Sales prospecting techniques that still work in 2026

A prospecting technique only works in 2026 if it clears three filters at the same time: a person's attention, the platform's rate limits, and the receiving system's abuse controls. Most advice optimises for the first and ignores the other two, which is why tactics that read well fail in practice — the message never arrives, or the account gets restricted before the tactic has a chance. The techniques below clear all three. Each one is written as a procedure with the decision rule at each fork, because the difference between these working and not working is almost always execution detail rather than concept.

Every technique has to survive three filters

Think of a prospecting message as passing through three gates in sequence, in this order.

Filter one: the abuse control. A receiving mail server, a phone carrier or a social platform decides whether to accept the message at all. This decision is made by machines using signals that have nothing to do with your copy — authentication, sending history, complaint rate, recipient engagement. Google's bulk sender requirements have been enforceable since 1 February 2024 for senders above 5,000 messages a day to Gmail addresses, requiring SPF, DKIM and DMARC on the sending domain, and Google's guidance asks senders to keep the spam rate reported in Postmaster Tools below 0.10% and to avoid ever reaching 0.30%.

Filter two: the platform cap. A rate limit you did not set and often cannot see. LinkedIn is the sharpest example: it applies invitation limits to every member, Basic and Premium, and its own help documentation does not publish the number, with LinkedIn Support stating it cannot disclose the type or reason for a restriction on an account. You are operating against a ceiling whose height is deliberately undisclosed.

Filter three: the human. Only now does copy matter.

The ordering is the point. A technique that wins filter three and loses filter one produces nothing, and produces it invisibly, because the metric that would have told you is delivery rather than reply. Diagnose in filter order, always: authentication and complaint rate first, caps second, message last.

Trigger-based prospecting, with the latency rule that makes it work

Trigger-based outreach means contacting a company because something specific changed, and citing that change. It works because it answers the only question that matters in a cold message — why now.

The triggers worth building around, and where each is published:

• Funding. In the United States, a Form D notice of an exempt offering is filed on the SEC's EDGAR system, and the SEC requires a company to file it "within 15 days after the first sale of securities in the offering". That deadline is also the latency: a round is already up to a fortnight old on the day the filing appears, and press coverage lags further.
• Hiring. Open roles are a public statement of where budget went. Ten open sales roles and ten open support roles are opposite signals and should produce opposite messages.
• Leadership change. A new head of a function has a mandate and a window in which changing tools is expected rather than disruptive. This is the highest-value trigger in most B2B categories and it is visible on public profiles.
• Expansion or a new office. A registry filing or a job posting with a new location attached.
• A public complaint about a competitor. Rarer, and the strongest of all when you find one.

The latency rule. Act inside the trigger's own half-life. Leadership changes stay relevant for about a quarter, because the mandate persists. Funding stays relevant for perhaps two months, because the spending happens fast. Hiring signals decay when the role is filled. A trigger past its half-life is still useful as context — "I saw you opened a Pune office earlier this year" — but it can no longer be the reason for the message.

The decision rule at the fork. If the trigger changes what the prospect needs, lead with it. If it merely proves you did homework, bury it in the second sentence and lead with the problem. Leading with a stale trigger is the single most common way this technique reads as automated.

The 90-second research pass that produces one specific line

Personalisation is inserting a value into a template. Research is finding one fact that changes what you say. Recipients tell the difference immediately, and only the second one moves anything.

Run this pass with a timer. Ninety seconds per prospect, no exceptions — the discipline is what makes it survive contact with a list of two hundred.

1. Fifteen seconds — the company's own words. Open the homepage and read the top headline. What do they say they do, in their vocabulary? Use their noun, not your category's noun.
2. Twenty seconds — the careers page. What are they hiring for right now? This is the fastest available proxy for where money is going.
3. Twenty seconds — the person's profile. How long in role? Under six months is a different message from four years. What did they do before?
4. Twenty seconds — one recent public artefact. A post, a talk, a podcast, a press mention, a product changelog. Anything dated.
5. Fifteen seconds — write the line. One sentence connecting what you found to the problem you solve. If you cannot write it, that is a signal: the prospect may not fit, and moving on is the correct decision rather than a failure.

The test that keeps you honest. Read your line and ask whether it could be sent, unchanged, to a different company. If yes, it is a merge field wearing a sentence, and it will be read as one. Worked examples of lines that pass and fail this test are in cold email personalisation examples.

At ninety seconds each, two hundred prospects is five hours of research. That is a real cost, and it is the reason this technique is genuinely defensible: it does not scale to everyone, so the people who do it are competing against templates.

Sequencing channels under caps you do not set

Multichannel prospecting works because a second channel reaches someone who ignored the first, at no cost to the first. It fails when the caps are treated as a single budget rather than five separate ones.

Every channel has an independent ceiling, and they are not interchangeable. Autocloz's per-channel daily caps make the shape visible: on the free plan a workspace gets 1,000 emails, 100 calls, 50 LinkedIn actions, 250 SMS and 250 WhatsApp messages a day; Growth raises those to 2,500 email, 500 call, 150 LinkedIn, 1,500 SMS and 1,500 WhatsApp; Pro to 12,500 email and 400 LinkedIn; Scale to 100,000 email and 750 LinkedIn. Notice which number stays small at every tier. LinkedIn caps are low deliberately, because the platform's own undisclosed limits bind long before a commercial limit does, and an account restriction costs more than a week of touches.

The sequencing rule. Order channels by recoverability, cheapest to recover from first. Email is recoverable — a bad email costs a small amount of reputation. LinkedIn is expensive to recover from, because a restriction affects the account rather than the campaign. Calling is intermediate, and carries the tightest legal constraints. So: email first, LinkedIn as a low-frequency parallel track, calling on the accounts that showed a signal.

The gap rule. Leave enough space between two channels that the second does not arrive as a second copy of the first. Same-day email and LinkedIn on a stranger reads as surveillance. Two to three working days apart, with different content in each, reads as persistence.

The one thing to get right before adding a channel. Replies from every channel need to arrive in one place. A LinkedIn reply nobody reads for four days is worse than a LinkedIn message never sent. Running all five channels from one workspace is the mechanical half of that; naming who watches the queue is the other half. Ready-made structures for the timing are in multichannel sequence templates.

Autocloz's free plan covers 5 users and 10 mailboxes with warmup and DMARC monitoring built in, and every channel is included on every tier with only the daily volume changing — start free and run a two-channel sequence before you pay for anything.

Calling still works, and the reason is arithmetic rather than nostalgia

Cold calling gets dismissed as dead by people measuring the wrong thing. It has a much worse cost per attempt than email and a much better response rate per contact, so whether it works is a question about your deal size, not about calling.

Here is the arithmetic, with illustrative figures you should replace with your own. A rep making 60 dials a day reaches perhaps 6 people and books perhaps 1 meeting. That is a day of a rep's time for one meeting. If a meeting is worth ₹15,000 in expected pipeline value, calling loses to almost anything. If a meeting is worth ₹3,00,000, a day of rep time for one is an excellent trade and email volume is the wrong instrument entirely.

The threshold, written as a rule: call when the expected value of a meeting exceeds roughly a day of fully loaded rep cost divided by your meetings-per-day rate. Below it, use calling only on accounts that have already shown a signal.

Two constraints to build in before the first dial. In the United States, 47 CFR 64.1200(c)(1) prohibits initiating a telephone solicitation to a residential telephone subscriber before 8 a.m. or after 9 p.m. local time at the called party's location, and the same rule covers numbers on the national do-not-call registry — a B2B direct dial to a personal mobile is frequently a residential line, so the exemption most people assume is thinner than they think. Separately, the time zone your dialler thinks it is in is not necessarily the recipient's; Autocloz resolves quiet hours from the campaign override, then the workspace default, then the sending account's timezone, so a workspace clock set to one region does not protect a prospect in another. Set the override per campaign when you sell across time zones. The dialler enforces whatever window you configure, not a window it infers from the number.

Disqualifying is a technique, not an admission of failure

The least-used prospecting technique is deciding not to prospect someone. It behaves like a technique because it produces the same output — more meetings — through a different mechanism.

Write three disqualifiers before you start any list, and make them observable rather than intuitive:

• A structural mismatch. They are on a platform you cannot integrate with, they are below your minimum size, they are in a market you cannot service.
• A timing mismatch. They signed a competitor's contract in the last six months. Note the renewal date, put them in a re-contact list, and stop.
• An authority mismatch. The person you can reach cannot start a purchase and cannot introduce you to someone who can.

Then remove them from the list rather than leaving them in it at a lower priority, because a low-priority row still consumes a touch when the sequence runs. The compounding effect is what makes this a technique: an hour of disqualification at the start removes hundreds of low-probability touches over the following month, and those touches were also costing you complaint-rate headroom against filter one.

The techniques that stopped working, and what killed each

Being specific about what died, and why, is more useful than a list of what to do.

• Buying a large list and sending to all of it. Killed by filter one. Complaint-rate thresholds are proportional, so a list that is 90% irrelevant produces complaints faster than volume produces replies.
• The "quick question" subject line. Killed by saturation. It was a pattern-interrupt when it was rare and is a pattern-match now.
• Automated connect-then-pitch on LinkedIn. Killed by filter two, and by the platform's User Agreement, which prohibits bots and unauthorised automated methods for adding contacts or sending messages at section 8.2.13. The restriction lands on the account, not the campaign.
• Open-rate optimisation. Killed by measurement itself. Mail privacy features load images automatically, so open counts include events no human caused; how open tracking actually works sets out what the number now contains.
• Sending at the mythical best hour. Killed by evidence, or rather by its absence — the effect is real but small compared with list quality and offer, which is the argument in the best time to send cold emails. The send-time calculator is worth using for time-zone hygiene rather than for a magic hour.

What changes between one prospector and ten

At one person, technique is everything and process is optional. At ten, that inverts.

Research quality becomes bimodal. Your two best reps do the ninety-second pass; the other eight do fifteen seconds and a merge field. The fix is not exhortation. Make the pass a required field: the sequence cannot enrol a lead until a research line exists on the record.

The caps become shared and invisible. A per-workspace daily cap divided by ten is not a per-rep allocation, and the tenth rep to press send is the one who finds out. Split it explicitly.

Duplicate contact becomes certain. Two reps working the same account from different lists will both send. Deduplicate by account, not by contact, and give each account a single owner.

Disqualification stops happening. Reps whose quota is measured in activity have no incentive to remove rows. If you want disqualification, measure meetings rather than touches. Which sources feed the list in the first place is covered in where to find B2B leads, and the platform-level comparison of running this on a per-seat tool versus a flat one is on the Autocloz and Outreach comparison.

What no prospecting technique will do for you

Some honest limits, because a technique list that promises outcomes is selling something.

No technique fixes an offer nobody wants. If a well-researched message to a well-chosen person gets no reply, the problem is upstream of prospecting and no amount of sequencing reaches it.

No software makes a platform-limit violation safe. Autocloz's LinkedIn caps are conservative by design, and they still cannot know what LinkedIn's undisclosed per-account ceiling is on your account today. The account risk is yours.

Autocloz will not decide which channel your buyers respond on. It will run five and report what happened; picking is a judgement about your market that the tool has no access to.

And it will not tell you whether a trigger is real. A funding filing is a fact; whether that company is now more likely to buy from you is an inference, and inferences are worth testing on fifty prospects before they become the basis of a quarter's plan.