Ashby vs Greenhouse vs Lever: The Three-Way Comparison at Series B

TL;DR

• Three platforms sold to the same buyer, and only one of them publishes anything you can read. That single fact shapes the evaluation more than any feature difference.
• Ashby's pricing page is a configurator rather than a list, so its headline figure moves with the company size and payment term you select and should never be quoted as a flat price.
• Greenhouse and Lever both publish no price at all, confirmed on their own sites on 10 October 2026. Also worth knowing: greenhouse.io now redirects to greenhouse.com.
• The question that decides it is where your bottleneck is. Consistency across many hiring teams, outbound sourcing and nurture, or not knowing because you have no data.
• Because you cannot benchmark against a published market price, you have to build your own benchmark, and the method for that is in this article.
• We have not audited any of the three products feature by feature, and no comparison should pretend to. The rows worth verifying yourself are listed explicitly.

Three Demos, One Buyer, No Prices

A company that had just closed a Series B round, 110 people and planning to hire around sixty in the following year, ran a proper evaluation. First head of talent in post six weeks, a scoring matrix with eleven weighted criteria, three demos.

All three demos were good. All three products were credible. The scoring matrix separated them by a margin so small that the head of talent described it, accurately, as noise.

Then the quotes arrived over a five-week period, each structured differently, two of them with modules priced separately and one as a bundle, and one of the three had already been revised once because the headcount band had been misunderstood. There was no published market price to check any of them against, so the only available benchmark for whether a number was reasonable was the other two numbers, which had been produced by vendors who knew they were competing. The company negotiated, got a discount it could not evaluate, and signed.

Recommended Tools

Best tools for Applicant Tracking (ATS)

B BambooHR HR management system L Lattice Performance management D Deel Global payroll & EOR

View all recommended tools

Eighteen months later the head of talent's honest assessment was that any of the three would have worked and the real variable had been the process they built on top of it. That is a more useful finding than any feature comparison, and it is the reason this article spends as much space on how to run the evaluation as on the products.

What This Comparison Can and Cannot Tell You

This section exists because the alternative is a feature matrix with invented rows, and a publication that takes no vendor money should say which rows it checked.

What we verified ourselves, with dates. Ashby's pricing page is a configurator: at the default company size and payment selection it returned around $400 per month for its all-in-one foundations tier when read on 10 October 2026, with candidate texting at $90 per additional thousand messages and single sign-on at $100 per month shown as add-ons. The figure depends on the size and term selected, so it is a configurator output rather than a list price.

The other two publish nothing. Greenhouse publishes no price, confirmed on its own site on 10 October 2026. Lever publishes no price, confirmed the same day. And greenhouse.io issues a redirect to greenhouse.com, which is worth knowing if you hold older links or internal documentation.

What we did not verify, and what nobody should take a general comparison's word for. Any specific feature claim, because feature sets in this category change quarterly and a comparison written against last quarter's product is worse than no comparison. Whether a given module is included in your quote, which is negotiated. Integration depth with your particular stack. And implementation timelines, which depend on your own data and processes far more than on the vendor.

So what follows is positioning, the question that decides between them, and a method. The positioning is drawn from how each vendor presents itself and from what each is consistently chosen for, which is defensible. The features are for you to check.

The Pricing Picture, Verified

Worth isolating because it is the most actionable thing in this article.

One of the three publishes figures and does so through a configurator. That is genuinely more transparency than its two competitors offer, and it is also the shape of transparency most easily misquoted: a configurator output is a price for one set of selections, and third-party articles routinely turn it into a flat monthly rate. If you see a single number quoted as what that platform costs, treat it as a figure somebody read at one default setting rather than a list price.

The other two publish nothing. Not a starting figure, not a band, not a per-employee rate.

Three practical consequences, and they are about method rather than product.

You cannot tell whether a quote is good. There is no published reference point, so the only comparison available is between the quotes you have obtained, from vendors who each know roughly what the others charge and that they are competing. That is a negotiation rather than a market price, and the outcome depends more on your preparation than on the product.

Sequential evaluation is actively harmful. A quote obtained in week two and a quote obtained in week nine are not comparable: the requirement has moved, the later vendor knows there is a competing offer, and the two quotes sit at different points in each vendor's own quarter. The fix is to run all three in parallel against one written requirement, which is the section further down.

Per-module pricing is the only comparable form. A bundle total cannot be compared across three vendors with three different bundles. Ask each for a line per module and a line for implementation, and the comparison becomes arithmetic rather than judgement.

The Question That Decides It

Where is your hiring bottleneck?

There are three answers at this stage of a company, and each points at a different one of these products.

Consistency across many hiring teams. You have twelve managers hiring, each running a different process, and the problem is variance: different interview loops, different standards, candidates getting different experiences, and no way to compare two candidates assessed by two teams. This is a process standardisation problem.

Finding people who are not applying. Your inbound volume is low or the wrong shape, and the hires that matter have to be sourced. The bottleneck is outbound work: building lists, running sequences, nurturing people who are not looking yet, and keeping a relationship warm across quarters.

Not knowing where the bottleneck is. You have no reliable data about your own funnel, so every conversation about hiring is anecdote. Which stage loses people, which sources produce hires, how long each team takes, what an offer acceptance rate looks like. This is a measurement problem and it is the most common honest answer at Series B.

Those three map onto the three products closely enough to be useful: standardisation towards Greenhouse, outbound and relationship management towards Lever, measurement and analytics towards Ashby. Because all three products do all three jobs to some degree, a feature comparison will show you three platforms that all appear to cover everything, which is why the bottleneck question has to be answered before the demos rather than during them.

Where Ashby Fits

Best for a company that wants its hiring to be measured and run by somebody who will use the measurement.

The positioning is analytics-led and all-in-one, and the reason it reaches shortlists at this stage is that a newly hired head of talent frequently has no data and needs some before they can change anything. A platform whose centre of gravity is reporting gives that person an argument to take to a leadership team, which is a specific and real benefit at Series B.

It is also the only one of the three that publishes figures, through the configurator described above, which is worth something beyond the number itself: a vendor that shows you a price before a conversation is easier to evaluate quickly.

Where it is weaker for this buyer. Analytics return value in proportion to volume, so a company hiring twenty people a year will not get much from the thing it is best at. The depth also implies somebody whose job is to look at it; without that person you are paying for capability that nobody opens. And the configurator's headline figure depends on your selections, so the number you see first is not necessarily the number you will be quoted.

Where Greenhouse Fits

Best for a company whose hiring is distributed across many managers and needs to be made consistent.

The positioning is structured hiring, and the reason it is chosen is that it is the most established answer to variance across teams. A company with a dozen managers each running their own loop has a real and expensive problem, because inconsistent assessment produces both bad hires and good candidates lost for reasons nobody can reconstruct. A platform built around a defined process is the right shape for that.

It publishes no price, confirmed on its own site on 10 October 2026. The redirect from greenhouse.io to greenhouse.com is worth noting if you hold older documentation or links.

Where it is weaker for this buyer. Structure is adoption work, and adoption work needs managers to change behaviour. A company that buys it and does not invest in the process change gets a more expensive version of what it had, which is the most common disappointing outcome in this category. It is also the choice most likely to be made for the company you expect to be rather than the one you are.

Where Lever Fits

Best for a company whose important hires have to be sourced rather than attracted.

The positioning combines tracking with candidate relationship management, and the reason it is chosen is that outbound hiring has a different shape from inbound. A person you contact who is not looking is a relationship to maintain over quarters, not an application to progress over weeks, and a product that treats them as the latter loses them.

It publishes no price, confirmed on its own site on 10 October 2026.

Where it is weaker for this buyer. The relationship capability assumes somebody spending hours a week on outbound work. At Series B that person often does not exist yet, or exists as a fraction of one role, and capability bought for a person you have not hired is capability you pay for and do not use. The honest test is whether sourcing is in somebody's objectives today.

When None of the Three Is the Answer

When your volume is too low

Below about twenty hires a year, all three are more platform than the problem requires and the per-hire cost becomes a genuine comparison with other uses of the money. The sensible options at that volume have flat pricing or are already in your HR platform, and Workable publishes a readable list for comparison: read on 10 October 2026, Standard is $299 per month or $3,588 per year, Premier is $599 per month or $7,188 per year and Enterprise is $719 per month or $8,628 per year.

When the problem is the job adverts

A surprising share of hiring complaints at this stage are about the quality and specificity of what is being advertised rather than about tracking it. If your bottleneck is that the right people are not applying and nobody is sourcing, a platform does not address it and two hours of somebody competent rewriting three adverts might.

When you are an agency

If you place people into other companies, all three are the wrong category, because they assume one employer, treat the job as the central object and have no concept of a fee. A recruitment CRM is the category that fits.

When the real problem is the interview loop

If managers disagree about what good looks like, that is a definition problem that precedes any tool. Write the scorecard on paper, agree it with three managers, use it twice, and then buy something to enforce it. Buying the enforcement before the agreement produces a form nobody completes.

Five Questions People Ask First

"Which one is cheapest?" Unknowable outside a sales process, and anybody telling you otherwise is quoting a configurator default or one customer's negotiated outcome. What you can control is obtaining three comparable quotes, which requires them to be run in parallel and itemised per module.

"Is Ashby only for companies that love dashboards?" The analytics depth is the differentiator and it needs somebody to use it, which is a statement about your team rather than the product. If your head of talent came from a data-led function and is asking for funnel metrics in week one, that is a strong signal. If nobody has asked for a report, it is capability you may not open.

"Is Greenhouse the safe choice?" It is the most established and that has real value, particularly if your investors or board have seen it work elsewhere. Safe is not free, though: the structure is its benefit and the structure is adoption work, so the risk is not that the product underperforms but that the process change does not happen.

"Do we need candidate relationship management?" Only if somebody is doing outbound work or will be within six months, with it in their objectives. Nurture capability without a nurturer is the clearest example in this category of paying for a person you have not hired.

"How long will implementation take?" Weeks rather than quarters for all three, and the variable is your data and your process definitions rather than the product. A company with clean data and an agreed interview loop will be fast on any of them. One with neither will be slow on all three and will attribute it to the vendor.

How to Run Three Parallel Sales Processes

This is the part that determines whether you get a good outcome, and it is almost entirely within your control.

Write a two-page requirement and send the identical document to all three on the same day. Headcount now and in twelve months, hires planned by team, number of hiring managers, whether anybody sources, the integrations that must exist, and the three things you want to be able to do in six months that you cannot do now. Two pages. Ask each vendor to structure their response against it.

Ask for itemised pricing: a line per module, a line for implementation, and the discount for annual commitment. Refuse a single bundle figure politely and persistently. An itemised quote is comparable, tells you what you are paying for something you may not use, and gives you something to remove in a negotiation.

Run all three demos within ten days, against your own scenario. Not the vendor's sample data. Bring one real role, one real candidate shortlist shape, and your actual interview loop, and ask each vendor to run it. Three demos in ten days are comparable because your own understanding has not moved between them.

Ask all three the same five verification questions in writing. The rows from the table earlier that a comparison cannot settle. Written answers are comparable and the willingness to commit something to writing is itself informative.

Set one decision date and tell all three what it is. Not to manufacture pressure but because the alternative is drift, and drift is how a parallel evaluation becomes a sequential one by accident.

And ask each for two reference customers at your headcount, hiring at your volume, with a similar manager-to-recruiter ratio. Shape matters more than sector. A reference at four hundred people hiring three hundred a year will tell you about a product you are not buying.

How to Choose: Five Questions Before You Talk to Any of Them

Which of the three bottlenecks do you actually have? Consistency, sourcing, or measurement. Write it in one sentence and have the hiring managers agree it. This single answer does more to narrow the field than the whole scoring matrix.

How many hires, by how many managers? Both numbers. Sixty hires by six managers is a volume problem. Sixty hires by twenty-two managers is a consistency problem, and they point different ways.

Does anybody source, and is it in their objectives? A yes changes the ranking. A no means relationship-management depth should be weighted low regardless of how good it looks.

Has anybody asked for a funnel metric in the last month? A proxy for whether analytics will be used. If the answer is yes and the asker has authority, weight measurement heavily. If nobody has asked, be honest that dashboards will be opened in the first fortnight and then rarely.

What will be different in six months? Three specific things, named. If you cannot write them, the evaluation will be decided by whichever demo was most impressive, which is the outcome the whole sales process is designed to produce.

The Comparison

Only rows we checked or that are matters of positioning. The right-hand column is the work.

Three of the seven rows say to verify it yourself. That is the real state of the information rather than evasion, and a table that filled them in confidently would be guessing on your behalf about the things most likely to have changed since it was written.

The Decision Table

Diagnosing Your Own Bottleneck in an Hour

The one-sentence bottleneck answer carries most of this decision, and companies arrive at demos without it because diagnosing it sounds like it needs the data they do not have. It does not. It needs three questions asked of five people.

Of the last ten roles you tried to fill, how many filled, and how long did each take from approval to acceptance? Approximate is fine. If most filled and the variance between teams is wide, you have a consistency problem. If several did not fill at all, you have an attraction or sourcing problem, and no tracking system addresses it.

For the roles that filled, where did the person come from? Applied, referred, or contacted by you. If almost nobody was contacted by you and the hard roles are the ones that did not fill, that is the sourcing answer. If most applied and the problem is what happened after they applied, it is not.

Ask three hiring managers to describe their interview loop from memory. Separately, in two minutes each. If the three descriptions differ materially, you have the consistency answer and you have the evidence for it, which is more persuasive to a leadership team than any vendor's benchmark.

Then ask yourself the fourth question: could you answer the first two from a system, or did you have to ask people? If you had to ask people, the measurement answer is also true, and at Series B it usually is alongside one of the others.

So rank them rather than picking one. Measurement is almost always second, because it is the condition for knowing whether the first fix worked. What matters is which is first, since that is the capability you should pay most for and the one your demos should be built around.

And write the sentence down with a date, then show it to the hiring managers. If they disagree with it, you have found something more important than a product, because a platform bought to fix a bottleneck the managers do not believe in will be adopted by the talent team alone, which is the pattern behind most of the disappointing outcomes in this category.

The Scorecard You Write Before Buying Enforcement

This article says twice that a definition problem precedes any tool, and that managers should agree a scorecard on paper and use it twice before buying something to enforce it. Here is what that document contains, because it is an afternoon and it changes the evaluation.

Four to six criteria, no more. Each one a thing you could observe in an interview rather than a quality you could assert about a person. Not strong communicator, which nobody disagrees with and nobody can assess consistently. Explains a technical decision to somebody outside their discipline, which two interviewers can separately judge.

One sentence per criterion saying what a weak, adequate and strong answer looks like. This is the whole value of the document and the part always skipped. Without it, a four-point scale means four different things to four managers, and the scores are not comparable even when the form is identical.

A named owner per criterion within the loop. Which interview assesses which thing. Two interviewers assessing the same criterion independently is useful by design; five interviewers all assessing culture fit is a loop with no information in it and three hours of wasted manager time per candidate.

A rule about what a single low score means. Does one weak criterion end the process or start a conversation? Agree it in advance, because deciding it per candidate is how bias enters a process that looks structured.

One free-text box, kept. Three or four sentences from each interviewer, written the same day. At this volume the written summaries are read more often than the scores and they are what makes a decision defensible in six months.

Then use it twice, on two real roles, before any demo. Two things happen. Managers discover that two of the criteria cannot be assessed in the time available, which is a finding you want before it is configured into a platform. And you arrive at the evaluation with a real artefact to ask each vendor to reproduce, which is a far better demo than the vendor's sample loop.

And if the managers will not agree the document, that is the most useful result available, because it tells you the constraint is not the software. A platform configured to enforce a loop nobody agreed will be worked around within a quarter, and the workaround is a verbal decision followed by somebody filling in the form afterwards, which is the state most structured hiring implementations quietly reach.

Building Your Own Price Benchmark

With two of three vendors publishing nothing, a benchmark has to be constructed. This is how, and it takes about two hours.

Convert everything to an annual figure and a per-hire figure. Every quote, including the implementation fee amortised over the first year. Then divide by your planned hires. Per-hire is the number that makes three differently structured quotes comparable, and it is the number a finance team will ask for anyway.

Use the one published figure as an anchor, correctly. Workable's published annual rates give you a defensible lower bound for what employer-side applicant tracking costs at a flat price point. That is not a claim that these three should cost the same; it is a reference point, and having any reference point is better than having none.

Price the modules you would not buy. Ask each vendor what the quote is without the two modules you are least likely to use. The answers tell you how much of the bundle is load-bearing and give you the most straightforward negotiation lever available.

Ask every vendor the same question about year two and year three. The uplift at renewal, and whether the discount is contractual or discretionary. This is where quote-based pricing diverges most and where a first-year discount is most often recovered.

Write down what you would pay. Before any quote arrives, decide the annual figure and the per-hire figure you consider reasonable, and the number above which you would walk. Doing this first is the only protection available against anchoring on whichever quote arrives first, and in a market with no published prices, anchoring is the main mechanism by which buyers overpay.

So the output is a single table with three columns of comparable annual and per-hire figures, a note of what each excludes, and your own pre-committed number. That is a benchmark. It is not a market price, because no market price is published, and it is enough to negotiate from and enough to defend to a board.

What Getting This Wrong Costs

The visible cost is overpaying, and with no published prices you will not know whether you did.

The second cost is the one the head of talent in the opening example identified: buying any of the three and not building the process it assumes. All three products encode a way of working. Greenhouse assumes managers will adopt a defined loop. Lever assumes somebody is sourcing. Ashby assumes somebody reads the numbers and acts. A platform whose assumption is not met in your company does not fail loudly; it quietly becomes a place where applications are stored, which is a function you could have had far more cheaply.

The third is time. An evaluation run sequentially across three quote-based vendors takes a quarter of elapsed time, much of it the first head of talent's, during which hiring is happening on whatever was there before. Running it in parallel compresses that to about six weeks. So the method is worth more than the product choice at this stage, which is an uncomfortable conclusion for a comparison article and is what the evidence supports.

And the reframing question: in six months, what will a hiring manager be doing differently? If you cannot name it, all three are the same purchase and you should choose on price, which brings you back to needing a benchmark you have had to build yourself.

When You're Ready to Move

The honest position is that at Series B all three of these are credible and the companies that regret their choice usually bought for a bottleneck they did not have.

The trigger worth acting on is not the funding round, although that is when the budget appears. It is the first time two hiring managers assess the same candidate differently and nobody can reconstruct why, or the first time somebody asks which source produced your best hires and the answer is unavailable. One of those is a reason to move. A round closing is a reason to have the budget, not a reason to have the requirement.

The sequence that works costs nothing and takes a week. Write the one-sentence bottleneck and have your hiring managers agree it. Count hires per manager rather than hires. Check whether sourcing is in anybody's objectives and whether anybody has asked for a funnel metric. Write the three things that will be different in six months. Then decide the annual and per-hire figures you would consider reasonable, before any vendor tells you a number.

Then run all three in parallel, with one requirement document sent on one day, demos inside ten days against your own scenario, itemised quotes, the same written verification questions, references of your shape, and one decision date. That process produces three comparable numbers, which in a market where two of the three vendors publish nothing is the entire advantage available to you, and a choice that rests on your own bottleneck rather than on which demo was best.

Frequently Asked Questions

HROpsLab takes no vendor money and publishes no paid placements, which is why this comparison names the rows it could not verify instead of filling them in.