The Twenty Questions
1. "Do you report on clicks or landing page clicks?"
Ask this first. LinkedIn's clicks field counts likes, comments, shares and profile clicks alongside real website visits. On thought leader ads the gap is 3–10×.
Good answer: they explain the difference immediately and say they report LP CTR and LP CPC as primary. Bad answer: confusion, or "we report CTR" without qualification. That means every report they have ever sent has been inflated, usually without them realising.
2. "Show me a mid-funnel retargeting campaign you've built."
Good answer: 180-day windows segmented by engagement depth — video viewers at 25/50/75%, ad engagers, lead form abandoners. Bad answer: one "all warm audiences" campaign, or a blank look. Most accounts have cold ads and demo requests with nothing between.
3. "Is your fee a percentage of my spend?"
If yes, ask what happens to their invoice when they cut £5,000 of monthly waste. Percentage-of-spend pays an agency more for spending more of your money. There are reasonable answers — some cap fees or share savings — but the question reveals how they think. See agency cost models.
4. "Who actually works on my account day to day?"
Bad answer: the person pitching is never seen again and an account manager relays messages to an unnamed executor. Ask to meet the person doing the work, and ask how many accounts they carry.
5. "Can you connect ad spend to pipeline?"
B2B buyers take 15+ touchpoints over three to six months and LinkedIn rarely gets last-click credit. An agency reporting only cost per lead will under-report LinkedIn's real contribution and optimise toward cheap leads that never close.
Good answer: CRM integration, influenced pipeline, a view of which closed-won deals had ad exposure before the opportunity existed.
6. "What will you switch off in the first week?"
Anyone experienced has a list: audience expansion, the LinkedIn Audience Network on most B2B accounts, campaigns with no destination, conversions that never fire. Bad answer: "we'd need to look" with no instinct at all. Everyone who has audited accounts has a mental checklist.
7. "What would make you tell me to stop spending?"
A genuinely good agency has conditions under which they would tell you LinkedIn is wrong for you — undefined ICP, transactional product, budget below the viable floor.
Bad answer: "LinkedIn works for everyone." It does not, and an agency unwilling to say so is optimising for closing you.
8. "How do you handle creative, and how often is it refreshed?"
UK audiences are small and saturate fast. Good answer: a rotation cadence tied to frequency and LP CTR decay, plus a view on thought leader ads versus company page ads. Bad answer: creative is your job, or refreshed "when performance drops" with no measure of what that means.
9. "What does month one actually look like?"
Good answer: audit, ICP and account list, tracking rebuild, then build. Honest about demand generation taking three to six months. Bad answer: campaigns live in week one and pipeline promised by day 30 on a cold account. That is not how B2B buying works.
10. "What happens to my account if we stop working together?"
Good answer: you own the ad account, creative, audiences and data, and spend runs on your own account. Bad answer: anything where the agency owns the ad account. Account history affects LinkedIn's algorithm, so leaving without it means starting from zero.
11. "What does your reporting actually look like?"
Ask to see a real client report with the numbers redacted. Good answer: it leads with pipeline or opportunities, uses landing page clicks, and states what changed and why. Bad answer: a platform screenshot with impressions at the top, or a refusal to show you anything at all.
12. "What have you got wrong recently, and what did you change?"
An agency that has never made a mistake has either not been running accounts long or is not being straight with you. Good answer: a specific example, what it cost, and the process change that followed. Bad answer: a non-answer, or a story where the client was at fault.
13. "What is the cost per landing page click on your accounts, and what is the benchmark?"
The number should come with a source. Kiin Labs' 2026 panel across 944 accounts: $17.54 median, $9.18 in the good quarter; single image on the website visits objective $11.33. Good answer: a figure for a comparable audience with a stated source. Bad answer: "around $4" (that is LinkedIn's clicks column), or "it depends" with no range.
14. "How much of my spend would go to the LinkedIn Audience Network, and why?"
Good answer: none, and they turn it off on day one and re-check after saving, because Campaign Manager re-enables it. Bad answer: "it extends reach". Across 107 accounts we audited, $274,000 had gone off-platform, in the worst case 99% of a budget.
15. "Which objective would you use for traffic, and which for thought leader ads?"
Good answer: website visits for traffic (the cheapest visit on the platform), engagement for thought leader ads on audiences under 30,000 and brand awareness above, video views for video. Bad answer: brand awareness for everything "because it is cheaper": it buys a $48 visit against $10.52.
16. "How do you bid?"
Good answer: manual, set low, checked daily, raised only when delivery stalls. The highest-bid campaign in an account pays 128% more per engagement than the lowest for the same audience. Bad answer: maximum delivery everywhere, or the suggested bid taken as given.
17. "Where does the demo ask go?"
Good answer: in a message ad (31 to 40% form completion, $175 a demo lead at the median) or behind a content offer; never a demo form on a single image in the feed (2.1% completion, $600 a lead, 48 form opens per lead). Bad answer: "we'd run a lead gen form on a single image to start with."
18. "What cost per lead would you report, and split how?"
Good answer: split by offer, content against demo, because they are different products at different prices ($164 and $343 at the median; $67 and $131 in the good quarter). Bad answer: one blended number, which describes the mix of offers rather than the performance.
19. "What is the account-level frequency, and when do you refresh creative?"
Good answer: they read frequency from the reach report at account level (campaigns report 3.5 when the account delivers 9.5), and refresh at one pass of the audience (every two weeks under 10,000 members, monthly over 100,000) because engagement is down 9% after one pass and 22% after five. Bad answer: "when performance drops."
20. "What do you do with the Companies tab?"
Good answer: read it monthly, exclude the companies that click and never convert, build retargeting lists from the engaged ones, and hand the engaged companies to sales or outbound with what they engaged with. Bad answer: a blank look. It is the one report in Campaign Manager that connects the ads to named accounts.
Ten Red Flags
- Guaranteed leads or a promised CPL before seeing your account. Nobody can know this.
- A single ROAS number quoted as a forecast. Averages are not predictions.
- No published pricing and evasiveness on a call. Fees are not commercially sensitive.
- Reports leading with impressions. Nobody leads with reach when they have pipeline to show.
- Owning your ad account. Occasionally framed as a convenience. It is leverage.
- Twelve-month lock-ins. Three months is reasonable because demand gen needs time. Twelve is protecting them, not you.
- LinkedIn as one of fifteen services. Fine for some companies, but do not expect depth.
- A fee that is a percentage of spend with no cap and no shared-savings clause. They are paid more when the account is less efficient; how agencies charge, with ranges, is on the agency cost page.
- No benchmark for any number they quote. "Good" without a source is an opinion. The 2026 benchmarks are public.
- A demo form on a single image in the feed as the plan. It is the most common setup on LinkedIn and the worst on cost per lead.
Green Flags
- They correct a wrong assumption you have, on the first call, at commercial risk to themselves
- They ask about your sales process and deal size before talking about ads
- They can name the specific thing they would check first, and why
- They tell you which of your ideas will not work
- They publish pricing
- They have written something you learned from before you ever spoke
Once you have chosen, hold them to what the first 90 days should look like; if you already have an agency, run the five-number audit first.
A Sensible Selection Process
- Shortlist three. One specialist, one full-service, one freelancer or offshore option, so you are comparing models rather than logos. Start from the UK shortlist.
- Ask all three the twenty questions. Same questions, same order. The differences will be stark.
- Ask each for one specific finding on your account. Anyone serious will spot something in twenty minutes. Compare what they find.
- Check the account ownership and notice terms before price.
- Then compare price — and compare all-in, including media, not fee alone.
If you would rather run the diagnostic yourself first, the eight-layer audit and the nine signs your current agency is wasting budget will tell you what to ask about.