The Mid-Loop Down-Level Call, Scored to Accept, Contest, or Walk

A recruiter has just told you they are "now thinking of you for the level below," and you have to decide quickly whether to go along with it, push back for the original level, or pull out before you put more prep time into a smaller pay band. This guide is for candidates anywhere between the recruiter screen and the debrief. It gives you six dimensions to score, a way to price the real gap between the two bands against your own floor, and a verdict of Accept, Contest, or Walk, with the reply to send for each.

What a down-level signal is, and where in the loop it gets set

A down-level signal is any message saying the employer now plans to hire you one level below the one you applied or interviewed for. It can arrive at the screen, after a hiring manager call, mid-loop, after the debrief, or only in the written offer. The stage tells you who made the call and how much room there still is to change it.

These terms come up throughout the guide:

• Pay band: the minimum-to-maximum pay range an employer sets for one level.
• Midpoint: the centre of the band. In the WorldatWork/Deloitte survey of salary structure practices, 96% of organizations tie range midpoints to the 50th percentile of the market.
• Midpoint progression: the percentage step between the midpoints of two adjacent levels.
• Range spread: how wide a band is. A 50% spread usually means the minimum is 80% of midpoint and the maximum is 120%.
• Overlap: how much of one band is shared with the next, usually expressed as a percentage of the previous band.
• Compa-ratio: a salary divided by its band midpoint. A compa-ratio of 0.80 means the salary is 80% of midpoint.

Who actually decides the level

Who sets the level differs by employer, and it decides who you need to persuade:

• Amazon: the Bar Raiser, not the hiring manager or recruiter, runs the debrief. A typical loop has about five interviewers: the hiring manager, two or three team members, and a Bar Raiser.
• Meta: candidates describe a broader review after the loop that mostly confirms the decision and settles the level.
• Microsoft: the team debriefs, and the hiring manager makes the decision with the recruiter.

Down-leveling usually comes from the system design or behavioral rounds, because those are where interviewers judge how senior you are. Coding rounds matter less for level. How often it happens is not published. One Bar Raiser estimated roughly 10 to 15% of candidates in their own experience, which is one person's view, not a rate. Early signals happen too. One candidate on Blind was down-leveled after a hiring manager call, before the onsite.

The six dimensions of a down-level score

Score each of six dimensions 0, 1, or 2. Three of them measure leverage, meaning how likely a contest is to work. The other three measure cost, meaning how much the lower level takes from you if you accept it.

I set the cut-offs in this rubric. Nobody publishes them. The 15% pay line is the top of published professional midpoint progression. The 30% line splits the real tech-ladder gaps in the pricing section below. Move them if your field's ladders look different, but decide where they sit before you score.

Stage

What it proves: how much new evidence you can still add. At the screen or mid-loop you still have rounds ahead of you, including the system design and behavioral rounds where seniority is judged. After the debrief, a re-evaluation needs new evidence, not a verbal argument.

When it lies: if you have already talked about salary or start date, score 0 whatever the stage. Exponent's guidance on reversing down-levels is blunt: "If you start talking salary or start date, you've implicitly accepted the level."

Reason specificity

What it proves: whether there is anything concrete you can answer. A named round tied to a named ladder criterion gives you a target to retake against.

When it lies: a reason that names a round but no criterion is still vague. A reason can also be specific and still impossible to contest. In one Stripe case shared on Blind, the hiring committee down-leveled a candidate "for my 'number of years of experience'" despite positive feedback. That is a tenure rule, and no retake changes it. Score it 0.

A reason you cannot answer with new evidence is not a reason you can contest.

Competing loops or offers

What it proves: that the market has priced you at the higher level. A recruiter-side source calls a competing offer at the higher level the fastest way to reopen the level conversation.

When it lies: the offer may move pay but not level. On Google, one Blind commenter put it as "Yea level won't change." Score a live loop 1 only if it is past the screen and has a dated next step.

Pay gap

What it proves: what you give up each year at the lower level. Score it on total compensation, including equity, not base salary. The next section shows why the two differ so much.

Floor versus the lower band

What it proves: whether you can live with the lower level at all. Your floor is the lowest total package you would sign for this move. Set it before you see their number.

When it lies: the lower band's maximum often looks comfortable, but new hires are rarely placed near the maximum.

Path back

What it proves: how long you pay the gap. Price it as time at the lower level multiplied by the total pay gap.

When it lies: a verbal promise of a quick review, or a lower level that is terminal at that company. Terminal means there is no expectation that you move up from it.

How to price the band gap against your own floor

Price the down-level on total compensation, then check whether a realistic new-hire offer in the lower band still clears your floor. Base-salary bands understate tech level gaps by a wide margin, and band overlap makes the lower level look better than the offer you will actually get.

Start from the published norms, then distrust them

Salary structure surveys put the step between adjacent levels at roughly 5 to 15%, and the step grows at higher grades:

• Administrative and operative grades: 5 to 10% midpoint progression.
• Professional and management grades: 10 to 15%, which matches the market-based structures in the WorldatWork/Deloitte survey.
• Executive grades: 15 to 20%.

A secondary source puts the most common progression at 10 to 14%. Some employers use broadbands, which are very wide bands with 80% to 200% range spreads and no defined midpoints. If your employer uses broadbands, the midpoint math below does not apply, and you should ask directly where in the band your offer will sit.

The real gaps at five tech ladders

Median total compensation for adjacent US software engineering levels, taken from public career-ladder compilations that cite Levels.fyi as the pay source:

The gap and gap-times-time columns are derived. They ignore raises and refreshers, so treat them as order-of-magnitude figures. Higher up, the gaps get bigger: Exponent puts the Google senior-to-staff gap at approximately $180K per year. Every one of these gaps is well above the 10 to 15% base progression norm, and that difference is the main reason to score pay on total compensation.

Where the top of the lower band lands

Assuming a 50% spread, with the minimum at 80% and the maximum at 120% of midpoint, here is how the two bands meet:

Read it this way. At typical progressions, the lower band's maximum reaches or passes the upper band's midpoint. That sounds reassuring, but it only helps people who are already at the top of the lower band. Employers commonly place new hires at a compa-ratio of 0.80 to 0.95, so your real offer gap sits close to the full progression, not the overlap.

The floor check

1. Get the lower level's midpoint, in total compensation if possible.
2. Multiply it by 0.80 and by 0.95.
3. If 0.80 of the midpoint clears your floor, score 0. If only 0.95 clears it, score 1. If even 0.95 misses it, score 2.

Sources disagree on how much healthy bands overlap, with figures of 25 to 50%, 30 to 50%, and 50 to 60% all published. That disagreement is one more reason to price against placement rather than overlap.

How the scores map to Accept, Contest, or Walk

Add the three leverage scores and the three cost scores separately, so each total runs from 0 to 6. A total of 4 or more counts as high. Plot the two totals on the matrix, then apply the overrides below, which take precedence over the matrix.

Overrides

• Floor scored 2 and leverage below 4: Walk. If you can't win the contest and can't live with the outcome, there is nothing left to decide.
• Reason is a tenure rule: contest only on the strength of a competing offer. Retaking rounds won't move a years-of-experience rule.
• Recruiter and hiring manager give different reasons at offer stage: treat it as Walk unless competing loops scored 2.
• Salary or start date already discussed: stage scores 0, and your reply has to reopen the level explicitly.

The down-level procedure, from signal to sent reply

Run these eight steps in order, starting the moment the signal arrives. Most candidates can get through the first seven in two or three days. The eighth runs as long as a contest takes.

Sources disagree on one point of order. Exponent says to settle level before any pay discussion. One Blind poster who was down-leveled after a hiring manager call chose to finish the onsite and judge the pay first. Both can work. What doesn't work is talking about pay and then trying to reopen the level, so if you judge pay first, do it privately with the bands and keep it out of the conversation with the recruiter.

Use this message for steps 2 and 3. It works as phone notes or as a follow-up email that confirms the call in writing.

Thanks for telling me directly. Before I respond on level, I want to understand the decision properly.

1. Which rounds raised the level question, and which criteria in the [original level] ladder were not met?
2. Is this about interview performance, or about a fixed rule such as years of experience?
3. What is the full range for [lower level], minimum, midpoint, and maximum, for base and for total compensation? Where in that range would an offer sit?
4. Who owns the level decision from here, and what new evidence would they consider?

I'd like to settle level first, before we discuss compensation or a start date.

The scope brief in step 8 is easier to write if you can see which ladder criteria your history already covers. Refolk scores how well your record fits a specific posting. Run it against the original level's posting before you write the brief, and you can see which criterion your evidence supports and where it is thin.

What to send for each verdict

Each verdict has its own reply, and the stage changes how a Contest reply is worded. Every one of them settles level before pay.

ACCEPT
Thank you for the clear explanation. I'm open to [lower level]. Before we move to numbers, I'd like the path to [original level] in writing: the review date, the criteria, and who makes the decision. Once that's settled, I'd like to talk about where in the [lower level] range the offer will sit.

CONTEST (screen or mid-loop)
I applied for [original level], and my recent work has been at that scope. I'd like the remaining rounds, especially system design and behavioral, assessed against [original level]. If the panel concludes [lower level], I'll take that evidence seriously, but I'd like the chance to show it first.

CONTEST (debrief or offer)
I'd like to ask for a re-evaluation at [original level]. I'm happy to retake the [round name] round, and I've attached a one-page brief mapping my work to the [criterion] criterion. I also hold an offer at [equivalent level] elsewhere, with a decision date of [date]. Is the level itself open for discussion, or only compensation?

CONTEST ONCE, THEN ACCEPT
Before I respond on [lower level], I'd like to make one case for [original level]: [one sentence of evidence tied to the missed criterion]. If the decision stands after that, I'm prepared to move forward at [lower level] with the review date, criteria, and decision owner in writing.

WALK
Thank you for being straight with me. The [lower level] range doesn't meet the floor I've set for this move, so I'll withdraw now rather than take more of the team's time. If a [original level] role opens, I'd be glad to hear from you.

The last question in the debrief-stage Contest reply is there on purpose. It makes the recruiter say whether you are negotiating level or pay. Without it, a competing offer can get you more money at the same level, and that can feel like a win until the first promotion cycle.

If your competing-loops score is 0 and you would rather contest than walk, the only fix is more live loops at the higher level. Refolk tailors your resume to each higher-level posting you apply to, which saves real time when you need those loops moving within the window of one decision.

How down-level scoring goes wrong

Most bad down-level decisions come from misreading one dimension, not from skipping the framework. Each failure below has a cheap check. Run the check before you trust the score.

1. Misreading the stage. A late signal is not automatically a bait-and-switch. The real red flag is a downgrade that first appears in the written offer, combined with an explanation that keeps shifting. Check: ask the recruiter and the hiring manager the same question separately and see whether the answers match.
2. Specificity that isn't real. A reason that names a round but no criterion only sounds specific. A years-of-experience rule, like the Stripe case, is specific but has nothing to do with performance, so a retake can't fix it. The AWS L6 to L5 case shared on Blind, where the candidate would "need to be 'mentored' for maintaining code", sits in between: it names a gap but not the ladder line it maps to. Check: ask for the exact criterion in the ladder that was missed.
3. Pricing on base salary. Base progression of 10 to 15% badly understates level gaps of 24 to 43% in total pay. Check: compare total compensation, including equity, for both bands before you score the pay dimension.
4. False comfort from overlap. The lower band's maximum can sit at or above the upper midpoint, but new hires land at 0.80 to 0.95 of midpoint. Check: ask where in the range the offer will sit, and score the floor against that figure, not the band maximum.
5. A competing offer that doesn't carry over. A higher-level offer may move pay and leave the level alone. Commenters report that Google may match a higher-level offer's pay without changing level. Check: ask outright whether the level or only the pay is up for discussion.
6. The six-month review promise. It sounds like a short path back, and it is often nothing. Check: get the date, criteria, and decision owner in writing, or score the path back as 2.
7. Down-leveled into a terminal level. At Google, L4 and above are listed as terminal, meaning there is no expectation to move up. At Stripe, the candidate noted that L2 is not terminal. Landing in a terminal level can stretch the time back. Check: look up the company's ladder and confirm whether the lower level is terminal before you score the path back.

One more misread is worth naming: treating the debrief as a negotiation. A recruiter-side source is explicit that committee re-evaluation requires new evidence, not a verbal argument. Going back with a better-worded version of the same case uses up goodwill and changes nothing. Go back with a retake offer, a scope brief, or a written competing offer.

What this framework cannot tell you, and when to re-score

This framework turns the call into a score, but several of its inputs are thin, and you should know which ones before you lean on the verdict. Re-score at every new stage of the loop, because stage and competing loops both change as the process moves.

Where the evidence is thin

• Success rates for contesting are not published. The routes are documented: retaking a round, talking to the hiring manager, escalating to a VP, and, at Google, a hiring manager resubmitting a packet to the committee after team match. How often any of them works is not established.
• The 10 to 15% down-level frequency is one Bar Raiser's estimate. Don't treat it as a base rate.
• Time-at-level figures come from third-party compilations, not the employers' own published ladders. Google promotes twice yearly, LinkedIn's minimum eligibility is one year at level, one generic ladder puts junior-to-senior at 18 to 24 months, and Qualcomm's standard is approximately three years per level. Check the ladder for your target company with someone who works there.
• Promotion placement varies. Workday targets the 75th percentile of the next level's range for promoted engineers. That makes the gap-times-time figure an overestimate there and possibly an underestimate elsewhere.

Disclosed ranges cover less than the law implies

Pay-transparency rules apply to the advertised role. In the US, a minority of states require a salary range, many of them in the posting itself. California defines a posted pay scale as a good-faith estimate of expected pay. The EU directive requires a range in the posting or before the first interview, but member states have been slow to transpose it, and one count found only 15.8% of German postings showing salary. No source shows any duty to disclose the lower level's band when you are re-leveled mid-loop. Refolk's index holds 15,313 Senior Software Engineers in Germany, 10 of 25 sampled in Berlin, and most of them will have to ask for both bands themselves. These rules keep changing, so check a current state-by-state tracker and an EU transposition tracker for your location before you assume a range has to be disclosed.

After you send the reply, log the date and plan to re-score at the next contact. A Contest verdict at the screen can become an Accept after a strong system design round. A Walk can become a Contest the day a higher-level offer arrives in writing.