Most influencer shortlists arrive sorted by follower count. It is the easiest number to collect, the easiest to put in a spreadsheet and the easiest to defend to someone who is not close to the work.
It is also the number with the loosest connection to whether anyone changes their mind about your brand.
The short answer. Choose on audience relevance, topic authority, engagement quality and track record of talking credibly about brands. Then judge performance against that creator's own typical results rather than against platform averages. Follower count only matters once those conditions are met.
You are borrowing credibility, not renting reach
When you pay a creator, you are not buying media space. You are asking their audience to extend some of the trust they have in that person to you.
That trust is specific. It is attached to a subject, a tone and a track record. A creator who is trusted on skincare is not automatically trusted on supplements. An operations director who is read closely on warehouse automation has no credibility on cybersecurity, however large their following.
This is why reach is a poor proxy. Reach tells you the size of the room. It tells you nothing about whether the person at the front is worth listening to on your topic, or whether the room is full of people who could ever buy from you.
So selection should start with the audience you need to influence and work backwards to the people that audience already listens to, rather than starting with a list of available creators and hoping the overlap is good enough.
B2B influence has quietly become a serious channel
Influencer marketing is still filed under consumer brands in most organisations. That is increasingly out of date.
B2B influence rarely looks like a sponsored post. It looks like a practitioner with 6,000 followers whose posts get quoted in internal Slack channels. A consultant whose newsletter is read by the exact 400 people who sign off on your category of purchase. A podcast host who spends 45 minutes with a specialist and produces the most useful thing published on that topic all year. An analyst whose opinion shapes the shortlist before you are ever invited to pitch.
The economics are different too. A consumer campaign might need scale to work. A B2B programme can pay for itself if it reaches 30 of the right people, because the value of a single opportunity is high and the buying group is small.
What makes B2B influence powerful is also what makes it fragile. These audiences are professionally sceptical. They can spot a paid post immediately, and they judge the creator, not the brand, for anything that reads as bought. A creator who takes obviously unsuitable work loses standing quickly, which is precisely why the good ones are selective.
Two things follow. The best B2B partners will interrogate your product before agreeing, and that is a positive signal rather than an obstacle. And some of the most effective B2B influence is not paid at all: customers willing to speak on record, employees with genuine expertise and standing in their field, and long-term relationships with people who would talk about you anyway.
Why creator choice now affects whether AI systems mention you
There is a second reason this decision matters more than it used to.
When someone asks an AI assistant which suppliers to consider, the answer is assembled from what has been published and indexed about your category, including social posts, community discussions, podcast transcripts, video and the coverage that creator activity generates. Brands repeatedly named in useful, substantive content about a topic are more likely to appear in those answers than brands only described on their own website.
That reframes what you are buying. A creator who produces one piece of genuinely informative content about a problem you solve, and names you in it, is contributing to a body of public evidence about what you do. A creator who posts a discount code and a product shot is not. It is one of the few places where AI visibility and brand partnerships meet directly.
There is a measurable version of this in search behaviour. In work for Tu Clothing, a partnership with the lifestyle platform SheerLuxe was accompanied by a 75% rise in Google searches for "Tu Denim" over the three days around the article going live, with 90% of the wider campaign coverage including a link to purchase. The point is not the size of the percentage. It is that a well-matched partnership changed what people went looking for by name, which is the same behaviour that builds the brand associations AI systems later draw on.
Six signals that tell you more than follower count
Audience composition, not audience size. Ask for the split by country, age and, for B2B, job title and seniority. A creator with 200,000 followers where 8% are in your market is weaker than one with 15,000 who are almost all in it.
Comment quality. Read the last 20 posts properly. Are people asking questions, disagreeing, sharing their own experience? Or is it emoji and one-word replies? Conversation is harder to buy than a like.
Topic consistency over time. Has this person been talking credibly about your subject for years, or did they discover it last quarter? Authority accumulates. It cannot be rented for six weeks.
How they talk about brands they are not paid by. Someone who recommends things unprompted, including competitors, has an audience relationship worth borrowing. Someone whose every positive opinion is sponsored does not.
Whether their audience does anything. Do people turn up to their events, subscribe to their newsletter, buy the things they make? Attention that converts elsewhere is more likely to convert for you.
How they handle the commercial conversation. Do they ask what you are trying to achieve, or send a rate card and wait?
What a poor fit looks like, even with strong numbers
A large but general audience with no clear centre of gravity. High follower growth with flat engagement, which often indicates bought or incentivised growth. A feed where sponsored posts sit back to back, so nothing stands out. Comments that are complimentary but generic. A tone your customers would find performative. And in B2B specifically, someone who has opinions about everything, because breadth without depth reads as commentary rather than expertise.
The category to watch most carefully is the account with real numbers and nothing behind them: genuine followers, respectable impressions, no discernible influence. The tell is usually that nothing measurable ever happens when they post. Ask directly what results their last three brand partnerships produced. A creator who tracks their own performance and can talk about it is a professional. One who only offers follower count is selling reach, which brings us back to where we started.
Choosing relevance over reach: what this looks like in practice
For Bing (Acamar Films), the task was to help parents through everyday parenting challenges. The obvious route was lifestyle parent-influencers with large followings. Instead the campaign partnered with parenting experts, including a BBC presenter and child psychologist, a clinical psychologist and a TikTok parenting authority, on the grounds that the subject demanded authority rather than audience size.
The campaign delivered nearly 5.3 million impressions and over 900,000 engagements on Bing's owned channels, along with more than 4,000 extra visits to the website.
That is a useful proof point because the selection criterion was expertise, and the reach followed rather than led.
The over-delivery problem, and how briefing causes it
Here is a failure that rarely appears in post-campaign reports: the creator tried too hard.
Less experienced creators, particularly those early in commercial work, often over-deliver on the brand side. They mention the product more than they would naturally, use your phrasing rather than theirs, hit every message in the brief and thank the brand twice. Every individual element is compliant. The overall effect is an advert, and their audience treats it accordingly.
The cause is usually the brief, not the creator. A brief that lists eight messages, mandates specific phrasing and demands the product appears within five seconds leaves no room for the thing you were buying.
A better brief does four things. It explains the audience problem rather than the product features, so the creator can find their own way in. It separates the genuine non-negotiables, which are typically legal, regulatory and disclosure requirements, from preferences. It names the one thing a viewer should understand, not the eight things you would like them to remember. And it states clearly what they must not do, which is far more useful than a list of things they must say.
Then leave the execution to them. If you find yourself rewriting their script into your brand tone of voice, you have converted a credible recommendation into an expensive advert.
Length of relationship matters here too. In a programme for the LYMA Laser, a small number of creators each ran six-month partnerships sharing honest personal accounts of their skin ageing, including before-and-after results. It achieved a 7% average engagement rate, above the average for sponsored content, and 441,000 organic video views across three creators.
Three creators, six months, unscripted personal stories. That is a structural choice rather than a creative accident: sustained partnerships let an audience watch something unfold, which is difficult to fake and hard to achieve in a single post.
What to measure when nobody clicks
Most influence does not produce a click. Someone watches a video, remembers a name, searches for it four weeks later and arrives through organic search. Your analytics will record that as organic. Attribution models will credit the last thing that happened.
This is why influencer programmes are often judged on the only numbers that are easy to collect, which are impressions and likes, and then quietly defunded when someone senior asks what it did for revenue.
Build the measurement before the campaign, not after it.
Set the benchmark against the creator, not the industry. Ask for their median performance on their last ten organic posts and their last five paid ones. Judge the work against that. Sponsored content typically underperforms a creator's organic baseline, so comparing it to a platform-wide average tells you almost nothing.
Capture self-reported attribution. Add an open "how did you hear about us?" field to enquiry forms and make sure sales ask the question on discovery calls and log the answer. In B2B this is often the only place a creator's name appears at all, and it is more reliable than most models.
Watch branded search and direct traffic during and after the activity. A rise in people searching for you by name is one of the clearest available indicators that someone made you memorable.
Use a holdout where budget allows. Running activity in some regions or audience segments and not others is the closest most programmes get to a clean read.
Track qualified pipeline, not leads, where the sale is considered. The meaningful question is whether opportunities appeared from accounts or job titles matching the creator's audience, and whether those opportunities progressed.
What good looks like depends entirely on the job you gave the campaign. If the objective was to be considered by an audience that does not know you, then prompted awareness, branded search and shortlist appearance are the right measures, and asking for immediate sales is a category error. Agreeing this before the contract is signed prevents the familiar argument six weeks later, when a campaign that did exactly what it was designed to do is judged against a metric nobody mentioned at the start.
When influencer marketing is not the answer
Some honest reasons to hold off.
If people already know you and are not buying, the constraint is probably not awareness. More attention on a proposition that is not landing simply produces more people who decline faster, and it is worth working out which problem you actually have first.
If the journey after the click is weak, influence will expose that rather than fix it. It is worth checking where interest currently disappears before paying to create more of it.
If you cannot say what you want someone to understand after seeing the content, no creator can rescue that. Vagueness in the brief becomes vagueness in the output.
And if your category genuinely has no community, which is true of some specialised B2B markets, your budget may work harder on customer advocacy, analyst relationships and getting your own experts published. Social media and content strategy builds the same credibility by a different route.
The test worth applying before you sign anything
Ask one question about every name on the shortlist: if this person recommended us without being paid, would anyone be surprised?
If the answer is yes, you are buying an advert and should price and measure it as one. If the answer is no, you have found someone whose endorsement will survive being questioned, and the follower count is now simply a detail about scale rather than the reason you chose them.
More on how we approach influencer marketing, or how it connects to winning attention and trust more broadly.
Not sure which creators fit your business?
We will look at who your buyers already listen to, what those creators' own baselines look like, and what you would need to measure to prove it worked. Then we will tell you whether influence is the right spend at all.
