We Built Salestrics — Then We Canceled HubSpot and Slack Internally (And Saved $1,200/Mo)

The hypocrisy we had to kill

Selling consolidation while paying HubSpot and Slack is a special kind of founder shame. We were not confused about the vision — we wrote the category. We were slow about applying it to ourselves because migration is annoying and shipping features feels more virtuous than canceling a renewal. The vendor spreadsheet forced the issue.

The breaking point was a Tuesday standup. Our AE said he had already answered a buyer in Gmail and would log HubSpot after lunch. Nobody was lazy. The stack trained us to treat our own CRM as homework. I had told a hundred founders the opposite — work where the record lives — while our internal record lived in three places.

The spreadsheet that started the argument

Our finance tab had a row called GTM software that grew every quarter without anyone owning the whole picture. HubSpot Starter had been fine when it was me and one AE updating stages after calls. Then we added Slack Connect channels for partners, a shared inbox nobody trusted, and Notion pages for security questionnaires that linked out to Drive anyway. The CRM said Proposal sent. Gmail said Waiting on legal. Slack said Did we lose this? Three truths, one customer.

I pulled twelve months of invoices on a Friday afternoon — the kind of task that feels productive and horrible at once. HubSpot with the seats we actually used, not the seats we bought “just in case.” Slack with the apps and Connect add-ons. The number landed around $1,200/month recurring. Not catastrophic. Not ignorable. Especially when you remember neither subscription included the inbox where negotiations actually happened.

What HubSpot still did well for us

Credit where it is due: HubSpot’s pipeline views are familiar, reporting is legible to investors, and plenty of recruiters recognize the name on a resume. For inbound-heavy teams with marketing hubs already paid for, I get the inertia. We were not inbound-heavy. We were eight people trying to close design partners and move upmarket without hiring RevOps to reconcile tabs.

The friction was not “HubSpot is bad.” The friction was work happened elsewhere. Every meaningful thread lived in Gmail. Every revised MSA lived in a folder someone named FINAL_v7_really_final. Slack held the quick questions that should have been on the record. We were paying for a system of record that was often the second place we looked.

What Slack cost us besides money

Slack is brilliant for hallway conversations at internet scale. It is a terrible archive for revenue decisions unless you enjoy archaeology. I cannot count how many times we rebuilt context for a renewal call because the pricing concession lived in a thread with three emoji reactions and no link to the opportunity.

• Search decay: six-month-old pricing approvals buried under product alerts
• Permission fog: guest channels, internal channels, DMs — which one is canonical?
• CRM tax: someone still had to copy outcomes into HubSpot after the fact
• Notification debt: sellers watching pipelines in one app and fires in another

We kept Slack longer than we should have because it felt like oxygen. Removing it internally forced uncomfortable conversations about where decisions should live. That discomfort was the point.

Running GTM on the product we ship

Once we committed, the language shifted internally. We stopped saying "migrate to Salestrics" like we were evaluating a vendor. We said "run deals on production" — Momentum CRM, Salestrics Mail, Workspace, Assistant, the same surfaces we ship. Week one was parallel-running against HubSpot. Week two was new buyer work only inside our own workspace. Week three was export, verify, cancel renewals.

We gave ourselves three weeks and a rule: no heroics on Friday afternoons. The playbook was the one we had been telling founders to use — we finally followed it ourselves. Hypocrisy does not scale when you are the company building the alternative.

1. Inventory open opportunities with next steps dated inside thirty days
2. Map active buyer threads per deal — if it is not on the opportunity, it is not real
3. Move live docs (security pack, order form templates) into Workspace with shared permissions
4. Train on Mail send-as and undo-send — small features prevent big apologies
5. Run one pipeline review entirely inside the workspace before flipping the off switch

The scariest moment was not data import. It was the first Monday when someone asked, “Where do I ping the team about this deal?” The answer became: on the deal. Comments, @mentions, attached docs — not a side channel that evaporates.

Mail on the record changed how we sell

Connecting Salestrics Mail on our own tenant felt incremental until it did not. When a prospect replied to a pricing question, the thread rolled up on the opportunity automatically. No BCC gymnastics. No “can you forward that?” Our AE stopped treating inbox zero as a personal art project and started treating the opportunity timeline as the source of truth — on the product his colleagues build.

I tested undo-send the hard way — drafted an enthusiastic paragraph with the wrong attachment, caught it in the buffer, swapped the file, sent. That sounds trivial until you have sent the wrong security doc to a buyer who is already nervous about vendors. Small safety rails matter when you are moving fast.

Calendar and meetings without tab hopping

We still use external calendars for some personal blocks, but customer meetings tied to accounts now leave breadcrumbs where the account lives. Prep notes in Workspace, meeting outcome as a note on the opportunity, follow-up drafted with Assistant from the same graph. The first time our lead investor asked for a pipeline walk and every stalled deal had a visible last-touch email, I exhaled.

Workspace replaced our Franken-folder situation

Workspace is not glamorous copy — it is where our order forms stopped mutating in mysterious Drive folders. Version history on docs, sheets for lightweight forecasting, slides for the narrative deck we reuse with tweaks. Attachments on mail draw from the same drive quota instead of duplicating blobs across systems.

Legal still redlines in Word sometimes. We are not delusional. But the approved PDF we send buyers lives on the customer record now. When someone asks “which MSA did we send Acme?” there is one answer, not a scavenger hunt.

Assistant earned trust on our own deals

I do not trust AI that hallucinates cheerful nonsense — including AI we ship. Our bar was simple: if Assistant could not cite the thread it used, we did not send the draft. Over a few weeks, our team used it for first-pass follow-ups, meeting summaries, and “what did we promise last time?” reminders — all grounded in mail and notes on the opportunity. We run the same build we ship; when it failed, we filed tickets like any customer would.

The funniest win was mundane: a renewal email that referenced an old discount cap we had actually documented in a note six months prior. The draft surfaced it without someone paging through Slack history. Boring magic is the best magic.

What we paid before we dogfooded

Our internal stack before we moved GTM onto our own product looked roughly like this:

Tool	What we paid for	What was still missing
HubSpot	CRM seats, basic automation	Native mail, docs, grounded AI
Slack	Team chat, Connect, apps	Deal-level record, mail archive
Notion + Drive (partial)	Playbooks, templates	CRM linkage, send workflow

After moving GTM work onto our own workspace, we dropped those two renewals and stopped treating docs as a separate religion. The ~$1,200/month we had been spending on overlap went back into infrastructure and engineering bandwidth. We were funding duplication while building the thing meant to end it.

What broke when we ate our own cooking

Dogfooding is not a marketing asset. It is a discipline tax. Mail threading missed a partner domain once during parallel run — we treated it like a production incident because it was one for us. Attachment routing from Workspace needed a fix before legal stopped side-eyeing internal sends. Export columns investors asked for twice finally landed because we needed them for a board deck, not because a roadmap slide promised them.

When undo-send saved a wrong PDF to a security-conscious buyer, the AE bought lunch for whoever built the buffer. That is the feedback loop you want: builders who feel annoyance personally ship boring safety rails that matter.

What we would do differently

• Migrate fewer open deals at once — twenty active opps, not eighty nostalgic ones
• Name a single record police for two weeks (rotating, not a permanent role)
• Schedule pipeline reviews in the new system earlier — ritual beats features
• Keep a read-only HubSpot export for ninety days; paranoia is free

We also would have written down every “we’ll fix process later” shortcut. Later is where CRMs go to die.

Why we published the internal receipt

Founders smell marketing fluff. I would. We build a Startup Revenue Workspace — we do not get a pass on running HubSpot because migration is inconvenient. Prospects ask whether we use Salestrics. We say yes and open our workspace. That answer is worth more than a trade show booth.

If you are evaluating your own stack, read how teams replace the Frankenstack and what a revenue workspace actually is. This post is the founder receipt for doing internally what we ask the market to do.

We still use specialized tools for engineering and finance. GTM collapse was the scope — not a fantasy of ripping out every SaaS line on the planet. The spreadsheet row we deleted was called GTM software. One primary line now. I look at it monthly instead of avoiding it quarterly.

I spread HubSpot, Slack, and Zapier invoices across the kitchen table while my kid asked why dad looked sick. Totals near twelve hundred dollars monthly for tools that still left Gmail as the court of final appeal. That night I opened a blank doc titled what we actually use and filled it in ten minutes of shame.

Our AE threatened polite mutiny if migration broke his top three deals. We granted veto power for week two. He used it once on a missing partner thread. After we fixed import rules, he became the loudest internal advocate — the signal I trust more than any NPS survey.

A pricing concession lived only in a Slack Connect guest channel named after a customer logo. Six months later nobody remembered which logo. We almost renewed at list price because archaeology lost to inertia.

Undo-send caught a wrong security PDF before it reached a nervous buyer. The AE said nothing in standup and everything in a private thank-you. Small buffers are brand insurance when you move fast.

An investor asked why a proposal-sent deal had no email in thirty days. I opened Gmail on screen share while HubSpot sat colorful and quiet. That afternoon I started the vendor audit for real.

When evaluating revenue stacks, treat forecast silence as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat import hygiene as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat trial sandboxes as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat seat math as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat connector veto as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat tribal knowledge as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat opportunity comments as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.

When evaluating revenue stacks, treat order form versions as a first-class signal — not a footnote on a pricing page. Founders who measure afternoons instead of adjectives catch misfit architecture before autopay renews it.