The Evidence Guide
What The Evidence needs to produce reliable results.
The Evidence reverse-engineers employer brand work into strategic frameworks. The sharper your description of the existing employer brand activity, the sharper the decoded strategy. This guide shows what good input looks like, and where descriptions usually fall short.
The Five Essentials for Good Evidence Input
1. The EVP or Employer Brand Idea
What good looks like: A clear description of the employer brand concept, not just the execution format. "A careers campaign that reframes working at this company as joining a movement, positioning the employer as the place where personal ambition and social impact align. The core idea is that your career can change more than your bank balance."
Red flags:
- "We want a cool careers page" (format, not idea)
- "A LinkedIn campaign" (channel, not concept)
- "Something that attracts top talent" (aspiration, not description)
2. The Experience
What good looks like: What does the candidate or employee actually see, hear, or interact with? "A 90-second film follows a day in the life of three different employees. Instead of showing perks, it shows the problems they solve. Each segment ends with the question they asked in their interview. End frame: 'Better questions lead here.'"
Red flags:
- No description of what the audience encounters
- Only the logo or tagline described
- "It would be very authentic" (tells nothing)
3. The Message
What good looks like: What takeaway or shift is the employer brand work trying to create? "The intended takeaway is that this company values intellectual curiosity over credentials. The emotional shift is from intimidation about not being good enough to excitement about learning."
Red flags:
- No stated message or intended effect
- "We want people to apply" (objective, not message)
- Message contradicts the described experience
4. The Channels
What good looks like: Where the work runs and why those channels were chosen. "The campaign leads with employee stories on LinkedIn, supported by a redesigned careers site. Internal comms mirror the external messaging through onboarding materials and internal newsletter redesign."
Red flags:
- No channels mentioned
- "It would work everywhere" (no channel thinking)
- Channel choice contradicts the implied talent audience
5. The Context
What good looks like: What prompted this work and what constraints shaped it. "This is an EVP refresh following two years of declining application quality in engineering roles. The CHRO mandated that the new EVP must work across all talent segments, not just tech. Budget covers creative development and LinkedIn activation only."
Red flags:
- No context for why this work exists
- No mention of constraints or mandatories
- "The CEO just wants something fresh" (no strategic context)
Good vs Weak: Side by Side
Weak Input
"Google employer brand. Great perks. Best place to work energy. Would be on LinkedIn and careers site."
Result: A generic analysis that could apply to any tech employer brand. Low confidence across all fields. The Evidence has to infer almost everything.
Strong Input
"A Google employer brand campaign targeting experienced engineers aged 30-45 who are considering leaving big tech for startups because they feel their impact is invisible at scale. The creative concept is 'Your Code Runs Here': a series of short films showing real infrastructure that Google engineers built, matched to the specific engineer who built it, filmed at the actual physical locations (data centres, undersea cables, rural connectivity projects). The films end with 'You built this. What will you build next?' Runs on engineering blogs, targeted LinkedIn, and YouTube pre-roll for tech audiences. Internal version shown at all-hands. The brief was to reduce senior engineering attrition, not attract new graduates."
Result: A precise analysis with high confidence. Clear talent tension (invisibility at scale), identifiable EVP (tangible impact), specific convention challenge (tech employer brands usually celebrate perks, not infrastructure).
Common Pitfalls
| Pitfall | Why It Weakens the Analysis | Fix |
|---|---|---|
| Describing the brief, not the work | The Evidence needs to analyse what was made, not what was asked for | Focus on what was produced, not the assignment |
| Evaluating instead of describing | "It's really authentic and bold" tells the AI nothing | Describe what makes it bold: what does it show, say, do? |
| Single-line descriptions | 100 characters is a minimum, not a target | 500+ characters dramatically improves analysis quality |
| Missing the talent audience | The AI cannot decode talent tension from "everyone" | Who specifically is this talking to? |
| No channel context | A LinkedIn campaign implies a very different strategy to an internal comms programme | Always include where the work runs |