// Recruitment Technology, Recruitment Strategies

Nine Things Recruiters Only Learn After Their First Campaign Goes Out

The bottom line

First campaigns rarely fail on copy. They fail on plumbing. Send caps split one launch into batches. Merge fields break in front of four hundred candidates. Opt-outs sit in a tool your database never hears from, and two consultants email the same person before lunch.

Each of those has a guardrail. Every one takes minutes to set up front and hours to unpick afterward. Treat the first campaign as a systems test rather than a launch.

Why the first campaign is really a test of your systems

Seventy-one agencies have told us a version of the same week-one story. The sequence was written, approved and loaded. Then it met the database.

Candidate outreach at scale exposes everything a desk has quietly tolerated. Duplicate records. Missing contact details. Opt-outs tracked in one person’s notebook. Three consultants who each believe they own the same contact. None of that appears in a preview pane.

The first campaign doubles as an honest audit of your candidate database. The result is rarely the one anyone hoped for.

What follows is the list most teams assemble the hard way. The problems tend to arrive in this order, and each one has a guardrail that prevents it.

Nine candidate outreach lessons agencies learn in week one

1. Your first campaign will not go out as one campaign
2. Merge fields break in public, not in testing
3. An unsubscribe only counts when the database hears about it
4. Part of your list has no email address at all
5. Two consultants, one candidate, the same morning
6. Replies land in an inbox, and the CRM never finds out
7. Open rates are the weakest number in the report
8. Nobody tells the sequence that the candidate was placed
9. Deliverability was decided before you wrote the first line

1. Your first campaign will not go out as one campaign

Send caps decide your launch date, not your plan. A segment of 600 candidates will not leave in one morning. Mailboxes and sequencers both throttle volume per user, per domain and per day.

Volume is rarely the real damage. Sequencing is. Step two reaches your first batch while the last batch still waits on step one. Replies then arrive in waves, follow-up calls stack unevenly, and one report ends up describing several different campaigns.

New sending domains tighten the window further. Volume has to be built up gradually rather than switched on. Our guide to email deliverability in recruitment covers the warm-up side in detail.

There is a desk-level version of the same problem. A consultant loads 400 contacts on Monday, then books a client day on Wednesday. The batch that sends while they are off the desk gets no reply handling at all.

Timing is part of the same decision. A sequence set to fire on a fixed schedule will reach somebody at three in the morning their time. Candidates read that correctly as volume, and a message built around a tight send window should respect the recipient’s working day.

Channel limits differ, and a multi-channel sequence inherits the tightest one. Email ceilings come from your provider. InMail runs on a monthly credit allocation. Connection requests carry their own daily limit. Phone calls are capped only by the hours in the day. Plan the campaign around the narrowest channel in it, then let the others run ahead.

The guardrail: set a daily ceiling per mailbox yourself. Size the segment to the send window rather than to the longlist. Split large lists into waves with their own start dates, and keep each wave small enough for one person to work the replies that day.

2. Merge fields break in public, not in testing

A merge field is only as reliable as the field behind it. Test records are clean, so the preview looks perfect. The live segment then introduces records created in 2019 by a consultant who left in 2021.

The failures are familiar. A greeting with no name after it. A raw token printed where the first name belongs. A current employer the candidate left two roles ago, quoted back to them as research. A city that was right three jobs ago.

None of that reads as a formatting slip to the person receiving it. It tells them exactly how the message was produced. That is the opposite of what the sequence was written to do.

Personalization built on a single name token was thin anyway. Personalizing candidate outreach past the first name is what separates a sequence from a mail merge. It depends on the record holding something worth referencing.

Senior candidates are the least forgiving audience for this. A broken token in a message about a board-level role ends the conversation at line one.

A pre-send field audit takes one filtered view of the segment. Show every record where first name, company or job title is empty. Most teams are surprised by the count the first time, and they never skip the check again.

The guardrail: require a fallback value on every token. Exclude records missing the fields your copy depends on. Preview against your oldest records rather than your newest, and have a second person read the rendered output before anything sends.

3. An unsubscribe only counts when the database hears about it

Opt-outs recorded in the sequencer and nowhere else are the most expensive gap on this list. A candidate unsubscribes. Three weeks later another consultant sends a spec CV from the CRM, because the CRM was never told.

That is a compliance exposure rather than an etiquette problem. The FTC requires senders to honor an opt-out request within 10 business days. Penalties run to $53,088 for each separate violating email, and the law makes no exception for business-to-business mail.

Suppression works when the opt-out and the candidate record are the same object. Atlas, an AI-powered recruitment platform that uses agentic AI to remove admin from recruiting workflows, runs multi-step, multi-channel outreach campaigns against records in the CRM itself. Campaigns draw from the live database rather than from an export.

Exported lists are where many of these problems begin. A CSV pulled on Monday knows nothing about Tuesday’s opt-out, Wednesday’s placement or Thursday’s complaint.

Opt-outs also need scope. A candidate who asks to stop hearing about contract roles in one city has not withdrawn from everything. Recording that nuance on the record keeps good people reachable, and it keeps the consultant who finds them next year out of trouble.

The guardrail: hold opt-out status on the contact record. Show it to every consultant who opens that record, and check it on every send. That includes the sends nobody calls campaigns, like one-off spec CVs and BD mail.

4. Part of your list has no email address at all

Every longlist shrinks at launch. Records with no email drop out. So do dead work addresses and addresses at companies the candidate left two years ago. The campaign then reports results against whatever survived.

That shrinkage is predictable, which makes it manageable. Data decay removes your easiest placements by turning known candidates into unreachable ones. It works fastest on the senior people a retained search needs most.

Keeping records current is the structural answer. Atlas tracks role and company changes across your network and updates candidate records automatically when someone moves. A job change also hands the desk a reason to make contact.

Candidates with no email are not lost, only mis-routed. They belong on a call list or in a LinkedIn InMail sequence. Leaving them in an email segment turns them into silent failures.

The count itself is management information. Teams that track reachable records by segment learn which parts of the database deserve enrichment budget. They also learn which parts should be retired rather than remailed every quarter.

The guardrail: count reachable records before you launch, not after. Report against the list you intended to reach. Route unreachable candidates to phone and social touches the same week, rather than to a tidy-up that never happens.

5. Two consultants, one candidate, the same morning

Nothing undoes a careful sequence faster than a colleague’s sequence landing twenty minutes later. The candidate notices. So does the client, when one agency puts the same person on two shortlists.

Duplicate records cause most of this. Desk boundaries cause the rest. A contact sourced twice exists twice, so a suppression rule applied to one copy does nothing for the other.

Consolidated contact data treats the underlying condition. Atlas connects to Gmail and Outlook and pulls every person your team has emailed into one CRM. That removes the private spreadsheets and personal address books where duplicates breed.

The commercial cost is larger than the embarrassment. A candidate pitched two roles in one morning reads the agency as disorganized. Candidate control is hard to win back once that impression lands.

Spec sends are where this bites hardest. Two consultants pitching the same candidate into the same account turns a strength into a credibility problem. Clients notice, and they tend to raise it at renewal rather than at the time.

The guardrail: dedupe before launch and suppress anyone already live in another campaign. Set a contact frequency cap across the agency rather than per desk. Settle record ownership before the campaign runs, not after the complaint arrives.

6. Replies land in an inbox, and the CRM never finds out

Replies go to whoever sits in the From field. Campaign results scatter across individual mailboxes within hours of launch.

Several failures follow. A reply that does not stop the sequence means a candidate gets step three after saying yes. A reply nobody logs means the next consultant opens a cold record on a warm candidate. An out of office counted as engagement inflates the whole report. Interest sent to one mailbox never reaches the person who owns the role.

Automatic capture is the fix, and it has to run in both directions. Atlas updates a candidate’s status automatically when they show interest by email or on LinkedIn. The reply changes the record instead of sitting in a thread nobody else can see.

This is also where campaign reporting drifts away from reality. The sequencer counts sends and opens. The desk counts conversations. Only one of those pays.

Cover matters too. Replies to a consultant on annual leave need a named second pair of hands, agreed before the campaign launches. A strong candidate who waits four days for an answer has usually started a process with someone else by then.

Shared mailboxes do not solve this either. A campaign sent from a generic address collects replies that nobody owns. Candidate control then dies in a queue, usually while the candidate is talking to someone faster.

The guardrail: require that a reply stops the sequence, syncs to the record and reaches the role owner the same day. Classify auto-replies separately, so a wall of out of office messages never reads as a strong week.

7. Open rates are the weakest number in the report

Open tracking stopped being evidence some time ago. Apple Mail downloads remote content privately in the background when a message arrives rather than when it is read. The pixel fires whether or not anyone looked.

Reply rate, meetings booked, CVs sent and interviews confirmed all survive that change. They are harder to inflate. They also map onto billings without interpretation.

Week one is when a team celebrates a 62% open rate. Week four is when someone asks how many of those opens became interviews.

Subject line testing suffers the same fate. Comparing two subject lines on opens tells you very little now. Comparing them on replies still works.

The metrics that hold up are the ones the desk already runs on. Time-to-send, submission-to-interview ratio, interviews booked and CVs sent per live role all tell you whether outreach is converting. Opens tell you a server somewhere fetched an image.

There is a second-order effect worth naming. Teams that optimize on opens start writing subject lines that earn opens. That is a different skill from writing messages that earn replies, and only one of them fills roles.

The guardrail: judge candidate outreach on replies and meetings. Treat opens as a weak secondary signal at best. Track the real measures by campaign and by consultant, because recruitment analytics at that level expose a failing sequence in days.

8. Nobody tells the sequence that the candidate was placed

Exit rules are the step most first campaigns skip. A candidate replies, interviews and gets placed. Step four of the original sequence still arrives, asking whether they are open to a new opportunity.

The same thing happens when a role is pulled. Outreach for a dead vacancy keeps running on schedule. The sequence has no idea the pipeline moved underneath it.

Pipeline stage should drive campaign membership. Where the sequencer and the pipeline are separate systems, someone has to remove people by hand. In a busy week nobody does.

This failure is quiet, which is what makes it costly. The candidate rarely complains. They conclude that the agency is not paying attention, and the next approach from that desk lands differently.

Placed candidates are the obvious case. The subtler one is a candidate already in process with a colleague on another role. A cold approach from the same agency tells them nobody internally is talking, which is an expensive thing to signal mid-process.

The guardrail: define exit conditions before launch. Cover a reply, a booked meeting, a placement, a closed role and an opt-out. Review live campaigns against the pipeline weekly, and pull anyone whose status has moved.

9. Deliverability was decided before you wrote the first line

Domain setup determines how much of a campaign lands, and those decisions sit upstream of the copy. Google asks senders to authenticate with SPF and DKIM, use a TLS connection, keep spam rates below 0.30% and support one-click unsubscribe on bulk marketing mail.

Agency mail runs into this because it looks like cold mail to a filter. Volume, templated structure, unfamiliar recipients and sudden bursts are the pattern those systems are tuned to catch.

Sending campaigns from your main company domain puts candidate mail and client mail on one reputation. A single badly targeted campaign then slows interview confirmations and invoices too.

Reputation also recovers slowly, which is the part teams underestimate. A week of enthusiasm can cost a quarter of deliverability on the domain your consultants use daily.

Scale changes the requirements as well. Once a team crosses into bulk sending volumes, providers apply a stricter standard than they do to one recruiter’s manual mail. The setup that worked for a single desk quietly stops clearing the bar.

The guardrail: authenticate every sending domain and monitor the spam complaint rate directly. Do not wait for reply rates to sag. Keep campaign traffic on a separate subdomain from the mail your consultants depend on.

What does running a candidate outreach campaign actually involve?

A candidate outreach campaign is a planned sequence of messages to a defined group of candidates or prospects. It runs across email, phone, LinkedIn and other channels, with rules for what happens when someone replies. The messages are the visible part. The rules decide whether it works.

Most agencies are already partway in. Atlas research across 1,000 agency recruiters found that 56.16% describe their technology setup as functional but fragmented, while 61.64% already use AI for candidate outreach or messaging. Those two findings explain the week-one pattern. The capability arrived before the connective tissue did.

Automated outreach fails at the joins rather than inside the steps. A sequencer handles the drafting and sending well. It knows nothing about the segment that fed it, the replies it produced or the pipeline it should answer to.

The same machinery carries candidate work and business development. A spec campaign built around one strong candidate uses the same segments, tokens, suppression rules and exit conditions as a sourcing sequence for a live role. Agencies that build the guardrails once get both, which is usually the argument that wins the internal case for doing it properly.

Ownership matters as much as tooling. Name one person responsible for replies, one for suppression, one for exit rules and one for reporting, even if that is the same person. Campaigns without an owner drift within two weeks, however good the sequence looked on launch day.

In practice, a campaign runs through five stages, and each one touches a different system.

1. Build the segment, then check how much of it is genuinely reachable
2. Draft the messages and decide what every merge field does when the data is missing
3. Stage the send against daily caps, with waves sized to the team that will work the replies
4. Handle replies, auto-replies and opt-outs so each one updates the candidate record
5. Close the loop by removing placed, pitched and withdrawn candidates from the running sequence

Which checks should you run before the next campaign goes out?

Check the plumbing in the order the campaign will hit it. Segment, fields, suppression list, send schedule, reply path, exit rules. Fifteen minutes there is worth more than another round of subject line edits.

Run the checks against live records rather than a clean sample. The messy records are the entire point of the exercise.

Running outreach inside the system of record changes output as well as hygiene. Ethea Solutions, a GTM search firm covering APAC, grew BD outreach by 500% and improved candidate response rates by 50% after moving its sequences into Atlas. One converted sequence produced a client that has billed more than 100k USD.

Then set the review before you need it. Look at it on day three for deliverability and merge errors. Day ten for reply quality. Day thirty for meetings and CVs sent, and day sixty for placements. Four dates in a calendar turn a campaign into a repeatable process rather than a one-off experiment.

Keep the checklist short enough that people use it. Nine questions get answered before a launch. A twenty-page process document gets skipped on the first busy Tuesday, which is exactly the week it was written for.

Nine questions worth answering before anyone presses send:

• How many records in this segment have a verified, reachable email address?
• Which tokens does the copy use, and what does each one render when the field is empty?
• Where does the opt-out list live, and does every consultant see it on the record?
• Who else in the business is currently messaging anyone on this list?
• What is the daily ceiling per mailbox, and how many working days will the full list take?
• Which domain is this sending from, and is it the one your client mail uses?
• What stops the sequence: a reply, a booked meeting, a placement or a closed role?
• Where does a reply land, and what updates the candidate record when it arrives?
• Which number will you judge this campaign on in 30 days?

The campaign is only as good as the system underneath it

Nine failures, one root cause. Each appears at a seam between two tools that do not share a record. That is why they arrive together in week one, and why better copy never removes them.

Agencies that solve this stop shopping for a sequencer. They run candidate outreach where the data already lives. The suppression list, the pipeline stage, the reply history and the campaign become one thing, so the guardrails hold without anyone remembering on a Friday afternoon.

Consolidation is the design principle behind Atlas. Candidate records, multi-touch campaigns, pipeline status and reporting sit in one AI-powered recruitment platform. Agentic AI keeps the records current, so your sequences act on what is true today.

Worth an hour before your next campaign goes out, if only to see what your current setup is quietly dropping.