Looker Conversational Analytics will model the Pipedrive half beautifully. But BI connects to databases and modelled sources. A sentence in a rep's inbox is not a modelled source and is not in the warehouse. It is not that the question is hard to express there. The evidence is outside what those tools can see at all.

What the join actually requires

Two things: a key, and a judgement call.

The key is the counterparty identity, and one key is not enough on its own.

In practice you match on contact email first, fall back to domain, and treat the domain match as a candidate that needs the content check rather than an answer.

The judgement call is the part no matching rule covers. Given the last inbound message on a thread, is there an open ask directed at us? An out of office is not. A calendar accept is not. "Sounds good, speak Tuesday" is not. "Can you confirm the SSO line" is. And if we did send something after that message, did it address the ask, or was it an unrelated check-in? That determination is reading comprehension over a paragraph of English, applied a few hundred times.

Answering it with Skopx

Skopx connects to nearly 1,000 SaaS tools plus your databases, and answers questions across them in chat with citations back to what it read. So the join above is one sentence:

Across our open Pipedrive deals in Proposal sent and Negotiation, find the most recent Gmail thread with each deal's contacts, and tell me which deals have an inbound message nobody on our side has answered. Quote what they asked for and how long it has been sitting.

What comes back is a list, ranked by deal value, where each row names the deal, the person, the number of days, and the actual sentence. Every row cites the Pipedrive deal it pulled and the Gmail message it read, so the rep who disagrees can open the thread and check in two seconds rather than argue about it.

The Monday review does not want that as a one-off. It wants it standing. Describing the same thing as a console gives you one: Internal Apps builds a console from a sentence typed in chat, so "make that a page I can open before every pipeline review" produces a table with deal, value, stage, owner, days waiting, and the quoted ask, sorted so the expensive silences are at the top.

The console reads, and it can act, once, when a person clicks. The one action worth having on a row here is logging a follow-up activity on the deal in Pipedrive, and clicking it shows exactly what will be written and asks for confirmation first.

Honest limits

It is not a monitor. Nothing runs overnight, nothing pings Slack at 8am, nothing watches for new unanswered messages. Someone opens the console and it reads fresh. If nobody opens it, nothing happens.

It stores nothing. There is no copy of your email sitting in a second database, no forms, no records created behind your back, no public link to hand to someone outside the company.

It sees what the connected mailboxes expose. Threads in a mailbox nobody connected, or in someone's personal account, are invisible to it, and so is anything that happened on WhatsApp or over the phone. Deals with no email trail simply do not appear on the list.

The judgement call is a model reading a sentence, so it will occasionally treat an out of office as an ask, or miss a request buried in quoted text three replies down. That is why every row carries the quote and the link. Use it as a shortlist for a human, not a verdict.

And it does not replace the Pipedrive Gmail integration. Keep the sync on. The more of the thread history that is already attached to the deal, the more reliable the join.

Team is $16 per seat per month with 2.3 million AI tokens included. Solo is $5.