The Smart Assessment turns a plain-English description of an AI or automation idea into an estimated annual value, using the measures that fit what the idea is for. SilkFlo’s AI drafts the answers from your description, asks for anything it’s missing, and calculates the value as you go: hours saved, cost avoided, revenue gained or risk reduced. You check and correct; you don’t start from a blank form.
It replaces the old Detailed Assessment.
Why assess ideas this way #
Most organisations assess AI ideas in one of two ways, and both break down:
- A long form, which analysts abandon halfway or fill with guesses to get it done.
- A spreadsheet per idea, each built slightly differently, so two ideas can’t be compared fairly.
The Smart Assessment fixes both. The AI does the first draft, so the effort goes into checking numbers rather than typing them. And every idea is valued with the same method for its Value Driver, so a £200,000 estimate on one idea means the same thing as £200,000 on another.
The Value Driver matters because different ideas create value in different ways. A cost-reduction idea is measured in hours and labour cost. A customer experience idea is measured in a score like NPS and how many customers it touches. A risk idea is measured in the cost and likelihood of the event it prevents. Asking every idea the same questions would force you to guess, so SilkFlo asks only the questions that fit.
Who can use it #
- Submitting a new idea straight into a Smart Assessment: users whose system role includes Submit CoE Driven Ideas. That’s the Authorised User, Program Manager and Account Owner roles. See Understanding User Roles in SilkFlo.
- Completing the assessment for an idea already in the pipeline: once an idea reaches the Assess stage, the Smart Assessment unlocks on its profile page. Admins can edit it at this point.
Standard Users submit ideas through the guided Modal instead. See Submitting an Innovation Idea.
Ideas submitted as a Smart Assessment skip the Review stage and go straight to Assess.
Start from the AI Opportunity Finder #
- On your Personal Dashboard, describe your idea in a sentence or two in the AI Opportunity Finder, then click Generate Idea.
- You go straight to the Smart Assessment page. Fields fill in live: name, summary, pain points, impact. Each one is flagged with a ✦ AI suggested badge.
- SilkFlo picks the most likely Value Driver from your description. Change it if it’s wrong.
- If your description didn’t include the numbers SilkFlo needs, a short card appears with one to three plain-English questions, such as “How many people currently do this task?” Answer what you can, or click Skip for now.
- As your answers land, the value sidebar updates and your headline number counts up. You also see a confidence rating (High, Medium or Low): how confident SilkFlo is in the fields its AI filled in.
- Check every field with a ✦ badge. Edit anything that’s wrong; the badge changes to show you’ve changed it.
- Click Save as Draft to come back later, or submit when you’re happy.
Tip: the more specific your description, the less you’re asked afterwards. “Our 4-person AP team keys around 300 supplier invoices a week into SAP, about 6 minutes each” will fill far more fields than “automate invoices”.
Start from a blank assessment #
Use the Submit Idea button in the side menu, choose CoE Idea, and fill in the fields yourself. The sections are the same as below. There are just no AI suggestions.
What’s in the assessment #
1. Core details (every idea) #
| Field | What to enter |
|---|---|
| Initiative Name | A short, plain-language title, e.g. “Reduce invoice processing time”. |
| Summary | 2–3 sentences on what you want to change and why, written for someone who doesn’t know the process. |
| Pain Point | What’s frustrating, slow or error-prone today. Specific examples help. |
| Negative Impact | What it costs the business if nothing changes. |
| Business Unit | The team that owns the problem, not the team that might deliver the solution. |
| Location | The office, site or region where the work happens. |
| Process Owner | The person accountable for the work day to day. |
| Collaborators | Anyone else who should help shape or review the idea. |
| Applications Used | The systems involved. Your admin manages this list in Settings > Platform Setup > Applications. |
2. Value Drivers #
Pick a Primary Goal: the one outcome that matters most. It decides which questions you’re asked next and how your business case is calculated. You can also add Secondary Benefits: any or all of the other five drivers.
| Value Driver | Use it when the idea mainly… |
|---|---|
| Productivity & Automation Improvement | frees up people’s time |
| Cost Reduction | removes labour, software or external service cost |
| Revenue Growth | moves a revenue metric such as conversion rate or deal size |
| Customer Experience & Improvement | improves NPS, CSAT or Customer Effort Score |
| Compliance and Risk Mitigation | reduces the cost or likelihood of a risk event |
| Employee Experience and Retention | improves eNPS, retention or time spent on tedious work |
3. Value driver questions #
You only see the questions for the drivers you picked.
Productivity & Automation Improvement and Cost Reduction each have a Simple / Detailed toggle.
- Simple asks for one headline estimate. Productivity asks for the total hours a year spent on the process. Cost Reduction asks for the estimated annual saving.
- Detailed builds the number from its parts:
- Volume: how often the task runs, the average volume each time, and minutes per transaction.
- People: how many employees do the task, and their average full cost per year.
- Working time: working days per year and hours per day. These default to your Account settings; adjust them if this team works differently.
- Your target: the percentage of the volume (Productivity) or cost (Cost Reduction) you expect the initiative to handle or remove. This is your estimate, not a system guess.
- Cost Reduction only: the primary source of cost (internal labour, external services, software, mixed). This decides which cost fields appear, such as annual service cost or annual software/licence cost.
- Optional quality and readiness questions (tier 3):
- Error rate
- Rework time
- Review/audit time and percentage reviewed
- Data structure
- Process and application stability
- Number of steps
- Decisions per transaction
Revenue Growth asks:
- the metric you’re moving: Conversion Rate, Average Deal Size, Customer Lifetime Value, or New Leads/Opportunities
- its current and target value
- the average revenue per unit of improvement
- how many units are in scope, and how often
Customer Experience & Improvement asks:
- the metric (NPS, CSAT or CES)
- the current and target score
- how many customers are affected each year
Compliance and Risk Mitigation asks:
- the risk category
- the estimated cost of one risk event
- its annual probability today, and the probability you expect afterwards
- whether this is a mandatory response, e.g. to an audit finding or regulation
Governance questions cover mission-critical systems, sensitive data (PII), and AI making automated decisions. Bias and explainability questions appear only if you answer Yes to automated decisions.
Employee Experience and Retention asks:
- the metric (eNPS, retention/turnover, or time spent on tedious work)
- the current and target score
- how satisfied people are with the process today
How the value is calculated #
The sidebar updates as you type. Here’s what sits behind each number.
Productivity & Automation Improvement and Cost Reduction
- Total annual hours (as-is): volume × frequency × processing time. Rework and review time are added where you’ve entered them.
- Annual hours saved: total annual hours × your handled/deflected %.
- Labour cost savings: hours saved × hourly cost. Hourly cost is the employee full cost ÷ (working days × hours per day).
- Total cost savings (Cost Reduction): labour savings + software cost removed + service cost removed. In Simple mode, it’s the annual saving you entered.
- FTE benefit: hours saved ÷ (working days × hours per day).
- Benefit per employee: hours or £ saved ÷ number of employees doing the task.
Revenue Growth
- Incremental annual revenue: (target − current metric) × annual volume in scope × revenue per unit.
Compliance and Risk Mitigation
- Annual value at risk reduced: cost of one event × (current probability − target probability).
Customer Experience and Employee Experience
- These are shown as strategic targets (for example, +15 NPS) rather than a £ figure.
Total Estimated Annual Value adds up the financial benefits: cost savings, incremental revenue and risk value reduced. Productivity labour savings count too, once the current value of the task has been calculated. It’s the headline number, so you can compare ideas with different drivers side by side.
Ease of Implementation estimates the effort involved, banded Easy, Medium or Difficult. It’s based on process and application stability, how hard the decisions are, the number of steps, the number of applications, and how structured the data is. You’ll find it on the Automation Viability card at the bottom of the results, shown only when Productivity is one of your Value Drivers.
ROI and payback need costs, so they’re calculated in the Business Case Builder once you add investment (CAPEX) and running costs (OPEX). See Building a Business Case.
These figures are forecasts based on your inputs. Once a solution is live, actual results are tracked against them on the Business Case Actuals and Variance Analysis tabs.
Tips #
- Assessing an existing process? Pull volumes from system reports, not memory. A week of real numbers beats a year of guesses.
- Not sure of a number? Use Simple mode now and switch to Detailed when you have the data.
- Be conservative with your target %. It drives every downstream figure, and reviewers will challenge it.
- Use Secondary Benefits to record value the primary calculation doesn’t capture, so it isn’t lost.
FAQ #
How do you calculate ROI for an AI initiative? Start with the annual value the initiative creates. For time-saving work, that’s hours saved × the hourly cost of the people doing the task. For cost work, add any software or service cost removed. For revenue work, it’s the metric uplift × volume × revenue per unit. Then subtract the annual running costs and compare the result with the one-off investment. In SilkFlo, the Smart Assessment produces the annual value, and the Business Case Builder adds investment (CAPEX) and running costs (OPEX) to give ROI and payback.
How do you value a compliance or risk project? Use expected loss. Estimate what one risk event would cost (fines, remediation, legal fees, disruption) and how likely it is to happen in a year, today and after the change. The annual value is the cost of one event × the drop in probability. A £200,000 exposure that falls from 10% to 2% likelihood is worth £16,000 a year in reduced risk. (Illustrative example.)
How do you measure the value of a customer or employee experience project? Some benefits don’t convert cleanly to £. SilkFlo records them as strategic targets against a named metric (NPS, CSAT, CES, eNPS or retention), with a current and a target score. That way they’re tracked and reported alongside the financial benefits, not lost.
Why didn’t my Smart Assessment idea get a Potential Score? The Potential Score comes from the quick Spark questions in the employee Modal. A Potential Score only appears when a Standard User has answered the Spark questions. Smart Assessment ideas skip that step because they already carry the detailed data. See Understanding the Idea Potential Score.
Where did the “Automation Potential” score go? It’s been replaced by your own handled/deflected % target. An estimate from the person who knows the process is more accurate than a system guess.
