Picture a 120-person fintech that posted a senior backend role 98 days ago, which is 14 weeks. It drew 140 applications, interviewed 9 people and made 2 offers. Both were declined, 1 for a counter-offer from the candidate’s current employer and the other for a higher number elsewhere. The CTO has paid 14 weeks of lost output and asks the recruiter why nobody has said yes. The recruiter blames the market, the hiring manager blames the recruiter and the budget owner suspects the pay band. This scenario is invented for illustration and doesn’t describe a real company or candidate.

This post sorts the delay into 8 causes you can test, with a cost model and a 30-day plan. It draws on Bureau of Labor Statistics wage data, a Stack Overflow survey retrospective and a SIOP summary of hiring-validity research. It also draws on 5 vendor and publisher pages from noon.ai, Proxify, EltexSoft, Karat and Rockstar Developer University, all of which sell something and several of which cite reports they don’t link. I flag their conflicts where they appear. Every cost figure beyond the cited ones is invented and recomputed in code.

TL;DR

  • The 8 causes fall into 3 groups: the spec, the process and the offer. An oversized requirements list and a stale pay band shrink the pool before anyone applies. Slow steps, a long interview loop and an unowned calendar drain it. Weak sourcing, a rigid location policy and offers that lose to counter-offers decide who’s left. Public time-to-hire figures for engineers run from 62 to 95 days and use different definitions.
  • Most delay is waiting, not evaluating. In my invented timeline, 37 days of a 98-day search are idle gaps between steps. An invented 7-stage loop asks a candidate for 11 hours and reviewers for 15.5, and a 4-stage version cuts those to 5 and 8.5. Noon.ai cites a Talent Board figure that 57 percent of candidates have quit a process because it dragged.
  • Price the delay, then fix the cheapest cause first. At an invented $15,000 a month, 95 days open costs $47,500, against $31,000 for 62 days. Offer acceptance data conflicts, with sources citing 51 percent and 82 percent, so measure your own. Confirm any pay or process change with counsel and the budget owner.

Start with what long means

Whether 98 days is long depends on who is counting. Vendor pages put senior engineering searches anywhere from 62 to 95 days, and the numbers aren’t measuring the same thing. As a broader benchmark, SHRM’s 2026 recruiting data reports a median time-to-fill of 39 calendar days for nonexecutive positions, but that covers many job types rather than senior developers specifically. A search that ran 10 weeks may be normal for a senior role and slow for a mid-level role, so compare against your own history for similar roles before you decide there’s a problem.

What matters more than the average is where the time goes. A search can be slow because few qualified people applied, because many applied and the process lost them, or because strong candidates were found and then lost at the offer. Each of those has a different fix, and mixing them up wastes weeks. The 8 reasons below are ordered along the path a candidate walks, from reading the posting to signing the offer, so you can check each stage against your own data.

Before you compare, write down your own definition. Decide whether the clock starts at the posting date, the first sourcing message or the approved requisition, and whether it stops at the accepted offer or the start date. Use the same definition for every search you track. A search that looks like 95 days under a given definition may be 70 under another, and the gap can hide the real delay. Then split the total into waiting time and working time.

Time-to-hire figures and why they disagree

The table lists the figures I found for engineering roles, the report each page names and the page I read them on. None of the underlying reports was linked, so I couldn’t check the method or sample, and each page sells something related to hiring. The numbers don’t line up, partly because of definitions and partly because of the sources. Treat the table as a range of vendor claims, and measure your own median from posting date to accepted offer.

FigureSource as citedPage I read it onCaveat
95 days, senior software engineersIndeed, 2025EltexSoft article on dev.toNo link; author runs a staff augmentation firm
90 to 110 days, senior; 100 to 130 days, AI and MLLinkedIn Talent Insights, 2025Same articleNo link or method
62 days, engineering and technical rolesGem, 2026noon.aiDoesn’t say mean or median, or fill versus hire
48 to 89 days, tech searches; 44 days, all positionsSHRM, 2025ProxifyNamed but not linked; Proxify’s headline says 35 days
About 42 days, software engineersGitnuxRockstar Developer UniversityDated 2023 in 1 place and 2024 in another

Proxify’s headline says hiring takes 35 or more days while its own cited figures run from 41 to 62. EltexSoft’s article is titled for 2026 but relies on 2025 data and shows conflicting update dates. Rockstar Developer University’s page says it was updated in October 2026 while citing February 2026. None of these problems makes the figures wrong, but they’re reasons to use the table as context rather than a target.

Reason 1: the role asks for too much

  • Must-haves multiply faster than candidates. Suppose each of 11 listed requirements is held by 70 percent of otherwise qualified engineers. If they were independent, which they aren’t, only 1.98 percent would hold all 11, against 11.8 percent for a list of 6. Skills correlate, so the real pool is larger, but the direction holds. Compare the role against a realistic IT job description framework and cut the list until each item would change your decision.
  • Years of experience stand in for judgment. A posting that demands 10 years of experience in a specific stack rules out strong engineers who learned fast and moved across tools. I found no public study linking requirement length or years of experience to time-to-fill, so this is my view. Write requirements as outcomes the person must be able to deliver, such as leading a migration or owning an on-call rotation, and test those directly.
  • The role is really 3 jobs. Specs often ask for an architect, a people manager and a hands-on coder in a single hire. Candidates strong in all 3 are rare and expensive, and the ones who exist have options. Decide which of the 3 matters most in the first 6 months, hire for that and plan how to cover the others, perhaps with a lead you promote or temporary capacity through IT staff augmentation while the permanent search continues.
  • The stack requirement is a proxy for onboarding fear. Requiring the exact languages and frameworks your team uses is understandable, but a senior engineer can usually pick up an adjacent stack. If your team can support a ramp-up of a few weeks, say so in the posting. If it can’t, say that too, and expect a smaller pool. Either way, state the trade-off so candidates can decide quickly. Say which case applies when you write the posting.

Reason 2: the pay doesn’t match the market

What public data shows. The Bureau of Labor Statistics reports a May 2025 median of $135,980 for software developers, with 1,717,800 people employed in 2025, but the page doesn’t split pay by seniority. A Stack Overflow retrospective reports that 2025 median compensation rose 24.0 percent for front-end engineers and 27.1 percent for mobile developers over 2024, against 5.1 percent for data engineers and 6.7 percent for cloud infrastructure. Pay moves unevenly by specialty, so a band set 2 years ago can be stale for a single role and fine for another.

Why a stale band costs weeks. If a band set in 2024 had tracked a 24.0 percent change, a $160,000 midpoint would now be $198,400. That’s an illustration of the arithmetic, not a forecast, because survey samples differ from year to year and respondents are self-selected. Before assuming the existing range still works, compare it with current IT salary benchmarks by role and experience level. The risk is real: a candidate who sees a number well below competing offers walks away without telling you why.

Total pay, and who can approve it. I found no public data on bonus and equity for senior developers, so ask candidates which parts of the package matter most. The retrospective lists autonomy as the top satisfaction factor, followed by pay and benefits, which suggests a high base won’t fix a role with little ownership. Proxify says compensation approval and multi-party sign-off add delay but gives no duration. Decide who can approve an above-band offer before you need it.

Reason 3: candidates wait too long between steps

Candidates experience your process as waiting. Proxify says the gap between screening and the first interview takes 7 to 14 days at the median and calls scheduling the single largest source of delay, with no source for the figure. Noon.ai cites a Talent Board report that 57 percent of candidates say they’ve dropped out of a process because it took too long. Neither number is verified here, but the pattern is consistent across the pages: most elapsed time is dead time between steps, not evaluation.

Take an invented timeline for fintech’s 98 days. Response to the application takes 4 days, scheduling the screen 5, the manager call 4, the take-home and its review 8, scheduling the onsite 7, collecting feedback 3, approvals 4 and sending the offer 2. Those gaps add up to 37 days, or about 38 percent of the search. Halving them recovers 18.5 days and brings the search to about 80 days. A senior candidate with other options will often accept elsewhere before your next step is scheduled.

Speed doesn’t have to mean carelessness. The federal government’s guidance on structured interviews describes asking candidates the same predetermined questions and rating responses against the same standards. A fast process works when every interviewer knows what they’re assessing, scores against written criteria and submits the score before discussion. Write the scorecard before the search starts and hold a 15-minute decision meeting within a day of the last interview.

Reason 4: the interview loop is too long

The table compares an invented 7-stage loop with a 4-stage version. It counts hours for the candidate and for reviewers, including preparation and debrief. Both loops cover screening, a conversation with the manager, a technical demonstration and a team conversation. The shorter loop drops repetition rather than signal, and a role-specific IT interview question framework can help you define what each stage is supposed to test before you add another meeting.

StageCurrent: candidate hoursCurrent: reviewer hoursTrimmed: candidate hoursTrimmed: reviewer hours
Recruiter screen0.50.50.50.5
Manager call1.01.01.01.0
Take-home4.01.5n/an/a
Technical interview1.53.0n/an/a
System design1.53.0n/an/a
Paired working sessionn/an/a1.53.0
Team panel2.06.02.04.0
Executive chat0.50.5n/an/a
Total11.015.55.08.5

The trimmed loop cuts candidate time by 54.5 percent and reviewer time by 45.2 percent. At $75 an hour, reviewer cost per candidate falls from $1,162.50 to $637.50, and for 5 finalists from $5,812.50 to $3,187.50. Proxify describes a company that replaced a 72-hour take-home with a 90-minute paired interview and halved drop-off while saving about a week, but that’s a single unsourced case.

A SIOP summary of Sackett and colleagues’ personnel-selection research reports corrected validities of .42 for structured interviews, .40 for job knowledge tests and .33 for work samples. I found no study showing that simply adding more rounds raises validity. Keep every stage that gives you new evidence and remove stages that repeat what an earlier interviewer already measured.

Reason 5: nobody owns the calendar

Feedback lag. Proxify cites a Greenhouse 2024 survey in which 67 percent of recruiters name hiring manager delays as their top bottleneck, without a link. The pattern is easy to recognize: the interview happens on Tuesday and the feedback arrives the following Monday, by which time the candidate has an offer from someone else. Set a rule that written feedback is due within 24 hours, and have the recruiter chase it. Measure the lag per interviewer, since a few people usually cause most of it.

Panel scheduling. A 3-person panel needs 3 calendars to line up, and each added interviewer multiplies the wait. Block recurring interview slots on the team’s calendars for the length of the search, so a candidate can be offered a time the same day. Proxify says automated scheduling cuts 5 to 7 days from time-to-fill, but it gives no source, and its parent business benefits from selling speed. Test it on your own process.

Approvals. Offers wait for compensation sign-off, headcount confirmation and sometimes a legal review. Agree on the band and the approver before the first interview, so a strong candidate doesn’t wait 4 days for a yes. The invented timeline above assigns 4 days to approvals, which is the same as the manager call wait. Name a single person who owns the search end to end, with authority to schedule, chase and escalate, and give that person a weekly report of the idle days.

Reason 6: sourcing depends on people already looking

  • Applications are a thin funnel. EltexSoft cites a Greenhouse 2024 figure of about 120 applications per hire, built from a 5 percent application-to-interview rate and a 17 percent interview-to-offer rate. Those 2 rates multiply to 118 applications per offer, not per hire, because they leave out acceptance. At the 51 percent acceptance some sources cite, 231 applications produce 1 hire. If internal sourcing isn’t reaching enough qualified people, compare the benefits of specialized tech recruitment agencies before simply buying more job-board traffic.
  • Referrals convert better, per vendor. Rockstar Developer University cites Gem 2025 for a hire rate of 30 percent for referrals against 7 percent for cold applications, and says 65 percent of tech hires come through referrals, citing The Interview Guys. Neither is linked, and the course seller benefits from the message that relationships beat volume. The direction is plausible. Ask your engineers whom they’d work with again, and pay for referrals in a way that fits your policy.
  • Senior engineers are often not looking. A specialized tech recruiter can add an outbound channel when the strongest candidates aren’t scanning job boards. Stack Overflow’s retrospective reports 24.5 percent of respondents happy at work, 47.1 percent complacent and 28.4 percent unhappy. A complacent engineer may still answer a thoughtful message about a specific problem, so plan for personalized outreach rather than waiting for applications.
  • Existing relationships are the cheapest source. Former colleagues, open source contributors and past interviewees who declined are warm leads that cost little to contact. Keep a list of strong candidates from earlier searches, with notes on what mattered to them, and reach out when a role opens. I found no public data on conversion for this source, so measure it yourself and compare it with what you pay for other channels.

Reason 7: the location policy shrinks the pool

The 2025 Stack Overflow Developer Survey shows substantial remote and flexible work among developers, including a 45 percent remote share among US respondents. The retrospective used in the original research reports that 32.4 percent of 2025 respondents worked fully remote and that 82 percent had some remote flexibility. Those figures describe respondents rather than the entire developer market, but they suggest an on-site-only senior role competes against employers offering more flexibility.

The cost of a rigid policy depends on your market. If your office is in a metro with a deep pool, an on-site requirement may cost little. If it’s elsewhere, you may be restricting yourself to a few thousand people. A fully remote posting widens the pool and your competition at the same time, since you’re now bidding against employers everywhere. I found no public study measuring the effect of work location on time-to-fill, so I tested it: post a hybrid version, track applications and compare quality as well as volume.

State the flexibility terms plainly in the posting: how many days in the office, what travel to expect, whether relocation help exists and whether the arrangement can change. Ambiguity costs time, because candidates who can’t tell either ask or leave. If your policy is under review, say so honestly. A senior engineer deciding between offers will weigh the policy heavily, and a clear answer on day 1 can save a week of back-and-forth later in the process.

Reason 8: offers lose before they close

What acceptance data says. The sources conflict sharply. EltexSoft cites Indeed for offer acceptance falling from 73 percent to 51 percent over 3 years, and Rockstar Developer University cites Second Talent and Huntr for 51 percent in Q2 2025, down from 74 percent in Q2 2023. Noon.ai cites Gem 2026 for an 82 percent average. These can’t all describe the same population. Karat’s close rates by city, 40 percent in Seattle, 44 percent in the Bay Area and 56 percent in New York, come from a page published in June 2021, though Rockstar Developer University dates them 2023.

Why the numbers disagree. They likely measure different things: all roles versus engineering, offers extended versus offers reaching final stage, and different years. None gives a method, so I can’t reconcile them. What they share is the direction that senior engineers hold more than 1 option, which EltexSoft states without a source, along with a claim that 40 percent accept counter-offers. Both are unverified. Track your own: of offers you extend, how many are accepted, and what did the decliners say?

What to do at the offer stage. Call the candidate before the offer to learn what they need, then make the offer fast and at the right number, with a short deadline you can defend. Ask the hiring manager to call personally. If a counter-offer is likely, ask the candidate what would make them stay and address it directly. Keep a second finalist warm until the first signs. A slow, low first offer is the most common way to lose a strong candidate.

What the delay costs: an invented model

The table prices an open senior seat at 3 durations and 3 monthly costs. The $15,000 figure is my invention. The $4,000 and $9,000 figures are the monthly lost productivity that Proxify attributes to SHRM, which I couldn’t verify because the report isn’t linked. EltexSoft gives a vacancy cost of $500 to $1,500 a day with no source, which would be $15,000 to $45,000 a month and shows how loose these estimates are. Replace the figure with your own estimate of lost output.

Days openMonthsAt $4,000 a monthAt $9,000 a monthAt $15,000 a month
451.5$6,000$13,500$22,500
622.07$8,267$18,600$31,000
953.17$12,667$28,500$47,500

The 33-day gap between 95 and 62 days costs $4,400, $9,900 or $16,500, depending on the column. The model leaves out the cost of the recruiter’s time, the interview hours from the earlier table and the risk that the team burns out covering the gap. It also leaves out the value of the work the new hire would have started. Use it to rank fixes, not to justify a budget on its own, and replace every assumption with your own numbers before you present it.

How to diagnose your own funnel

  • Count every stage, with dates. List applications, screens, interviews, offers and acceptances for this search. In the invented fintech, 140 applications gave 9 interviews, a rate of 6.4 percent, and 2 offers from those 9, a rate of 22.2 percent, with no acceptances. Compared with the 5 and 17 percent figures EltexSoft cites, the top of the funnel looks healthy. The loss happens at the offer, which points to reasons 2 and 8 rather than reason 6.
  • Measure the waiting days at each step. Pull timestamps from your applicant tracking system and compute the median days from application to screen, screen to manager call, interview to feedback and final interview to offer. A few outliers usually explain most of the delay. Share the numbers with the interviewers who cause them, and ask what gets in the way. Often the answer is a calendar problem that a standing time slot fixes.
  • Ask decliners and withdrawals why. A 10-minute call with each person who turned you down, or dropped out, tells you more than any benchmark. Ask what they chose instead, what the number was and what in your process helped or hurt. Code each answer to 1 of the 8 causes and tally them after 5 or 6 calls. Be ready to hear things about pay or about the hiring manager that nobody wanted to say.
  • Compare against your own history, not vendor medians. Look at your last 5 senior hires and record days to fill, interviews per hire and acceptance. If this search is 20 days slower than your usual, find what changed. If it’s typical, the cause may be the market, and the fix is wider sourcing or a better offer. Vendor numbers are too loose, and too self-interested, to serve as your target.

What to fix first

This week: ownership, feedback and approvals. Name a single owner for the search, set a 24-hour rule for written feedback, confirm the pay band and the approver and block standing interview slots. These changes cost almost nothing and address reasons 3 and 5. In the invented timeline, recovering half of the 37 idle days is worth 18.5 days, which is $9,250 at the invented $15,000 a month. Report idle days weekly so the improvement stays visible.

This month: spec, loop and pay. Cut the requirements list to the items that would change your decision. Replace the longest stage in the loop, usually a take-home, with a shorter working session, and test the pay band against what declined candidates accepted. These changes take effort from the hiring manager and the budget owner, and they address reasons 1, 2 and 4. Expect some pushback on the loop, since interviewers like to keep their slot.

This quarter: sourcing and policy. Build a referral process, a list of warm candidates from earlier searches and a consistent outreach routine for passive candidates. If that still doesn’t create enough qualified conversations, test software engineering staffing as an additional sourcing channel rather than changing 5 variables at once. Revisit the location policy with real applicant data, and write a spec template that starts from outcomes.

When outside help makes sense

A recruiting firm can widen your sourcing, but it can’t fix a stale band or a slow loop. EltexSoft, itself a staff augmentation firm, states recruiter fees at 15 to 25 percent, which would be $27,000 to $45,000 on a $180,000 salary, without a source. Fees vary by arrangement, so get a written quote from any IT staffing agency you’re considering. Compare that range with the cost of delay in the invented model only after you’ve fixed the causes inside your control.

Proxify and EltexSoft both sell staffing, and both articles frame faster hiring through their own networks as the answer, so weigh the advice accordingly. If you do hire a firm, compare tech recruiting firms on what they’ll change about the search, how many candidates they’ve actually placed in roles like yours and what the guarantee says if the hire leaves. Give the selected firm the pay band, the process and an approver on day 1.

A 30-day recovery plan

  • Days 1 to 5: diagnose. Count the funnel, pull the waiting days, call 5 declined or withdrawn candidates and list the answers by cause. Confirm the pay band with the budget owner and ask whether an above-band offer is possible, and who can approve it. Write down the search owner and the 24-hour feedback rule, and tell the interviewers. Estimate the monthly cost of the vacancy and put it at the top of the page.
  • Days 6 to 15: remove the easy delays. Block standing interview slots, trim the requirements list, shorten the loop and set a weekly pipeline review of 20 minutes. Rewrite the posting around outcomes and state the pay range if your policy allows it. Start personalized outreach to 30 or more passive candidates, including past finalists. Track replies by channel, and ask every candidate how long each step felt. Review the posting with a recent hire.
  • Days 16 to 25: run a faster loop. Put the new loop in place for every active candidate, offer same-day scheduling and require feedback within 24 hours. Prepare the offer before the final interview, with the approver on call. Ask the hiring manager to speak with each finalist personally. Keep a second finalist warm, and keep the process fair by applying the same stages and the same scoring to everyone.
  • Days 26 to 30: close and review. Make offers within a day of the final decision, with a short deadline you can defend and a plan for a counter-offer. Compare days open, idle days and acceptance with your earlier history, and write down what changed. Share the result with finance and the leadership team. Ask counsel to review the offer terms and any new process steps before you roll them out to the next search.

FAQs

How long should it take to hire a senior developer?

There’s no reliable public figure. Vendor pages cite 62 days for engineering roles from Gem, 95 days for senior engineers from Indeed and 48 to 89 days for tech searches from SHRM, none of them linked, and they use different definitions. Compare against your own last 5 senior hires, measured from posting to accepted offer, and treat anything far above your history as a signal to check each stage. Update the baseline each quarter as you hire.

Why do senior developer searches take so long?

The causes fall into 3 groups: a spec and pay band that shrink the pool, a process with long idle gaps and a weak offer stage. Proxify says most delay is dead time between steps, and noon.ai cites a Talent Board figure that 57 percent of candidates have dropped out of a slow process. Both are vendor claims, so measure your own waiting days. Start with the cause that removes the most idle days at the least cost.

How many interviews should a senior developer loop have?

I found no study that identifies the right number. A SIOP summary of a 2022 meta-analysis reports corrected validities of .42 for structured interviews, .40 for job knowledge tests and .33 for work samples, and says nothing about the number of rounds. My invented trimmed loop has 4 stages and 5 candidate hours. Keep every stage that tells you something new, and drop repetition. Ask interviewers what each stage adds before you keep it.

What’s a good offer acceptance rate?

The sources conflict. EltexSoft cites Indeed for a fall from 73 to 51 percent over 3 years, and Rockstar Developer University cites 51 percent in Q2 2025, but noon.ai cites Gem for an 82 percent average. None gives a method. Track your own rate and ask every decliner why. If your rate is low, check pay, speed and the hiring manager’s involvement before you blame the market. Compare it with your last 5 senior offers.

Does allowing remote work fill a senior role faster?

I found no public study that measures it. Stack Overflow’s 2025 retrospective reports that 82 percent of respondents had some remote flexibility and 17.9 percent were strictly on-site, so an on-site-only role competes against employers who offer more. A remote posting widens the pool and the competition. Test a hybrid or remote version, and compare the quality of applicants as well as the volume. Remember that the survey is self-selected and shows current arrangements, not preferences.

Should I use a take-home test for a senior developer?

It’s a trade-off. Proxify describes a company that replaced a 72-hour take-home with a 90-minute paired interview and halved drop-off, but that’s a single unsourced case. A long take-home adds candidate hours and waiting time, and it may exclude people with little free time. A paired working session of 90 minutes costs reviewers more but finishes faster. Ask counsel about any assessment. Whichever format you choose, give every finalist the same instructions and scoring.

How much does an open senior developer role cost?

Estimates are loose. Proxify attributes $4,000 to $9,000 a month in lost productivity to SHRM without a link, and EltexSoft cites $500 to $1,500 a day without a source. At an invented $15,000 a month, 95 days costs $47,500 and 62 days costs $31,000. Use your own estimate of lost output, and rank fixes by how many idle days they remove. Count the idle days in your own process, then multiply by your estimate of the monthly loss.

When should I use a recruiter for a senior developer search?

When your own sourcing has run out of reach and your process is already fast. EltexSoft states recruiter fees at 15 to 25 percent without a source, which is $27,000 to $45,000 on a $180,000 salary. A firm can widen the pool, but it can’t fix a stale pay band or a slow loop. Get fees, the guarantee and the replacement terms in writing before you sign. Ask for references from clients who hired similar roles.

By

Leave a Reply

Your email address will not be published. Required fields are marked *