Skip to content

Makes

Microsoft Ads for B2B

Sets up LinkedIn profile bid adjustments by company, industry and job function on your Microsoft Ads search campaigns, capped by what a lead can afford.

Other tools

Add this skill to Claude

Claude takes skills as a ZIP file. Your download has started; if it didn't, use the button below.

  1. 1In Claude, open Customize, then Skills.
  2. 2Choose +, then Upload a skill, and pick microsoft-ads-b2b.zip.
  3. 3Ask for it by name, or let Claude use it when a task matches.

Skills need code execution turned on in Claude's settings. On a Team or Enterprise plan, an admin can share it with everyone.

Download the folder

Try it now, without installing

See the skill work on a made-up input before you add it. Nothing is installed, and you can swap in your own input.

Open in the Claude app

Opens a new chat in the Claude desktop app with the prompt filled in. Nothing is sent until you press enter.

Includes the whole SKILL.md and the sample input. Paste it into Claude, ChatGPT or any chat.

Influenced by
LinkedIn profile targeting
Needs
Microsoft Ads search campaigns with conversion tracking · Campaign clicks, conversions and average cost per click · The most a lead may cost
Works with
Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md
Version
1.0.0 · Last checked · License MIT
Maintained by
Sabih AhmedPaid media
SkillPaid media

Microsoft Ads for B2B

Example output

Set profile bid adjustments
SegmentAdjustmentCall
Software development+67%Bid up
Financial services-33%Bid down
Higher educationExcludeCan't buy

Sets up LinkedIn profile bid adjustments by company, industry and job function on your Microsoft Ads search campaigns, capped by what a lead can afford.

The skill

The whole skill, in plain text.

Nothing is hidden. Read it, change it to fit your team, or take the whole folder.

Scripts
Optional Python scripts for the counting. It works without them and says which way it ran.
Network
Opens web pages or your tools if your agent can. Works from pasted data if not.

3 test cases

microsoft-ads-b2b/SKILL.md

In the folder

  • microsoft-ads-b2b/SKILL.md
  • microsoft-ads-b2b/references/microsoft-facts.md
  • microsoft-ads-b2b/scripts/profile_bids.py
---name: microsoft-ads-b2bdescription: >-  Sets up LinkedIn profile bid adjustments (company, industry, job function) on a B2B SaaS company's Microsoft Ads  search campaigns: checks the prerequisites (conversion tracking, the Google import, market availability), picks  which dimension to start with, computes each adjustment from the campaign's own conversion rates, caps it so no  segment pays more per click than its leads can afford, checks how stacked adjustments multiply, and writes a bulk  plan the reader applies. Returns READY, READY WITH GAPS, NOT YET or BLOCKED. Use it after Microsoft Ads search  campaigns are live, or when importing them from Google. Not for search term cleanup (use search-terms-triage) or  for setting the cost per lead ceiling (use lead-value-math).license: MITcompatibility: >-  Works in Claude, ChatGPT, Codex, Cursor and any agent that reads SKILL.md. Uses a connected Microsoft Advertising  account if the agent has one, otherwise report exports the reader uploads. Python 3 is optional, for the  adjustment math.metadata:  author: SHAKEUP  version: "1.0.0"  stage: make  plugin: paid-media  homepage: https://www.shakeuphq.com/skills/microsoft-ads-b2b--- # Microsoft Ads for B2B Leans Microsoft search bids toward the companies, industries and job functions that buy, without paying more per click than a lead is worth. ## Purpose - One job: produce the LinkedIn profile bid adjustment plan for one Microsoft Ads account, and check it.- Stage: make. It ends with its own check, at most 2 fix rounds.- It never applies, edits or uploads an adjustment, a bid or a budget. It writes the plan and a bulk sheet; the reader applies them in Microsoft Advertising or its Editor.- It never sets an adjustment from a guess: each one has data behind it, or is labelled as a starting value. ## Use when / not | Situation | Use || --- | --- || Adding company, industry or job function bid adjustments to Microsoft search campaigns | this skill || Importing Google campaigns to Microsoft and setting them up for B2B | this skill (prerequisites in step 2) || Reviewing an existing set of profile adjustments after a read window | this skill || Cleaning up search terms on Microsoft | `search-terms-triage` (it reads Microsoft exports) || Setting the most a lead may cost | `lead-value-math` first || Planning LinkedIn Ads themselves | `linkedin-plan` | ## Inputs | Input | Required | Ways to provide, in order | If missing || --- | --- | --- | --- || Microsoft Ads search campaigns with conversion tracking | yes | a connected account; a campaign report export; the reader | `NOT YET` until a UET tag and a goal (or offline conversions) are live || Campaign clicks, conversions and average CPC (last 30 days or more) | yes | the campaign report | `BLOCKED` for data-driven adjustments; only starting values for target accounts || Max CPL | yes | `lead-value-math` handoff; the reader | ask once; without it, no cap can be set, so `BLOCKED` || Profile reports: company, industry, job function (clicks, spend, conversions) | no | Microsoft's audience or demographics report by profile; an export | no data-driven adjustments; starting values only || Target account list | no | the CRM (target accounts); a CSV; `linkedin-plan` handoff | no company adjustments || Closed-won by industry or job function | no | the CRM | adjustments from platform conversions only, coverage `partial` || Google Ads status | no | the reader; `lead-value-math` handoff | ask: Microsoft copies of Google campaigns that don't work yet only spread the problem || Selling markets | no | `## Company` "Regions" | ask; profile targeting availability varies by market | ## Profile read From `company-profile.md`: `## Company` (regions, category), `## Buyers` (ideal customer, bad fit, buying group: the industries and functions that buy), `## Channels` (Microsoft Ads status, budget, Google Ads status), `## Targets`, `## Accounts and data` (whether Microsoft can receive CRM stages), and every `## Learnings` entry for `microsoft-ads-b2b` with status `noted`, `fact` or `accepted`. No profile: ask in one message for the markets, the most a lead may cost, the industries and job functions that buy, and whether a target account list exists. Say that `company-profile` saves this next time. ## Steps 1. **Apply learnings.** A segment that produced opportunities, or one whose adjustment was cut, changes its starting point. Name each one that fires. 2. **Check the prerequisites** against `references/microsoft-facts.md`:    | Check | Passes when | If not |   | --- | --- | --- |   | Google Ads works first | the Google campaigns being copied produce conversions the CRM agrees with | `NOT YET`: fix Google first |   | Conversion tracking | a UET tag and a goal, ideally CRM stages sent back as offline conversions | `NOT YET` |   | Import schedule | a scheduled Google import can't overwrite Microsoft-side changes | stop the re-import or limit it to new items before adding adjustments |   | Market | LinkedIn profile targeting is available in the selling markets | plan only the markets where it is |   | Policy | the product isn't in a category barred from using demographics (financial, insurance, education, employment, housing) | check Microsoft's policy before using profile targeting |   | Base CPC | the campaign's CPC is at or under max CPL x its conversion rate | `NOT YET` for positive adjustments: fix bids or keywords first |   | Distribution and assets | partner sites reviewed; autogenerated assets reviewed or off | note it; it doesn't block the plan | 3. **Pick where to start.**    | Situation | Start with | Why |   | --- | --- | --- |   | A target account list of up to 1,000 accounts per campaign | company | closest to who you sell to |   | Buyers defined by the industries they work in | industry | wide enough to gather data quickly |   | Buyers defined by function across many industries | job function | |   | Several fit | one dimension now, the next after one read window | adjustments multiply, so adding all 3 at once makes the effect unreadable |    With enough data on 2 or 3 dimensions already, adjust all of them, then run the stacking check in step 5. 4. **Compute each adjustment.** Run `python3 scripts/profile_bids.py --csv segments.csv --campaign-clicks <n> --campaign-conversions <n> --cpc <avg CPC> --max-cpl <max CPL>`. Rules (the script's docstring has them in full):    | Case | Adjustment | Why |   | --- | --- | --- |   | Segment has at least 2 / campaign conversion rate clicks, and conversions | segment rate / campaign rate, minus 1 | it pays more where a click is worth more |   | Enough clicks, 0 conversions, spend at least 2 x max CPL | -90%, or exclude if the segment can't be a buyer | it has spent what 2 leads may cost with none |   | Thin data, on the target account list | a starting value: the reader's, or +15% (Microsoft's own default for new associations) | a lean toward target accounts until data arrives |   | Thin data, not a target | 0 | no evidence either way |   | Any case | capped at max CPL x the segment's rate (the campaign's rate when data is thin) / base CPC, minus 1, and kept within -90% to +900% | no segment pays more per click than its leads can afford |    Weight by closed-won where the CRM has it: a segment whose leads close at twice the campaign's rate earns up to twice the adjustment, still under the cap. Without Python, compute each row by hand, show the formula, and say so. 5. **Check stacking.** A searcher can match a company, an industry and a job function at once, and Microsoft multiplies the adjustments. Multiply the highest positive adjustment of each dimension: the CPC that results must stay under max CPL x the best conversion rate among those segments. If it doesn't, lower the adjustment with the least evidence first (a starting value before a data-driven one). 6. **Exclude only what can't buy.** A segment that can't be a buyer (students, an industry the profile names as a bad fit) with 0 conversions on 2 x max CPL of spend is excluded; anything else gets a negative adjustment. Never exclude a target account. 7. **Write the bulk plan.** One row per adjustment: campaign, ad group (or campaign level), profile type, name, adjustment %, effective CPC, cap, basis and the data behind it. Ad group rows override campaign rows for the same profile, so keep each profile at one level. Keep each campaign or ad group at 1,000 companies or fewer. 8. **Set the read window.** Re-read conversion rates by segment after 30 days and the change log of what was applied. Read pipeline (cost per opportunity against Google's) after the judgment window, 1.5 x the sales cycle. Track the share of clicks from target companies before and after. ## Check and verdict | Check | Fails when || --- | --- || Range | any adjustment outside -90% to +900% || Cap | any segment's effective CPC above max CPL x its rate || Stacking | the highest adjustments together pass the cap || Evidence | an adjustment without data that isn't labelled as a starting value || Limits | more than 1,000 companies in a campaign or ad group || Levels | the same profile set at both campaign and ad group level without saying the ad group wins || Exclusions | a target account excluded, or a segment excluded with conversions | | Verdict | Condition (first match wins) || --- | --- || `BLOCKED` | No campaign data and no target accounts, or no max CPL after one question || `NOT YET` | A blocking prerequisite in step 2 fails (Google not working, no conversion tracking, base CPC above what a click can afford) || `READY WITH GAPS` | The plan stands, but checks are still open after 2 rounds, or only starting values exist for lack of data (each named) || `READY` | Every check passes on data-driven adjustments | Named stops: `BLOCKED` asks for the missing input. `NOT YET` names the prerequisite and its fix. No other stops. ## Loop 1. Run the checks.2. Fix: cap an adjustment, lower the least-evidenced one in a stack, move a profile to one level, split a company list across ad groups, relabel a guess as a starting value or drop it.3. Check again. At most 2 rounds, then hand over with what is still open, labelled `open`. ## Learn Append to `## Learnings` when: - a read window ends: a `fact` per segment with conversion rate and cost per opportunity before and after;- a segment's leads close at a rate far from its platform conversion rate (CRM): a `fact`, and a `proposal` to weight it;- an import overwrote applied adjustments: a `fact`, so the next run checks the schedule first;- the reader overrides a cap or a starting value: a `proposal` with their reason. Example: `- 2026-10-24 | microsoft-ads-b2b | fact | Software development industry converted at 5.0% against a 3.0% campaign rate; +67% held cost per opportunity under Google's | Why: Microsoft profile report and HubSpot, Sep to Oct 2026 | Status: noted` ## Output ```Verdict: <READY | READY WITH GAPS | NOT YET | BLOCKED> · Coverage: <complete | partial (what's missing)> · Math: <script | by hand> Prerequisites| Check | Result | Fix | Starting dimension: <company | industry | job function>, because <reason>Campaign baseline: <clicks>, <conversions>, <rate>, base CPC <value>, max CPC at that rate <value> Bulk plan| Campaign | Level | Type | Name | Adjustment | Effective CPC | Cap | Basis and data | Stacking: <segments, multiplier, effective CPC, max CPC> · Exclusions: <list and why>Read window: <dates for the 30-day read and the pipeline read>Check: <each check, pass or fail> · Rounds: <n>Open: <labelled items>Learnings applied: <list, or none>``` Then the handoff block. ## Hand off ```handoffskill: microsoft-ads-b2bversion: 1.0.0verdict: READYcoverage: completefor: <company> Microsoft Ads searchoutputs:  dimension: industry  adjustments: 7  exclusions: [higher education]  stacked_multiplier: 2.19  read_on: 2026-11-24next: [search-terms-triage, pipeline-report]open: [reader applies the bulk sheet and logs the date]``` ## Failure modes | Problem | Do this || --- | --- || The profile report shows most clicks as unknown profile | Expected: no adjustment applies to unknown profiles. Base the plan on the known share and say what it covers || A target company isn't in Microsoft's list | Note it; it can't be targeted by name || The account runs Target CPA and the adjustments seem ignored | Microsoft's automated strategies use adjustments as signals, not fixed multipliers; judge on segment conversion rates, not CPCs || The reader asks for +900% on target accounts | Cap it at what the leads can afford, and show the cap || Microsoft data is thin because the campaigns are new | Starting values for target accounts only; schedule the first data-driven read after 30 days | ## Influenced by - Microsoft Advertising's documentation on LinkedIn profile targeting and bid adjustments: what can be targeted, bid only rather than narrowing, the -90% to +900% range, the +15% starting value and how profile adjustments multiply: https://learn.microsoft.com/en-us/advertising/msa-help/hlp_ba_conc_linkedintargeting- SHAKEUP's B2B Google Ads playbook: a click is worth what its leads can pay (max CPL x conversion rate), and one change at a time keeps the result readable.

How it thinks

Inputs in, a verdict out, and a stop only where you decide.

What we built

Microsoft can aim search bids with what people list on LinkedIn, but adjustments multiply and most teams set them by feel. We price each segment from its own conversion rate, cap it by what a lead can afford and check the stack before anything is applied.

It reads your company profile if you have one, or asks for only what it needs. Then it works through its steps, checks its own work and ends in a verdict.

The steps

  1. 01

    Read your learnings first

    Segments that produced opportunities before, or adjustments that were cut, change the starting point.

  2. 02

    Check what has to be true first

    Google Ads working, conversion tracking live, imports that can't overwrite your changes and profile targeting open in your markets.

  3. 03

    Start with one dimension

    Company if you have a target list, otherwise industry or job function. Adjustments multiply, so one at a time keeps the effect readable.

  4. 04

    Price each segment from its own data

    A segment that converts at 5% against a 3% campaign rate earns about +67%, capped so a click never costs more than its leads can pay.

  5. 05

    Check how adjustments stack

    A searcher can match a company, an industry and a job function at once. Their adjustments multiply, so the highest ones together must stay under the cap.

  6. 06

    Write the bulk plan

    One row per adjustment with its basis and effective cost per click, for you to apply. A 30-day read and a pipeline read are booked.

What you need

  • Required

    Microsoft Ads search campaigns with conversion tracking

    A connected Microsoft Advertising account → A campaign report you export

  • Required

    Campaign clicks, conversions and average cost per click

    The campaign report

  • Required

    The most a lead may cost

    The handoff from lead value math → Your answer

  • Optional

    Profile reports by company, industry and job function

    Microsoft's profile or demographics report → An export you upload

  • Optional

    Your target account list

    Your CRM → A CSV you upload

What it can return

  • READY

    Every adjustment has data behind it, sits inside its cap and stacks safely.

  • READY WITH GAPS

    The plan stands, but some adjustments are labelled starting values until data arrives.

  • NOT YET

    Google Ads isn't working yet, tracking is missing, or the base cost per click is already too high.

  • BLOCKED

    There's no campaign data, no target list or no cost per lead ceiling.

Where it stops for you

  • When campaign data, a target list or your ceiling is missing
  • When a prerequisite such as tracking fails

Example output

What you get back.

A run for a made-up company. Your output depends on your inputs.

Here's a plan for a made-up B2B SaaS company's Microsoft search campaign. It ran 4,000 clicks and 120 conversions last month at a $6.50 average CPC, and the most a lead may cost is $285.

Verdict

READY

Coverage: complete

SegmentDataAdjustmentCost per click afterCap
Software development45 of 900 clicks, 5.0%+67%$10.83+119%
Marketing and advertising18 of 500, 3.6%+20%$7.80+58%
Financial services6 of 300, 2.0%-33%$4.33-12%
Higher education0 of 250, $1,625 spentExcludenonenot a buyer
Marketing job function54 of 1,200, 4.5%+30% (from +50%)$8.45+97%
Sales job function15 of 600, 2.5%-17%$5.42+10%
Stacking

a software company's marketer would get +67% and +50%, about 2.5 times the bid, or $16.28. That's above the $14.25 a click there can afford, so the job function adjustment drops to +30% and the stack lands at $14.11.

The handoff block the next skill reads
skill: microsoft-ads-b2b
version: 1.0.0
verdict: READY
coverage: complete
for: Microsoft Ads, NB Core
outputs:
  dimension: industry and job function
  adjustments: 5
  exclusions: [higher education]
  stacked_multiplier: 2.17
  read_on: 2026-11-24
next: [search-terms-triage, pipeline-report]
open: [apply the bulk sheet and log the date]

What good looks like

  • Every adjustment shows the data behind it, or says it's a starting value
  • No segment's cost per click above what its leads can afford
  • A stacking check on the highest adjustments together
  • Exclusions only for segments that can't buy
  • A bulk plan you apply yourself, with read dates booked

Install

Use it in the tool you already have.

Pick your tool. Each tab has the shortest path we know of.

  1. 01
  2. 02
    In Claude, open Customize, then Skills.
  3. 03
    Choose +, then Upload a skill, and pick the ZIP. Code execution must be on in your settings.
  4. 04
    Ask for it by name, or let Claude use it when a task matches.

Questions

Before you add it.

Does LinkedIn profile targeting narrow who sees the ads?

No. Microsoft treats it as bid only. Searchers outside the companies, industries or functions you pick still see your ads at the normal bid.

Why cap the adjustments?

Because a higher bid only pays if the segment's leads are worth it. The cap is your max cost per lead times the segment's conversion rate.

What happens when adjustments overlap?

Microsoft multiplies them. +67% on an industry and +50% on a job function come to about 2.5 times the bid, so the skill checks the highest ones together.