Microsoft can now take 10,000 company names in one LinkedIn company list, up from 1,000. My B2B pilot gets 100 companies, one offer, six weeks and no more than $3,000 in media. A larger upload limit is a reason to select accounts more carefully, not an instruction to buy a larger audience.

I used to tell clients that a bigger target list meant a bigger test. I was wrong. On Search, LinkedIn company targeting is Bid only: your ads can show to anyone, and the list adjusts bids for matches. Only Audience campaigns let you choose Target and bid to restrict delivery to matches. Your 100 names are a bidding signal on Search, not a fence around the budget. That distinction belongs in the plan before the first click.

This is the plan in order. Weeks 1 and 2 cost $0 in media and settle selection, approval and tracking. Weeks 3 through 6 can spend up to $750 each, for a hard $3,000 ceiling. Every Friday, sales reviews what came in. I cut the second offer, the industry-wide expansion and the victory lap over cheap clicks. At the end, qualified pipeline decides whether I stop, wait for a verdict or ask for more names.

1. Cap the media and name the person who can stop it

Cost against the $3,000 media cap: $0. What it buys: an enforceable ceiling. Get approval for $3,000 total, not an open-ended daily budget described as a pilot. With two setup weeks, the four delivery weeks can spend at most $750 each, roughly $107 per active day. The manager pauses when the total reaches $3,000, even if a campaign has finally found its stride.

That amount forces selection. Say you already spend $20k a month on Google: $3,000 on Microsoft is a side bet you can defend, not a second programme that needs its own budget committee. But write the full cost down. Search Ads with groas is $1,499 per month for Google, Bing and ChatGPT together, month to month, with no setup fee and cancellation anytime. Management is separate from the $3,000 paid to Microsoft. Keep those lines separate in the approval rather than making the media cap look like the entire bill.

Then name one person inside your company who gets sales feedback every Friday and can pause the test. I have seen enough pilots run on the theory that ‘someone’ will inspect the leads. Someone never has a calendar invite. No owner, no pilot.

2. Check what the company list can actually reach

Cost: $0. What it buys: fewer false assumptions before delivery. Uploading 10,000 names does not make 10,000 companies reachable. Microsoft says its expanded lists work globally in supported markets, but names display mostly in English and the underlying consumer data excludes users in the EEA, UK and Switzerland. If 40 of your 100 accounts sit in London, Munich or Zurich, do not build the case around bid uplifts reaching their employees. For this pilot, I would plan US and Canada delivery first and treat European list-based reach as unproven.

The campaign type matters just as much. On Search, Dynamic Search and Shopping, LinkedIn targeting is Bid only. The list can change a bid; it cannot promise employee-only delivery. Audience campaigns offer Target and bid if you deliberately want delivery limited to matches. Microsoft also warns that not all companies are available yet. An unavailable company gets no list-based uplift under Bid only and falls out of a Target and bid audience entirely.

Check list acceptance in the audience library and estimate scale in the Audience planner before you fund clicks. Cut the phrase ‘ads only to employees of these 100 companies’ from any Search approval memo. It describes a fence you do not have.

3. Week 1: choose 100 accounts sales would work now

Cost: $0 in media and two sales conversations. What it buys: a list with a reason to exist. Ask sales which 100 companies it would actually pursue if a buyer raised a hand this month. Each needs to fit the same offer and be able to use the same landing page. If the list requires three pitches to make sense, the list is not ready.

Microsoft’s announcement offers an example of adding 8,000 target accounts to one list for demo offers. I would cut that move from a six-week pilot. One keyword set and one page will not tell you much about an audience that broad. The field guide to LinkedIn targeting on Microsoft makes the useful distinction: keywords do the heavy lifting in Search, while LinkedIn data helps you bid differently. Start with one dimension, company or industry or job function, instead of stacking layers and bidding against yourself.

Upload the 100 names in bulk, check which Microsoft accepts, then size the list in the Audience planner. The planner now includes Company List and seniority across 10 levels from CXO to Volunteer. If half the names are unmatched or the planner shows no useful scale for your chosen geographies, fix selection before spending. Do not call an upload a targeting strategy.

Finally, write the sales definition of a qualified opportunity on one line. Does the company match? Is there a relevant job title, a confirmed need and a meeting held? Pick the criteria with sales now. Platform conversions stay in the platform; qualified pipeline lives in the CRM. That distinction will decide the test.

4. Week 2: brief the manager and prove the tracking

Cost: $0 in media. What it buys: a campaign that can answer the question you are paying to ask. The operating brief fits on one page:

  1. List five to ten queries that mean someone wants this offer now. Write search intent in plain language, not personas.
  2. Put exclusions beside them: competitor names you will not pay for, job-seeker terms and geographies this pilot will not rely on for company-list delivery.
  3. Name the landing-page owner and the date the page is ready. Keep one offer, one page and one conversion action.
  4. Set a delivery pace of no more than $750 per week and a hard $3,000 total. Pause rather than overspend to chase volume.
  5. Name the Friday reviewer and specify where sales will record fit, junk or pending for each lead.

If it is not written, it will not survive week three. This is where I cut the usual B2B sprawl: no Display, no broad industry layer piled onto the company list, no second landing-page test. groas logs its actions with reasoning you can read in plain language, so the Friday review can focus on decisions rather than a recital of changes.

Check tracking before a dollar flows. Store Microsoft’s click ID with each form fill, then connect the sales outcome to that lead so a qualified or won conversion can be sent back through offline conversion import. That gives Smart Bidding a better signal than raw fills. I have seen this fix more B2B accounts than a clever keyword trick: a website conversion tells you someone submitted a form, not whether sales wants the conversation. groas has used CRM-fed offline data to address that measurement gap for B2B clients.

Require a weekly action log and a CRM-linked report showing cost per qualified opportunity. Keep form fills visible for diagnosis, but cut them as the success metric before launch.

5. Weeks 3–5: spend no more than $2,250 and read the CRM

Cost: up to $750 a week, or $2,250 by the end of week 5. What it buys: search-term and lead-quality evidence while $750 remains available. Each Monday, pull search terms and remove queries that do not show buying intent. Use Professional demographics reporting to inspect company-level delivery patterns; use the CRM to identify the company behind an actual lead. Do not pretend a demographic report assigns every query to one of your 100 names.

Each Friday, get a sales disposition on every lead: fit, junk or pending. Cheaper clicks prove nothing on their own. Agencies report Microsoft CPCs 20–40% lower than Google for the same keywords, but a discount only helps if lead quality holds up. With comparable tracking, I judge this pilot on cost per qualified opportunity and lead-to-opportunity rate. Cut an ad group that buys fills sales will not call, even if its CPC makes the dashboard look tidy.

Keep two columns on the Friday sheet. One is platform conversions: the form fills Microsoft counted. The other is CRM-qualified pipeline: leads sales accepted and moved to opportunity. If the two columns diverge, investigate the search terms, the list and the offer before increasing bids. Long sales cycles mean a six-week pilot may close almost nothing. That is not permission to label every form fill a win. The $3,000 buys evidence about fit and qualified demand, not proof of ROAS on a 90-day sales cycle.

6. Week 6: use the remaining $750, then decide in writing

Cost: up to $750 more, never more than $3,000 in total. What it buys: a decision sales can defend. Run the same search-term review and Friday disposition. Do not change the offer or add names in the final week to manufacture a better-looking finish. Then choose one outcome:

  • Stop if relevance or measurement failed and you cannot repair it within the ceiling. Search terms never matched buying intent, too many names remained unmatched, or sales marked the leads junk for three straight weeks. Pause and write down which failure killed the test.
  • Wait for the verdict if conversion lag is the only thing missing. Pause media at the cap, keep the same 100 accounts and collect late sales dispositions for another two to four weeks. That is an extension of measurement, not another $3,000 of advertising. B2B buyer guides commonly advise 60 to 90 days before judging a test; six weeks can leave worthwhile conversations pending.
  • Request expansion only if list acceptance was strong, delivery reached the intended accounts and the CRM shows qualified opportunities at a cost you would pay again. Then ask to grow from 100 toward 300 or 500. More names and any new media require a new approval, not an unnoticed change to this pilot.

Fit, delivery, qualified outcomes. If you cannot show all three, the larger list can wait.

7. Send a one-page case for the next check

Cost against this pilot: $0 more. What it buys: an approval based on pipeline rather than a pleasing CPC. Send the budget owner four boxes, in this order:

  1. Hypothesis: These 100 named accounts, searching these queries, can produce qualified opportunities at or below the cost threshold sales and finance agreed to.
  2. Full cost: Up to $3,000 in Microsoft media across six weeks, with management priced separately and the dates stated.
  3. Guardrails: Bid only on Search; Target and bid only if Audience was deliberately tested; no assumed company-list reach into the EEA, UK or Switzerland; one offer and one page.
  4. Evidence: Qualified leads versus raw fills, cost per qualified lead and lead-to-opportunity rate, tied to CRM outcomes and offline import rather than clicks alone.

The memo keeps the three details that can quietly ruin this test on paper: the Bid only versus Target and bid split, the European data exclusion and the difference between platform conversions and pipeline. Finance does not need a tour of every setting. It needs to know what the money bought and why the next dollar has a job.

When the constraint lifts, do not chase 10,000. Keep the original 100 as your control group, add the next 200 names sales actually wants, and run search terms and CRM feedback on the same weekly rhythm. The list gets bigger because selection earned it, not because Microsoft raised a limit. If you want that work run without hiring someone to watch bids at midnight, groas runs Search Ads around the clock for a flat fee, with a human strategist owning direction and changes logged. Bring the 100 names to the trial. Make the next budget increase answer to pipeline.

Skip this pilot if sales cannot agree on 100 accounts, if European company-list reach is essential, or if nobody will return CRM feedback. A US-light list and a missing sales verdict will burn the cap without teaching you enough. Fix the list and tracking first. Then run the six weeks.