The Harvest price increase that has been landing on renewals through 2026 is not a percentage on the sticker price. Bending Spoons, the Milan company that closed its acquisition of Harvest's parent Iridesco LLC on 31 July 2025, has changed how the bill is calculated. Some accounts now pay for things that were never counted before, and the counting happens in a place customers cannot see until the invoice lands.
That distinction matters, because it changes what you can do about it. A rate rise you can budget for. A change in what gets measured, with the measurements unpublished, you cannot.
This piece is an attempt to describe the mechanics accurately. Almost every article currently ranking for this topic contains the same category error, and it is not a small one: they describe Flex and Unlimited as Harvest's plans. They are not plans. Getting that backwards leads to advice that is exactly wrong for the customers most affected, so it is worth spending a section on before anything else.
Harvest Flex vs Unlimited: billing modes, not tiers
Harvest's pricing page lists four tiers: Free, Teams, Enterprise and Enterprise Plus. Flex and Unlimited appear nowhere on it. They live in the help centre, where they are described as the two ways a plan can be billed.
So the two choices are stacked, not parallel. You are on a tier, and that tier is billed in one of two modes. "Teams versus Enterprise" is a question about features. "Flex versus Unlimited" is a question about how your usage is converted into money.
The second consequence is the one that trips people up. Unlimited is the fixed-price mode. The metering, the allowances, the overages, all of that belongs to Flex. If you have read that Unlimited is where the meter runs, you have read it backwards, and any cost model built on that reading will point you in the wrong direction.
Here is Harvest pricing in 2026 as published on 7 September, per seat per month:
| Tier | Annual | Monthly | Notes |
|---|---|---|---|
| Free | $0 | $0 | 1 seat, 2 projects |
| Teams | $9 | $11 | |
| Enterprise | $14 | $17.50 | Timesheet approvals, activity log, SAML SSO |
| Enterprise Plus | Custom | Custom |
Worth noticing where timesheet approvals sit. If your consultancy cannot invoice until someone has formally signed off the hours, the approval step is not a nice extra, and on Harvest it starts at Enterprise.
What Flex meters, and the numbers Harvest does not publish
Start with the absence, because it is the single most useful thing anyone can tell you about this pricing and no comparison article seems to mention it.
Harvest publishes no usage allowances and no overage rates. Not on the pricing page, not in the help centre, not anywhere public. The help centre confirms that allowances exist. It gives no numbers for them. It says the rates are shown in Settings, under Billing, which means the only way to see what your usage costs is to already be a paying customer with an account open.
Sit with what that does to a purchasing decision. You cannot model the bill before you sign up. You cannot compare Harvest against another tool on total cost, because half the formula is behind a login. You cannot check whether your renewal quote is arithmetically correct, because you have no rate card to check it against. And you cannot show a finance director a forecast, because the forecast has an unknown in it.
What Harvest does publish is the list of what gets counted. Under Flex you pay a per-seat rate plus usage on four dimensions:
- active projects
- active clients
- active tasks
- invoicing
Two details in the definition do real work. First, "active" means something specific: only items with time tracked against them during the period are counted. A dormant client on your books is not billable; a client someone logged twenty minutes against is. Second, the usage charge is calculated from the prior billing period. You are always paying for last period's activity.
One point of care on the fourth dimension. Harvest's own wording is "invoicing" and "additional invoices", which reads as a count of invoices rather than a share of their value. The BBC reported the metering as running on revenue invoiced. Those are different claims with very different consequences for a consultancy that raises a few large invoices rather than many small ones, and we are not going to pretend the ambiguity is resolved. If your billing is concentrated in a handful of high-value invoices, that difference is the first thing to establish with Harvest support before you renew.
Harvest's naming is not entirely consistent with itself, either. Its transaction-fee documentation refers to accounts "not on Harvest's Premium or Unlimited usage billing plans", a vocabulary that does not appear on the pricing page at all. When a vendor's help centre and pricing page use different words for the same thing, the confusion in the market is not really the market's fault.
Why the bill jumps at renewal rather than at sign-up
Harvest supplies its own worked example of the shock, and it is more candid than most vendors would be. The help centre notes that a bill can jump from a seats-only $44 to $404 once usage is added.
The mechanism is the prior-period calculation. Usage is billed from the period before, so a new or newly converted account starts out looking like a per-seat product. The seats are all there is to bill for, because there is no prior period yet. Then the first full cycle of tracked projects, clients, tasks and invoices lands, and the number changes shape.
That is why so many of the stories share a structure: nothing happened for months, and then renewal happened. The increase was not applied to the customer at some announced moment. It arrived when the first metered period reached the invoice.
Harvest told the BBC that customers receive 30 days' notice and then a further 10 days' notice before renewal. Whether 30 days is long enough to migrate a consultancy's time-tracking system is a fair question, and most people answering it discover they are also negotiating under a deadline.
The 7 November 2024 date, and what it does not protect
This one is worth correcting because the internet has it close to backwards.
The date is real and it is Harvest's own. The help centre says accounts upgraded from a trial before 7 November 2024 "may be on one of our Legacy plans". That is a statement about eligibility for a legacy plan, not a promise about price.
It is widely repeated as grandfathering, as though customers on the right side of the date were shielded. Harvest told the BBC the opposite: legacy customers faced the larger increases. If you have been on Harvest for years and have been reassured by the date, that reassurance is pointing the wrong way, and you are in the group with the most to check.
There is a second claim circulating that deserves the same treatment. You will read that the increase lands at each customer's own renewal anniversary rather than all at once. It is plausible, it fits the reported pattern, and Harvest has published nothing that says it. It appears to originate in competitor marketing. Treat it as a reasonable guess about your renewal date, not as a fact about your contract.
The payment fee, and the rate you have probably read
If you take payment through Harvest's invoicing, there is an additional fee on card and ACH payments. The help centre article on transaction fees says the fee applies to accounts "not on Harvest's Premium or Unlimited usage billing plans". It states no rate.
A figure of 0.5% does circulate, and it has a real source: it appears on a separate help page about passing fees on to clients, inside a worked example alongside Stripe's 2.9% plus $0.30. A worked example is an illustration, not a published rate, and the same page says the fee is recalculated at current rates on each recurring invoice. So 0.5% is the number to start a conversation with, not one to put in a spreadsheet.
Anyone quoting you a firm Harvest surcharge percentage has gone further than Harvest's own documentation does.
What the increases have looked like
The reported cases cluster in a wide band, and quoting the top of it as typical would be its own kind of dishonesty. Here is the verified spread.
Salentis, a UK consultancy, went from $130 to $2,110 a month. Its director Richard Haldenby spoke to the BBC on the record (BBC News, Laura Cress, 20 August 2026). The same report carries a US customer whose annual charge went from $2,800 to $23,000.
Bloomberg covered the wider backlash across Bending Spoons' portfolio a week earlier, on 13 August 2026, including a freelancer whose annual renewal went from $211.20 to $2,547.60, a twelvefold jump (Bloomberg, Bergen and Boyne).
Subscription Insider reported on 22 August 2026 that a seven-user business went from $69 to $821.50 a month. Silicon UK, on 31 August 2026, described increases of "more than 15 times", citing a Reddit user at seventeen times.
The developer thread on Hacker News ran on 20 August 2026 and drew 112 points and 110 comments (item 49374920).
So: roughly eight to seventeen times, across the cases anyone has documented. Not a percentage you can apply to your own bill, because the multiplier depends entirely on how much of your work is spread across projects and clients rather than concentrated. A three-person team on one long client engagement and a three-person team on twenty small retainers will get very different answers from the same rate card.
One thing nobody outside Harvest can tell you is how many customers actually left. Bending Spoons publishes net revenue retention at group level in its prospectus, and breaks it out for AOL, Evernote, Remini and StreamYard. Harvest is not among them. The group figure of 94% covers the first quarter of 2026, months before any of this, so it says nothing about the response to the repricing. Anyone citing a churn number for Harvest has made it up.
Modelling your own bill before you have to decide
You cannot compute this precisely, because the rates are not public. You can still work out roughly where you sit, and more importantly whether the metered dimensions are ones your consultancy has many of or few of. That is what determines whether you are an eight-times account or a seventeen-times one.
Start with the count that is easiest to get wrong. Active clients is not the number of clients on your books, it is the number anyone tracked time against last month. Pull last month's timesheets and count the distinct clients. Most consultancies find this number is higher than they expected, because it includes the dormant retainer somebody logged half an hour to and the prospect whose scoping call went on a timesheet.
Do the same for active projects. Consultancies that open a project per engagement will have a small number. Consultancies that open one per workstream, per phase, or per month of a retainer will have a large one, and that is a structural choice made years ago for reporting reasons that now has a price attached to it.
Active tasks is the dimension people forget entirely. If your team logs against granular task lists, discovery, design, build, review, handover, on every project, the task count is projects multiplied by task types. This is the number most likely to surprise you.
Then invoicing. Count what you raised last month. If you bill monthly per client, this is roughly your active client count. If you bill per project or per milestone, it is higher.
Now do the comparison that actually matters. Take those four numbers and ask what happens to them if you grow 30%. On a per-seat tool, a 30% bigger team costs 30% more. On Flex, a 30% bigger team probably means more clients, more projects, more tasks and more invoices, all at once, and all four meters move together. The metered model gets more expensive in a way that accelerates, and it accelerates fastest for exactly the growing consultancy that can least afford a surprise.
Finally, check the Unlimited quote. Since Unlimited is the fixed-price mode, it is the option that makes the bill forecastable again, and for a consultancy with a lot of small clients it may well be cheaper than Flex as well as more predictable. It is worth asking for that number explicitly rather than accepting the default.
If you decide to leave
Migrating a time-tracking system mid-year is unpleasant but not hard. What makes it go badly is discovering, three weeks in, that something you needed was never exported.
- Export the historical time data first, before you cancel anything. Time entries with dates, people, clients, projects, tasks, hours and notes. This is the asset. Everything else can be rebuilt.
- Export invoices as documents, not just as data. You may need to produce the actual PDF a client received during an audit or a dispute, and a row in a CSV is not that.
- Write down your rate card by hand. Billable rates by person, by project and by client, plus cost rates if you use them. These are usually the least well documented thing in any time-tracking system and the most annoying to reconstruct.
- Note who approves what. Not the software configuration, the actual answer: which manager signs off which team's hours, and what happens when they are on leave. Most consultancies have never written this down.
- Pick the cutover at a period boundary. Month end, or the end of a billing cycle. Running two systems across a half month means reconciling two partial timesheets for every person, and someone will get it wrong.
- Keep the Harvest account read-only for one full billing cycle after you move. Cancel it after you have raised and been paid on one complete set of invoices from the new system, not before.
- Check what the notice period actually is. Harvest told the BBC customers get 30 days' and then 10 days' notice ahead of renewal. Work backwards from your own renewal date.
What to look for in a replacement
The lesson of this repricing is not that usage billing is wrong. It is that a price you cannot compute in advance is a risk you have taken on without pricing it. So the first question about any replacement is not what it costs, it is whether you can work out what it costs.
Three things to check, in this order.
Is the whole price published? Every rate, every allowance, every threshold, visible without an account. If any part of the formula is only visible in Settings, you are buying the same problem again from a different vendor.
Does the meter run on anything except people? Clients, projects, tasks and invoices are all things you want more of. A tool that charges for them is a tool that charges you for growing.
Does approval sit behind an upgrade? If your invoices depend on signed-off hours, the approval workflow is core function, not an enterprise feature, and paying an enterprise rate for it changes the comparison considerably.
On price, here is what the alternatives published on 7 September 2026, per user per month. Two caveats before the table, because both of them are the sort of thing this article has been complaining about. Toggl's $14 Premium is an introductory annual rate that renews at $18, and quoting the $14 alone would be exactly the sin under discussion. Clockify prices in euros and Scoro in pounds when read from the UK, so converting either into dollars quotes a figure nobody was actually shown.
| Vendor | Free option | Plan | Annual | Monthly |
|---|---|---|---|---|
| Toggl Track | Yes, limited | Starter | $9 | $12 |
| Premium | $14 first year, then $18 | $20 | ||
| Clockify | Yes, to 5 users | Basic | €3.99 | €4.99 |
| Standard | €5.49 | €6.99 | ||
| Pro | €7.99 | €9.99 | ||
| Enterprise | €11.99 | €14.99 | ||
| Everhour | Yes, to 5 seats | Team, minimum 5 seats | $8.50 | $10 |
| Timely | No, trial only | Starter | $9 | $11 |
| Premium | $16 | $20 | ||
| Unlimited | $22 | $28 | ||
| Productive.io | No, trial only | Essential | $10 | $12 |
| Professional | $25 | $29 | ||
| Scoro | No, trial only | Time billing, minimum 5 users | £13 as printed | not published |
| Hubstaff | No, trial only | Starter | $4.99 | $7 |
| Grow | $7.50 | $9 | ||
| Team | $10 | $12 |
Scoro deserves a note: it has no monthly toggle on that plan and sells combinations of applications rather than a straightforward tier ladder, so it is harder to compare than the row suggests.
TimeSubmit, which we build, is one option among those and worth naming honestly rather than at length. It charges for people and nothing else: no charge for adding a client, opening a project, creating a task or raising an invoice. Core is $42 a month covering 3 members, or $33 a month billed annually, with $14 a month for each additional member. Margin is $110 covering 5 members, or $90 billed annually, with $22 for each member above that. Practice is $39 a month covering 5 members with 1 client and 1 project, and the first two months are at no cost, with a card required to start. Approvers hold accounts and count as members, which is the honest way to put it, though they never see billing rates or invoice totals. Approval workflows are on every plan rather than behind an enterprise tier, and invoicing generates from approved hours. What it does not have: single sign-on, an API, Xero or QuickBooks sync, or a client-facing approval portal. If those are on your list, it is not the tool.
For a broader comparison of what else is available, we keep a roundup of Harvest alternatives for consultancies and a direct comparison with Harvest, plus a page on switching from Harvest specifically.
The short version
Flex and Unlimited are billing modes, not tiers, and Unlimited is the fixed-price one. The tiers are Free, Teams, Enterprise and Enterprise Plus. Flex meters active projects, clients, tasks and invoicing, counted from the previous period, which is why the bill changes at renewal rather than at sign-up. The allowances and overage rates for all four are unpublished and visible only inside a paying account. The 7 November 2024 date marks legacy-plan eligibility and protects nobody; Harvest told the BBC legacy customers saw the larger rises. Documented increases run roughly eight to seventeen times.
If you take one thing from this: ask Harvest for the Unlimited quote in writing before your renewal notice arrives, and count your active clients, projects and tasks today so you can read that quote properly when it does.
Harvest, Bending Spoons and competitor pricing verified against primary sources on 7 September 2026. Software pricing changes quickly; check the vendor before acting on any figure here.