The Pipedrive deadline that was never Pipedrive's, and the migration that only looks finished

A deadline passed and nothing happened

If you run Pipedrive through an automation platform, you were probably told that 31 July 2026 was the day the old API stopped working. Zapier said so in March, in an article whose title begins "Action required". Make said so on its help centre. Twenty-eight Zapier steps were renamed to carry [DEPRECATING JULY 31 2026] in the name, so you would see it inside your own workflows.

The date came and went. Searching the Make community in early September for anyone whose scenarios stopped after 25 July: nothing. Searching the n8n forum for the same: nothing. As of early September, Zapier had published no article about Pipedrive since May, and the warning inside its editor was still written in the future tense, five weeks after the date it names.

There is a reason for the quiet, and it is not that everybody migrated in time. It is that 31 July was never Pipedrive's date.

Whose deadline was it?

Pipedrive's changelog does not contain the string "31 July 2026" anywhere. What it contains, dated 29 July, is a reminder with a very different tone:

The API v1 endpoints announced for deprecation in one of our previous changelog are now in an out-of-support state.

Although these endpoints may remain functional, Pipedrive no longer provides technical support, maintenance, bug fixes, or service-level guarantees for them. Their continued availability and behavior are not guaranteed, and they may be changed or removed without further notice.

• "Reminder: Deprecated API v1 endpoints are now out of support", Pipedrive developer changelog, 29 July 2026

Effective from 1 August. Pipedrive had named one date, once: the original announcement in April 2025 said the deprecated endpoints would "remain accessible until December 31, 2025", after which their availability and functionality "will no longer be guaranteed". That is an end of guarantee rather than a switch-off, and nothing has replaced it. No shutdown date, then or since.

The 31 July date has a traceable origin, and it is worth following because it explains why two platforms told their users something firmer than the vendor was saying.

In January, Make notified its users that Pipedrive modules would stop working on 31 January — about fifteen days' notice. The forum thread that followed ran to forty-eight posts and over a thousand views. One user, who describes having "hundreds of modules to update", put the question plainly:

One year ago I asked Make when they'd be updating to the V2 Pipedrive API. They said at the time they'd give sufficient notice when this happens. Is 15 days sufficient?

• a user, Make community, 16 January 2026

Three days later, Make apologised — "We understand that the tight timeline has placed pressure on your teams, and for that, we are truly sorry" — and then, four hours after that, announced this:

Update: Pipedrive API v2 Transition Deadline Extended to July 31, 2026

[…] in response to your feedback, we successfully coordinated with Pipedrive to push back the API v1 sunset, and have secured an extension of the grace period for the API v2 migration. Existing Make scenarios using affected API v1 endpoints will remain functional until July 31, 2026 […]

• a Make community team member, Make community, 19 January 2026

That is where the date comes from. It is a grace period one platform negotiated for its own tenants, announced on its own forum. Zapier published the same date seven weeks later without saying where it got it — the two dates matching is almost certainly not a coincidence, but the connection is not something anyone has stated publicly, so treat it as a coincidence with an obvious explanation rather than a proven chain.

What Pipedrive actually deprecated

Two corrections to the version of this story you have probably heard.

It is not the whole of API v1. From the original announcement in April 2025: "We are deprecating specific API v1 endpoints (listed below) that now have direct replacements in API v2 — this does not affect the entire API v1 platform." The list covers Activities, Deals, Persons, Organizations, Products, Pipelines, Stages and the item-search endpoints. Files, Notes, Leads, Users, Filters and — remember this one — Webhooks are not on it.

And "out of support" is a defined term, not a euphemism for off. Pipedrive spells out what it means:

Integrations that continue to use the affected API v1 endpoints may still operate. However:

• Pipedrive does not guarantee their availability, reliability, performance, or response behavior.

• Issues affecting these endpoints may not be investigated or resolved.

• Fixes and improvements will be delivered only for the corresponding API v2 endpoints.

• These endpoints are not covered by Pipedrive's standard service-level commitments.

• Pipedrive developer changelog, 29 July 2026

And, in a box marked Important: "The fact that an out-of-support endpoint remains accessible does not mean it is safe to continue using it. Pipedrive does not guarantee an additional migration period before these endpoints are changed, restricted, or removed."

So: no support, no fixes, no SLA, and the right to remove them whenever, with no notice. What there is not, is a date. Which is precisely the situation that produces no urgency and no action.

Three platforms, three different sentences

Put what each of them told users next to what the vendor said, and the disagreement is not subtle.

The practical consequence is upside down from what you would expect. Anyone who believed their platform migrated on time. Anyone who went to the primary source found "may remain functional", no date, and every reason to leave it until later.

What each platform paid for the same deprecation is a cleaner way to see the difference than what each platform broke.

Make paid in reputation. It was late by its own users' account — "Pipedrive has had API V2 available for years, but Make hasn't updated modules" — gave fifteen days' notice, apologised, and had to go back to the vendor to buy time. Its own list of what it would do with the extension included "identify and prioritize any missing or regressed functionality from the v1 experience", which is an admission that the replacements were not ready when the originals were marked for removal.

Zapier paid in its users' labour. It moved early and did the job properly: every affected step renamed so it is visible inside your Zap, a warning in the editor, an article with per-step replacements, V2 steps ready to go. Twenty-eight step definitions carry that deprecation label, and all twenty-eight are hidden from the picker, so you cannot add a new one — but the ones already in your Zaps are still there. The work of swapping them is yours.

n8n paid nothing at all, which is why its users are the ones still in the dark.

What migration actually costs, counted in steps

The expensive part of this migration is not the endpoint change. It is one line in the v2 migration guide:

Related objects have been removed from API responses to prevent too eager fetching of unnecessary data. Use subsequent API calls to fetch them if you still need them.

• Pipedrive API v2 migration guide

In v1, one call to fetch a deal brought back the person and the organisation attached to it, with their names and details. In v2 it brings back their IDs. Everything else is a second call.

Both platforms hit this, and both describe it the same way from opposite ends. Zapier, in its own guidance:

The V2 API returns less data than the V1 API. Some fields that were previously available in trigger or action outputs are no longer included. […] If your Zap used the New Activity trigger and mapped the deal_title field (which is no longer available), you'll need to: Add a Get Deal search step after the trigger.

• Action required: Update your Pipedrive workflows, Zapier Help Center, published March 2026, last updated May 2026

A Make user, discovering it by hand in January:

my first replacement try is not really a success as I cannot get the details as before in "Watch deals" module: now we only get related person and organisation IDs but the names and all the details we had in collections before in 1 module […] Do you aim at making us multiplying modules?

• a user, Make community, 19 January 2026

The same person's request to Make, in the same post, is the sentence to quote at anyone estimating this job: "we need […] to recover the functionalities we had before and without replacing 1 deprecated Pipedrive module by 3 modules!"

Both platforms bill by run — Zapier by task, Make by operation. So one field that used to arrive free now costs a step, on every run, forever. That is the number to work out before you start: not how many modules you have to rebuild, but how many extra calls the rebuilt version makes every time it fires.

The other breaking changes are ordinary migration work and worth listing so nothing surprises you: offset pagination (start and limit) becomes cursor pagination, PUT becomes PATCH, custom fields move into a separate custom_fields object, timestamps become RFC 3339, field selectors are gone, user_id is renamed owner_id, and deleting via the update method no longer works.

What n8n told its users

Nothing. Not "not much" — nothing.

The old Pipedrive node has https://api.pipedrive.com/v1 written into it in one place, with no switch and no fallback. It calls, among others, /activities, /deals, /deals/search, /organizations, /persons, /products and the nested collections under deals, and it updates everything with PUT. Every one of those is on Pipedrive's deprecated list.

The single message the node displays to anyone who opens it reads, in full:

New node version available: get the latest version with added features from the nodes panel.

Not "this uses a deprecated API". Not a date, not a link. Added features. The node's documentation page contains no mention of deprecation, and no mention of v1 or v2 either.

In January, a user did the work n8n had not. They opened the source, found the hardcoded base URL, and asked three specific questions on the forum: is the team aware, has anyone spoken to Pipedrive about a timeline, and is a migration planned. The only reply came from another user — a well-regarded one, but not staff:

The n8n team should be aware of the Pipedrive API v1 deprecation. I believe i saw it elsewhere on github or something someone discussing it. The team is likely working on plans to migrate the Pipedrive node to API v2 to avoid any disruptions. … (Not Positive tho)

• a forum member replying on community.n8n.io, January 2026

The thread got 139 views and auto-closed three months later. A v2 node did arrive, eleven weeks after that question — built by an outside contributor rather than by n8n, and released in April. n8n never answered the thread.

The v2 node still calls v1

Here is the part that matters more than any date in this article, because it is the one thing you cannot see from the interface.

Switching to the Pipedrive node v2 does not take you off the deprecated endpoints. It takes you off most of them. Four operations — searching deals, persons, organizations and products — still call the v1 search endpoints, in both their paginated and non-paginated paths. Three of the four carry a comment explaining why, of the form:

// Search uses v1 API — v2 search endpoint is not available for deals

As of the master branch in early September, that comment is not correct. Pipedrive's own OpenAPI specification for v2 contains /deals/search, /persons/search, /organizations/search, /products/search, /itemSearch and more. The deprecation changelog lists them as the direct replacements, one for one — GET /v1/deals/search becomes GET /api/v2/deals/search — and has done since April 2025, nearly a year before the v2 node was written.

The rest of the node is careful work: everything else uses the new cursor pagination properly, and the resources Pipedrive did not deprecate — files, notes, leads — are correctly left on v1, which is where they belong.

The contributor who built it was explicit in the pull request that search was staying on v1, and about how the node was built: "Based on spec and co-authored with Claude Opus 4.6". Over the three months after release, three separate fixes landed against the new node — a date format, a user_id that should have been owner_id, a person that needed to be sent as a participant. Two of those, the timestamp format and the owner field rename, are items Pipedrive had listed in its migration guide in advance.

There was no open issue about the search endpoints when this was written. There is no indication in the interface. A workflow that shows Pipedrive node v2 in the panel, with a search step in it, is calling an out-of-support endpoint every time it runs, and the person who migrated it believes the job is done.

Why this costs money and not just principle

The token budget is what turns "no support" into a number.

Pipedrive rate-limits by tokens rather than by request count. The daily budget is "30,000 base tokens × subscription plan multiplier × number of seats […]", reset at midnight in the server's timezone, which Pipedrive notes "may not be aligned with timezone of customer location". And in its own words, the v2 endpoints are "performance-optimized, resulting in lower token costs compared to the original v1 endpoints" — the original announcement put the difference at 50 per cent.

So a search left on v1 costs roughly twice as many tokens as the same search on v2. Not twice the money; twice the share of a fixed daily allowance that, when exhausted, stops every integration on the account until midnight somewhere else. A workflow that has been comfortably inside the budget for a year can cross it because the deal list grew and the same nightly sync now walks twice as many records — and the thing that breaks is not the search. It is everything.

The trigger is fine, which is the opposite of what you would guess

The Pipedrive Trigger node also calls api.pipedrive.com/v1, so it looks like the biggest bomb in the workflow. It is not. The only endpoint it touches is /webhooks, and webhooks are not on Pipedrive's deprecated list. The contributor who built the v2 node said the same in the pull request, explaining why the trigger was out of scope: "v1 webhooks not deprecated".

Worth knowing before you spend an afternoon replacing something that was never at risk.

Checking whether your migration actually happened

The useful question here is not "have I migrated" — everybody believes they have. It is "what is my account still calling", and there is exactly one place that answers it honestly.

1. Open the API Usage Dashboard in your Pipedrive company settings. Pipedrive points at it directly: use it "to identify whether your organization is still calling deprecated endpoints". It reports what actually happened, which makes it immune to every assumption in this article, including mine.
2. If you use search steps in n8n, assume they are still on v1 regardless of which node version the panel shows, until the dashboard tells you otherwise. This is the one that will not show up anywhere else.
3. In Zapier, look for [DEPRECATING JULY 31 2026] in your step names. They were renamed precisely so this search would work. The definitions were still live in the app in early September and hidden from the picker, so you cannot add new ones — whether the ones already in your Zaps still execute is not something anyone has demonstrated publicly, so test one rather than assume either way.
4. In Make, check whether your scenarios use the modules marked deprecated, and note that Leads and Projects legitimately remain on v1 because Pipedrive never deprecated them.
5. Count the extra calls before you rebuild anything. Every field that used to arrive attached to a deal now needs its own step. On per-operation billing that is a recurring cost, and it is the number that decides whether this is an afternoon or a project.
6. Do not wait for an outage to prioritise this. There is no scheduled one. The endpoints stop when they stop, without notice, and the vendor has said as much in writing.

When this is a job to hand over

Every other deprecation piece ends with a countdown. This one cannot, and that changes what you are actually deciding.

Nothing is on fire. Nothing is scheduled to catch fire. What you have is an integration running on endpoints the vendor has stopped maintaining, at double the token cost, with a stated right to be switched off without notice — and no calendar entry anywhere that will ever remind you. Deprecations like this do not get fixed because they became urgent. They get fixed because somebody decided to, or they get fixed at two in the morning.

The size of the job is set by one number, and it is not how many workflows you have. It is how many fields your automations read that v2 no longer returns in the same call. Ten workflows that only create deals are close to free. One workflow that reads a person's name, the organisation's address and the owner's email off a single deal lookup is three extra calls per run and a mapping rebuild, and it will cost more in platform operations afterwards than it does to fix.

That count is knowable from your own scenarios and the API usage dashboard, before anyone opens a workflow — which is how the price gets settled up front. A good deal of the integration work we take arrives in this state: not broken, just quietly running on something the vendor has walked away from.

The thing worth keeping from all of this is smaller than the deprecation and more useful. Three platforms, one vendor change, and the only shutdown date anybody could name belonged to a private arrangement between two companies. Your automation platform is not a neutral window onto the API it wraps; it is a party with its own timetable, its own negotiating position, and its own reasons to tell you something firmer or vaguer than the truth. When it says the connector changed under you, that is worth checking against the vendor. When it says nothing at all, that is worth checking too.

Sources

• Pipedrive developer changelog - read 8 September 2026. The deprecation announcement of 14 April 2025 stating that this does not affect the entire v1 platform and listing the endpoint groups with their v2 replacements one for one, and the reminder of 29 July 2026 with the out-of-support definition, the "may remain functional" wording, the four bullets on what out of support means, and the note that no additional migration period is guaranteed. Effective dates of 1 January 2026 and 1 August 2026. No shutdown date appears anywhere, and the string 31 July 2026 does not appear at all.
• Pipedrive API v2 migration guide - read 8 September 2026 through a text proxy, the page being behind a bot check. Cursor pagination replacing offset pagination, PUT becoming PATCH, custom fields moving to their own object, related objects removed from responses, RFC 3339 timestamps, removal of field selectors, and the rename of user_id to owner_id.
• Pipedrive API v2 OpenAPI specification - read 8 September 2026. The presence of /deals/search, /persons/search, /organizations/search, /products/search, /itemSearch and related paths in v2, which is what establishes that the comment in the n8n source is incorrect.
• Pipedrive rate limiting documentation - read 8 September 2026 through a text proxy. The token budget formula, the midnight reset in the server's timezone, and the statement that v2 endpoints cost fewer tokens than v1.
• Action required: Update your Pipedrive workflows before the V1 API deprecation - Zapier Help Center, created 10 March 2026, last updated 29 May 2026, read 8 September 2026 through the help centre API. The 31 July date, the claim that all V1 endpoints are affected, the renaming convention, the warning that V2 returns less data, and the worked example of adding a Get Deal step to recover a lost field.
• Pipedrive integrations on Zapier - read 8 September 2026. Twenty-eight step definitions carrying the deprecation label, all of them hidden from the picker, and the in-editor warning still written in the future tense.
• Pipedrive API Deprecations thread - Make community, January to April 2026, read 8 September 2026. The fifteen days' notice before the 31 January cutoff, the user accounts of hundreds of modules and of one module becoming three, the two apologies from Make's community team, and the announcement of the extension to 31 July 2026 negotiated with Pipedrive.
• Pipedrive API v1 to v2 transition - Make help centre, created 19 January 2026, last updated 10 March 2026, read 8 September 2026. The statement that scenarios will run until 31 July 2026 and that new scenarios cannot use the deprecated modules. Not updated since March.
• Pipedrive app documentation - Make, read 8 September 2026. The note that Leads and Projects legitimately remain on v1.
• n8n source, Pipedrive node - read 8 September 2026 from the raw files on the master branch. The hardcoded v1 base URL in the version 1 node with no fallback, the endpoints it calls, the old-version notice text in full, the version switch in the v2 transport layer, the four search operations pinned to v1 in both their paginated and non-paginated paths together with the comments explaining why, the correct cursor pagination used elsewhere, and the trigger node calling only the webhooks endpoint.
• n8n pull request #27905 - read 8 September 2026 through the GitHub API. Merged 6 April 2026 and released the following day in version 2.16.0. The author is an outside contributor rather than n8n staff. The pull request states that search operations remain on v1, that the trigger was out of scope because v1 webhooks are not deprecated, and that the work was based on the specification and co-authored with an AI model.
• n8n Pipedrive node documentation - read 8 September 2026. No mention of the deprecation, or of v1 or v2 at all.
• Pipedrive node API v1 deprecation - migration plans? - n8n community, 19 January 2026, read 8 September 2026. A user's three questions after finding the hardcoded base URL himself, and the single reply, from another community member rather than from n8n, saying the team is likely aware.
• n8n issues and pull requests mentioning Pipedrive - read 8 September 2026 through the GitHub API. Four fixes to the v2 node in the three months after release, covering date formats, the owner field rename and participant handling, all of them items listed in Pipedrive's migration guide beforehand; and no open issue about the search operations remaining on v1.
• community.make.com and community.n8n.io - searched 8 September 2026 for reports of breakage after 25 July 2026. None found on either platform.