How to Find Winnable GeM Tenders: A Bid Triage Framework

The Government e-Marketplace publishes over 4,000 new bids on a typical weekday. For any given seller, maybe 2% of them are actually winnable. The other 98% fail you on turnover, MSE preference, OEM authorization, geography, or a buyer who already has a preferred vendor in mind. If you're reading every bid document that matches your category, you're burning four to six hours a day to find the handful that deserve a real proposal.

GeM tender filtering is not a search problem. It's a triage problem. The search box on bidplus.gem.gov.in will happily return 300 results for "earthwork services" in a state you operate in. The real question — "which of these 300 can I realistically win, and in what order should I work them?" — is what separates firms with a 20% hit rate from firms with a 2% hit rate.

At Arched, we've processed thousands of GeM tenders against real seller credentials, and the patterns are consistent. This guide lays out the seven filters experienced bid managers use to decide what to bid on, a worked example applying them to a real GeM bid type, and a triage matrix you can use starting tomorrow. No fluff about how to register or where the login page is — if you're here, you already have portal access.

Why GeM's firehose is the core problem

GeM is designed to be open. Any registered seller in a category can view and download any bid in that category, which is great for transparency and terrible for focus. The platform does not tell you "this bid is a fit for your firm". It tells you "here are 300 bids in your category this week, good luck."

The three failure modes we see repeatedly:

• Seller chases volume — submits 40 bids/month with generic responses, wins 1–2, blames GeM. Real issue: no filtering, so 35 of those 40 were never winnable.
• Seller chases only the big numbers — filters by value, ignores mid-size bids with 3 competitors in favor of ₹10 crore bids with 47 competitors.
• Seller over-researches — spends 8 hours evaluating a bid before noticing the BOQ requires OEM authorization they don't have.

The fix is a disqualification-first workflow. Kill the no-gos in 90 seconds. Only spend real time on the bids that survive.

Filter 1 — Pre-qualification hard-stops (90-second screen)

Every GeM BOQ-based bid (BID mode) has an "Eligibility Criteria" section with MSE Exemption flags and experience requirements. Read this before anything else. If you fail here, nothing else about the bid matters.

The five hard-stops that kill most bids

Hard-stop	Where to find it	Skip if you fail?
Minimum average annual turnover	ATC / Tender document, section on financial criteria	Yes, always
Similar work experience (value + count)	ATC "Past Performance" or "Experience Criteria"	Yes, always
OEM authorization / MAF	BOQ, item-level specs	Yes, unless you can get MAF in time
Specific registrations (MSME, Udyam, GeM category, BIS)	ATC + seller profile match	Yes — retroactive fixes don't count
Geographic restriction (state-specific, local content)	ATC clause on "Local Supplier" or PPP-MII	Yes, unless you qualify as Class-I/II Local Supplier

The one-glance test: if your last three audited turnovers can't each clear the minimum (some buyers require each year, not just average), stop. If the "similar work" clause asks for three completed orders of ₹5 crore each in the last five years and your biggest was ₹3.2 crore, stop. These aren't negotiable.

On turnover specifically, pay attention to how the requirement is phrased — "average annual turnover of ₹X in the last 3 financial years" is different from "turnover of ₹X in each of the last 3 financial years." The second form is far stricter and we see sellers misread it constantly.

For a deeper treatment of how to pre-check these criteria systematically, see our guide on residual bid capacity — the concept extends cleanly to GeM eligibility.

Filter 2 — Economic fit (does the math work?)

You survive Filter 1. Now: is this bid worth winning? We've seen sellers celebrate winning a GeM order that cost them money after EMD lockup, performance security, and 180-day payment cycles.

The economic screen

Check these five numbers before you invest in proposal work:

• EMD as % of estimated value — GeM caps EMD at 2% and MSEs get exemption, but even 1% on a ₹2 crore order is ₹2 lakh locked up for the bid cycle. How many bids can you run in parallel at that rate?
• Performance security — typically 3% for GeM (reduced from 5–10% in most buyer categories). Locked for the entire contract + warranty period.
• Expected margin after logistics — GeM "delivery at consignee's location" means your freight to Itanagar is your problem, not the buyer's. Calculate landed cost, not factory-gate price.
• Payment terms — GeM mandates 10-day payment post-CRAC (Consignee Receipt and Acceptance Certificate) for most categories. In practice, state-government consignees stretch this. Check the buyer's past payment behavior if you can.
• Liquidated damages / penalty clauses — 0.5% per week of delay, capped at 10%, is standard. Tight delivery schedules + penalty = margin risk you should price in.

Rule of thumb: if expected gross margin after logistics is under 8%, skip. The probability of something going wrong (delivery delay, partial rejection, payment stretch) exceeds your margin cushion.

Filter 3 — Competitive density

A bid with perfect fit for your firm is worthless if 40 other firms also have perfect fit. GeM doesn't show you competitor count, but you can estimate it in under two minutes:

1. Check the category's BPA or Direct Purchase history on GeM analytics dashboards. High-velocity categories (stationery, basic IT hardware, standard PPE) routinely see 30+ bidders. Niche categories (specialized lab equipment, custom infrastructure) often see 3–8.
2. Check similar past bids from the same buyer — the "Bid Ranking" is published after award. If this buyer's last five similar tenders each had 20+ L1 candidates, expect the same.
3. Check seller count in the GeM category directly — if your category has 4,200 registered sellers, assume competition is high; if 38, assume it's niche.

A rough mental model:

Estimated bidders	Win probability (if you're fully qualified)
1–3	30–50% — worth full effort
4–10	10–25% — worth full effort
11–25	4–8% — worth a strong but not custom response
25+	Under 4% — bid only if L1 strategy is viable

High-density bids aren't automatically bad — they're just L1 (lowest price) games, which means your effort should go into cost modeling, not technical differentiation.

Filter 4 — Geographic and categorical fit

GeM implements PPP-MII (Public Procurement Policy for Make in India) and state-level local content preferences that materially change the competitive picture. A Class-I Local Supplier (50%+ local content) gets first right of refusal against a non-local L1 in many categories. If you qualify, it's a structural advantage; if you don't, you're bidding against someone who does.

Categorical fit also matters in ways that aren't obvious:

• Reserved categories — 25% of GeM procurement is reserved for MSEs, with further sub-reservations for SC/ST and women-owned MSEs. A non-MSE firm bidding in these is wasting EMD.
• State-run buyer preferences — some state departments use GeM but strongly prefer sellers registered in that state, through local clauses in the BOQ (e.g., "service delivery must be through a local office in [state]").
• Consignee location clusters — a bid with 14 consignee locations across five states demands a logistics profile many sellers can't deliver on. Read the full consignee table, not just the headline.

If you're in a reserved category and you don't qualify, skip. If you're in a category with an MII preference and you're not Class-I/II, your win rate is structurally lower than your competitors'.

Filter 5 — Red flags: the pre-wired bid

This is the filter experienced bid managers run on instinct. Some GeM bids are written to be won by a specific vendor. The buyer's hands are tied by procurement rules, so they publish the bid, but the specs, timelines, and eligibility criteria are constructed to leave exactly one viable bidder.

Signals of a pre-wired tender

• Over-specific technical specs that match exactly one OEM's product line. A generic laptop bid that specifies "14-inch anti-glare display, 3:2 aspect ratio, 400-nit brightness, carbon-fiber palm rest" is describing a ThinkPad X1 Carbon, not a category.
• Brand-agnostic language that's really brand-specific. "Or equivalent" is standard, but if every single spec matches one vendor's data sheet, "equivalent" is fiction.
• Unreasonably tight delivery schedules that only a vendor with pre-existing stock can meet.
• Narrow experience criteria matched by very few firms — "experience supplying [exact product] to [exact type of buyer] in [exact state] within the last 3 years" — narrows the field to a handful.
• Repeated cancellation-and-republish patterns on the same tender. When a buyer cancels on technicalities and republishes with slightly different specs, they're often tuning the specs until only the intended vendor qualifies.
• Very short bid submission window (7–10 days for a complex procurement) — the incumbent has been preparing; you haven't.

None of these are smoking guns individually. Two or three in combination should move a bid to the bottom of your priority list. You can still bid, but treat it as a "log an observation, submit a minimal-effort response" exercise, not a flagship pursuit.

Filter 6 — Buyer behavior and reliability

Not all GeM buyers are equal. Some departments pay on time, run clean processes, and reorder consistently. Others cancel 40% of their bids after opening, stretch payments past 90 days, or consistently award to the same vendor regardless of who wins L1.

Before you bid, spend three minutes on the buyer's GeM history:

• How many bids have they published in the last 12 months?
• What % were awarded vs cancelled?
• Are awards clustered to 2–3 vendors, or spread across many?
• Is there a "buyer reputation" signal in your own network (your sales rep has heard this buyer stretches payments, etc.)?

If the buyer has cancelled 6 of their last 10 bids in your category, your expected value is half what the face-value suggests. Price it in.

Filter 7 — Time budgeting and the bid triage matrix

The point of filters 1–6 is to decide how much effort to invest, not just go/no-go. Here's the matrix we recommend:

Tier	Criteria	Effort budget	Response type
A — Flagship	Perfect pre-qual, strong economics, ≤10 competitors, trustworthy buyer, no pre-wire flags	8–16 hours	Custom technical + sharp pricing
B — Standard	Pre-qual fit, workable economics, 11–25 competitors	2–4 hours	Template response, careful pricing
C — Volume	Pre-qual fit, commodity bid, 25+ competitors	30–60 minutes	L1 pricing exercise, minimal tech narrative
D — Skip	Any hard-stop fail, pre-wire flags stacked, negative economics	0	Do not bid

A realistic weekly target for a mid-sized firm: 2–3 Tier A bids, 5–8 Tier B, 10–15 Tier C, and 100+ Tier D filtered out before they touch your BD team. If the distribution looks different — say, 30 Tier A bids per week — your filters aren't strict enough.

Worked example: GeM bid for "Desktop Computers — 47 units"

Let's walk through a real bid type (edited for anonymity) to show the filters in action.

The bid:

• Buyer: State Directorate of Education, Maharashtra
• Quantity: 47 desktop computers with specific specs (i5-12400, 16GB RAM, 512GB NVMe, 23.8" monitor)
• Estimated value: ₹42 lakh
• EMD: ₹84,000 (2%)
• Delivery: 21 days from PO, consignee location in rural district HQ
• Eligibility: ₹2 crore average annual turnover last 3 FYs, 1 similar order of ₹20 lakh+ in last 3 years, OEM authorization required
• Payment: 10 days post-CRAC

Filter 1 (pre-qual): Our hypothetical seller has ₹3.2 crore turnover (pass), 3 similar orders of ₹25 lakh+ (pass), dealer authorization from Dell and Lenovo (pass, assuming one of them matches the spec). Survives.

Filter 2 (economics): 47 units × ₹89,000/unit = ₹41.8 lakh. Dealer cost ~₹78,000 → gross margin ~12% before logistics. Consignee is rural district HQ, add ₹15,000 logistics, on-site install + warranty support = another ₹20,000. Net margin ~10%. Survives, but tight.

Filter 3 (competition): Desktop procurement on GeM is high-density — typically 20–35 bidders. Tier C territory.

Filter 4 (geography/category): Maharashtra buyer, Class-II Local Supplier preference likely applies. If our seller qualifies, small edge; if not, structurally disadvantaged.

Filter 5 (red flags): Specs are generic, delivery window is reasonable for 47 units, no unusual experience criteria. Clean.

Filter 6 (buyer): State education directorate — typically reliable payer, but expect 20–40 day payment stretch beyond the 10-day CRAC mandate. Price it in.

Filter 7 (triage tier): Tier C. Run the L1 pricing exercise, submit a template response, don't over-invest. Expected value: ₹42 lakh × (1 / 25 competitors) × 10% margin = ₹16,800 in expected gross profit, against maybe 45 minutes of effort. Worth bidding.

That calculation is what a good bid manager does in their head in 90 seconds. The filters make it repeatable.

Common mistakes we see

• Skipping Filter 1 to read the BOQ first. The BOQ is seductive — lots of detail, feels productive. If you fail pre-qual, none of that detail matters.
• Ignoring payment terms in the margin calculation. A 10% margin at 90-day payment is very different from 10% at 10-day payment. Cost of capital is real.
• Treating every "or equivalent" spec as genuinely open. Look at the whole spec sheet. If 11 out of 12 specs match one OEM and one is "or equivalent," it's still a single-OEM bid.
• Bidding everything you qualify for. Every bid has an EMD lockup cost. At some point, you're better off concentrating on Tier A and passing on Tier C.
• Not tracking win/loss patterns by buyer. If you've lost your last 5 bids to the same buyer with the same L1 winner, something is structural. Stop bidding that buyer for 6 months.

For the broader operational layer around this — pipeline management, response workflow, proposal quality — see our guides on bid management best practices and what bid management actually involves.

Tools, automation, and what manual triage costs

Manual triage across the filters above takes a skilled bid manager 4–6 hours a day just to process the GeM firehose — before any actual proposal work. Multiply by five working days and you've spent the equivalent of a full FTE just to decide what not to bid on.

This is where tooling changes the equation. Good bid intelligence software applies pre-qualification filters automatically — turnover, similar work, registrations, OEM authorization, geography — and filters the 300 bids down to the 6–10 that survive Filter 1 before a human reads them. Government tender alert software adds the real-time layer so you don't miss a bid with a short window.

At Arched, we built our matcher specifically around this workflow: it reads every new GeM tender as it's published, runs it against each seller's actual credentials (financials, past work, registrations, OEM MAFs), and surfaces only the ones where the pre-qualification clears. The typical outcome for our users is manual triage dropping from 4–6 hours a day to under 30 minutes — not because AI is doing the bid, but because 90% of the bids were never winnable anyway and a good filter catches them before the human does.

If you're evaluating other procurement methods GeM uses alongside L1 — consulting procurement, for instance, uses QCBS (Quality and Cost-Based Selection) — the filtering logic changes somewhat, since quality weightage means L1 isn't the whole game.

FAQ

How many GeM tenders should I bid on per month to win consistently?

There's no universal answer, but the firms we see winning consistently run a concentrated strategy: 15–25 bids per month across Tier A and B, with a 12–20% win rate on Tier A and 4–8% on Tier B. Firms bidding 60+ per month with generic responses typically win at 1–3%. Quality and selectivity beat volume.

Can I tell if a GeM tender is pre-wired to a specific vendor before I invest time?

Not with 100% certainty, but the signals stack. Look for: over-specific technical specs matching one OEM's data sheet, narrow experience criteria that fit very few firms, unreasonably tight delivery windows, or a rebid pattern where the same tender was cancelled and republished with tweaked specs. Two or three of these together is a strong signal.

What's the minimum turnover GeM buyers typically require?

It varies by category and contract value, but a common pattern is average annual turnover of 50–100% of the estimated order value over the last 3 financial years. For a ₹1 crore order, expect ₹50 lakh–₹1 crore turnover requirement. MSE exemptions can reduce this, but not eliminate it.

Does GeM favor MSEs and how much does it matter?

Yes, significantly. 25% of GeM procurement is reserved for MSEs, with further sub-reservations (3% for women-owned, 4% for SC/ST-owned). MSEs also get EMD exemption, tender fee exemption, and a 15% price preference (they can match L1 even if they quoted up to 15% higher). If you qualify as MSE and haven't completed Udyam registration, you're leaving structural advantage on the table.

How do I check a GeM buyer's payment history before bidding?

GeM doesn't publish buyer payment behavior directly, but you can triangulate: check their bid volume (high-volume buyers usually have stable payment processes), check industry networks and WhatsApp groups for your product category (seller word-of-mouth is surprisingly accurate), and if possible call a past awardee listed in the buyer's recent award history. Build a buyer-quality database internally over time — it compounds.

What's the difference between Bidplus bids and direct purchase on GeM?

Bidplus hosts BID-mode tenders with full tender documents, EMD, BOQ, and a defined bid window — these are the "winnable tenders" this guide is about. Direct Purchase (under ₹25,000) and L1 Purchase (₹25,000–₹5 lakh) are different: the buyer picks directly from the catalog. Getting onto the catalog with competitive listing prices is its own strategy, and wins different kinds of revenue.

Should I bid on GeM tenders outside my home state?

It depends on the category. For products shipped in bulk (stationery, PPE, IT hardware), yes — logistics are manageable. For services that require on-site presence (AMC, installation, training), cross-state bids become expensive once you factor in travel, local presence, and state-specific compliance. Check the consignee locations and delivery terms carefully before assuming your pan-India registration means pan-India economics.

How long should I spend evaluating a single GeM tender?

For Tier A, up to 16 hours across proposal prep. For Tier C, 30–60 minutes including the pricing exercise. For the initial go/no-go triage — 90 seconds. If you're spending more than 90 seconds deciding whether a bid passes Filter 1 (pre-qualification), your filters aren't sharp enough. The whole point of this framework is to make "skip" decisions fast so you can spend real time on the bids that deserve it.