The Evidence

Reverse-engineer strategy from existing employer brand work work.

"Most creative work contains an implied strategy. The Evidence makes that strategy visible, testable, and presentable."

The Core Insight

Agencies often create before they strategise. Ideas get approved, campaigns launch, and only later does someone ask: "What was the strategy behind this?" The Evidence works backwards from finished creative to surface the strategic framework that the work implies.

This is not reverse-engineering in an academic sense. It is practical strategy extraction that lets you present creative work with strategic backing, stress-test ideas against rigour, or compare what you built against what was briefed.

Why This Tool Exists

Three common situations call for The Evidence:

• The Drift Check: Creative moves during a pitch. Import the locked strategy and the Evidence shows you, field by field, where the work has wandered, so you decide whether to pull the creative back or update the line.
• Creative stress-testing: Before presenting, check whether the creative actually serves the strategy it claims to serve.
• Post-hoc strategy: You have employer brand work the organisation loves but no formal strategy document. The Evidence builds one.

How It Works

Step 1: Organisation Context

Provide organisation details: name, sector, and region. Optionally add a description, competitors, and website.

You can also import from a completed pitch to pre-fill fields and enable the Drift Check.

Employer Context Documents

Upload materials that explain the organisation as an employer: EVP documents, employee research, engagement surveys, employer brand guidelines. These enrich the AI's understanding of the employer brand and improve decode accuracy. Up to 5 files.

Organisation Validation

When you click "Validate Organisation", the tool checks two things:

Validation produces one of three outcomes:

Sector validation is blocking: you cannot proceed with an invalid sector. Organisation recognition is advisory: you can always proceed, but the tool is honest about what it can and cannot do.

Step 2: Creative Input

You can provide creative input in two ways: describe the work in text, upload images of it, or both.

Image Uploads

Upload images of the employer brand work directly: careers page screenshots, job ad visuals, employer brand campaigns, recruitment marketing materials, EVP creative. The AI analyses the images using vision capabilities. When images are uploaded, the text description becomes optional. You can skip straight to decode or add context to sharpen results.

Text Description

If no images are uploaded, a text description of at least 100 characters is required. Richer descriptions produce better decodes. See the Decode Guide for what makes a good creative input.

Structured Prompts

Expand the "Want a sharper decode?" accordion for guided input fields:

• What is the creative idea? The core concept, separated from the execution format.
• What does the audience experience? What they see, hear, or interact with.
• Where does it run? Channels, media, formats.
• What response is it trying to provoke? The intended emotional or behavioural shift.

These fields are optional but significantly improve decode quality. They are concatenated with the main description before analysis.

Creative Work Uploads

Upload the actual employer brand materials you want to decode: careers site content, job advertisements, recruitment campaigns, EVP documents, onboarding materials, employee communications, employer brand guidelines. Images (JPG, PNG, GIF, WEBP) are analysed visually. Documents (PDF, Word, Excel, PowerPoint, text) have their text extracted. Up to 5 files.

For best results, upload materials from a single campaign or execution. Multiple campaigns should be decoded separately.

Step 3: Decode

The AI analyses your creative work against the organisation context and produces a complete reverse-engineered employer brand strategy. This covers eight fields drawn from across the Threader Suite: problem, convention, talent tension, positioning, direction, exclusions, organising thought, and brand attitude.

Step 4: Results

All decoded fields are editable. Each carries a confidence rating (High, Medium, Low) based on how much evidence the creative input provided. Fields decoded from rich descriptions receive high confidence. Fields requiring significant inference receive low.

The Decoded Strategy synthesises everything into a single editable sentence.

Re-decode with Edits

If you edit decoded fields and then re-decode, your edits are preserved as "golden threads". The AI treats your refinements as the strategic direction and refines around them rather than starting from scratch. This lets you iteratively sharpen the decode without losing your work.

The Drift Check

The Readiness Check finds the objections. The Drift Check finds the drift. The Readiness Check runs in the Signal when you lock the pitch; the Drift Check runs here, whenever creative drafts arrive. Lock it checked, keep it checked.

When you import a completed pitch, The Evidence compares the decoded strategy against the pitch strategy, pair by pair. This produces:

• Six field comparisons: Real Talent Challenge, Talent Tension, Position, Direction, Organising Thought, and Exclusions. The Exclusions pair checks whether the creative does anything the pitch explicitly ruled out, and whether it has quietly added constraints the strategy never made.
• A status per pair: Aligned, Partial, or Divergent, with a gap note under every Partial and Divergent row explaining exactly what moved.
• An overall alignment percentage.
• A strategic recommendation that commits to one of three actions: pull the creative back to the line, update the strategy through the cascade, or accept the divergence deliberately.

Reading the score

Reading the score: 60 and above, the work is carrying the strategy; shore up the flagged rows. From the mid-20s to around 60, the work has drifted; pull it back using the gap notes. Below the mid-20s, the work is likely following a different strategy, or violates what was locked; expect the recommendation to say restart. In every case the recommendation, not the number, is the instruction: the rows and gap notes are the check, the score is the summary, and two checks with similar scores can deserve different actions.

Rich creative can support more than one reading, so run-to-run variation in the score is expected. Read the rows.

The report

Download the completed check as a standalone Word document from the Drift Check panel, formatted like the other Threader reports. The check is also appended as a Drift Check section to the decode Word export and PDF report.

Plans and requirements

The Evidence is available on Lite and Pro, and decoding is full-capability on both. The Drift Check needs a pitch locked through the full cascade (Diagnosis to Signal), so in practice it is a Pro feature: the cascade is where the locked promise comes from, and locking requires the Pro cascade tools. Check one campaign or execution at a time; separate campaigns should be decoded and checked separately.

Strategic Confidence

The confidence score uses the same three-factor system as all Threader Suite tools:

Hover the ? icon on the confidence badge to see the full breakdown. Uploading employer brand documents and EVP materials directly increases the Input Quality score, which is the most actionable way to improve confidence.

Tips for Best Results

• Describe, don't evaluate. Tell the tool what the creative does, not whether it is good.
• Include channels and context. Where the work runs affects the decoded strategy significantly.
• Use the structured prompts. They force you to separate idea from execution, which produces sharper decodes.
• Upload employer brand documents for unfamiliar organisations. If the validation shows amber or red, EVP documents or employer brand guidelines make a significant difference.
• Upload the brief. If you have the original brief, upload it to the brand context section. It grounds the decode.
• Edit the output. The decode is a starting point. Your strategic judgement refines it. Re-decoding after edits preserves your direction.

Mode & Access