.png)
.png)
.png)
How to Test Whether an Agency Recommendation Uses Your Marketing Data
By Blackbird Marketing
Yes. A Hertfordshire business can test an agency recommendation by asking the agency to connect each proposed action to a specific source of evidence, explain the limits of that evidence and show how success will be measured. If the recommendation cannot be traced back to your enquiries, website data, advertising results, sales records or clearly stated assumptions, it may be a standard package rather than a response to your business.
In short: Ask what data supports each recommendation, check that the data is reliable and complete, then make the agency explain what decision would change if the evidence changed.
Start by separating evidence from assumptions
A useful recommendation should distinguish between what the agency has observed, what it has inferred and what it is proposing to test. These are different things, and confusing them makes a report sound more certain than the available data allows.
Observed evidence: recorded enquiries, search visibility, website actions, advertising spend, conversion figures or sales data.
An inference: a reasoned interpretation, such as a high number of visits to a service page suggesting that the page deserves closer attention.
An assumption: an unconfirmed belief, such as a particular audience being ready to buy or a new channel producing better leads.
A recommendation: the proposed action, including what it is expected to change and how that change will be assessed.
Ask the agency to label these categories in its report or presentation. A recommendation can still be worthwhile when the evidence is limited, but the uncertainty should be visible. This matters especially for smaller businesses, where a handful of enquiries can make percentages appear more significant than they really are.
Build a simple data pack before reviewing the proposal
You do not need a large analytics system to test whether a recommendation reflects your business. You do need enough information for the agency to understand where enquiries come from, which services matter commercially and what happens after someone makes contact.
Collect the following where available:
Website analytics for important pages, traffic sources and key actions.
Search Console data showing impressions, clicks and the searches connected with your website.
Google Business Profile information, if local visibility is relevant to the business.
Paid search or social advertising spend, leads, clicks and recorded conversion actions.
Phone, form and email enquiries, with the date and the service requested.
A basic record showing whether each enquiry became a qualified opportunity, quotation or sale.
Average job values, margins or priority services, if you are comfortable sharing them.
Seasonal factors, capacity limits and areas you do not want to target.
The last three points are often missed. An agency may identify a channel that produces enquiries, but that does not prove it produces profitable work, suitable customers or work you can currently deliver.
Check whether the data can be trusted
Before debating the recommendation, ask how the figures were collected. A report based on duplicated conversions, unfiltered spam enquiries or incomplete call tracking can lead to the wrong conclusion even when the calculations are correct.
Useful checks include:
Are phone calls recorded separately from form submissions, or could one enquiry appear twice?
Are test submissions, recruitment enquiries and supplier contacts excluded?
Does the reported conversion action represent a genuine enquiry or merely a button click?
Are dates, attribution windows and reporting periods consistent?
Can the agency explain missing months, tracking changes or sudden jumps in performance?
Do website and customer records tell broadly the same story?
If tracking is incomplete, the correct recommendation may be to improve measurement before increasing spend or changing the whole marketing strategy. That is not a failure of the data; it is a reason to be cautious about strong conclusions.
Ask the agency to map every recommendation to a business problem
For each proposed activity, ask the agency to complete a short chain: problem, evidence, action, expected change and measurement method. This turns a persuasive presentation into something you can inspect.
For example, a recommendation to improve a service page should answer:
What problem has been identified: low visibility, weak engagement, poor enquiry quality or something else?
Which page, search query, device type or customer journey evidence supports that diagnosis?
What will actually change on the page?
What result is expected, and over what reporting period should it be reviewed?
Which measure matters most: relevant enquiries, qualified opportunities, booked work or another agreed outcome?
Do not accept a traffic increase as the only answer if the business needs suitable leads. More visitors can be useful, but the commercial test is whether the extra attention reaches the right service, location and type of customer.
Test whether the recommendation fits your priority services
Compare the proposed activity with the work you actually want. If your priority is a particular service, ask whether the data shows demand for that service or whether the recommendation is based on broad traffic figures.
This is also where local context matters. A Hertfordshire business may receive visits from outside its practical service area, from people researching rather than buying, or from searches for a service it does not want to prioritise. Ask the agency to separate useful demand from attention that cannot become commercially relevant work.
Look for the difference between lead volume and lead quality
A recommendation can appear successful because it produces more contacts while making the sales team deal with more unsuitable enquiries. Test the agency's definition of a lead before accepting its performance claims or forecast.
Ask whether the data distinguishes between:
A contact who only requested general information.
A person seeking a service the business does not provide.
An enquiry outside the area or capacity the business can handle.
A realistic opportunity that fits the service, budget and timing.
A completed job or sale.
If the agency has only website conversion data, it may not be able to judge lead quality yet. You can still proceed with a test, but agree how offline outcomes will be fed back into future decisions. A form submission should not automatically be treated as equal to a profitable customer.
Ask what the agency cannot conclude
A credible recommendation should include limits. For example, a short reporting period may show that one campaign generated more recorded enquiries, but not whether it will continue to do so or whether it produced the best commercial return.
Ask these direct questions:
What is the smallest or weakest part of the evidence?
Which figures are estimates rather than recorded results?
What alternative explanation could fit the same data?
What would make you change this recommendation?
What should we avoid deciding until tracking improves?
The answers are more useful than a confident forecast with no conditions attached. They show whether the agency is treating data as a decision tool rather than as decoration in a sales document.
Check the proposed measurement plan before approving the work
A recommendation is not properly testable unless you know what will be measured, who can access it and when the result will be reviewed. The measurement plan should be agreed before the work starts, not created after the first disappointing report.
Confirm:
The starting point: What are the current figures for the chosen measure?
The action: What exactly is changing, and what is not changing?
The comparison: Will results be compared with an earlier period, a similar campaign or another agreed baseline?
The commercial measure: How will the business identify a suitable enquiry, quotation or sale?
The reporting owner: Who checks the data and explains unusual results?
The review decision: What happens if the action improves one measure but weakens another?
Be wary of measurement plans that contain many figures but no decision rules. A report can show impressions, clicks, sessions and engagement rates while leaving you unsure whether to continue, adjust or stop the activity.
Do not confuse correlation with proof
If enquiries rise after a website change, that may be a useful sign, but it does not automatically prove that the change caused the increase. Other factors may have changed at the same time, including demand, seasonality, advertising, reviews, offline activity or the availability of the business.
Ask the agency to explain what can reasonably be attributed to the work and what remains uncertain. This does not require perfect testing; it requires a fair description of what the data can support.
Compare the recommendation with the customer journey
Marketing data should connect to what a potential customer does next. A page may attract attention but make it difficult to choose a service, understand the next step or make contact. Conversely, a low-traffic page may still be commercially important if it serves visitors who are close to enquiring.
Review the path from first interaction to enquiry:
How does the customer find the business?
Which service or information do they reach first?
What might stop them trusting the page or understanding the offer?
Is the next action clear and technically usable?
What happens after a form, call or email is received?
Can the business identify whether the enquiry was suitable?
For a separate perspective on this check, see How to Test Whether an Agency Recommendation Matches Your Enquiry Journey. Treat the customer journey and the marketing report as connected evidence, not as two unrelated documents.
Know when a recommendation is too broad to approve
Some recommendations are reasonable as starting points but too vague to approve as a complete plan. “Improve SEO”, “increase social media activity” or “run paid advertising” does not explain which problem is being solved or how the business will judge the result.
Ask the agency to narrow the proposal by specifying:
The audience, service or location being addressed.
The channel and activity involved.
The data that supports the choice.
The work that will be completed first.
The measure that will indicate progress.
The information still needed before making a larger investment.
It may be sensible to start with measurement, a focused landing page, local SEO work, a paid campaign test or a website improvement. The right choice depends on the evidence and on the business's capacity to respond to demand. There is no sound reason to choose a channel simply because it is included in a standard package.
Use a short decision scorecard
Before signing off the recommendation, score each major proposal from one to five against the questions below. The score is not a substitute for judgement, but it exposes weak areas that a polished presentation can hide.
Evidence: Is there a clear source for the recommendation?
Data quality: Are the figures complete enough for the decision?
Business fit: Does the activity support the services and customers the business wants?
Commercial relevance: Will the measurement reflect useful enquiries or sales, not just activity?
Practicality: Can the business handle the likely enquiries and provide the information or access needed?
Reviewability: Is there a clear point at which the recommendation will be assessed?
A low score does not necessarily mean reject the agency. It may mean ask for a narrower first stage, better tracking or a clearer explanation of assumptions. The important point is that the scope and risk are visible before money and time are committed.
Questions to ask before choosing the next step
Take these questions into the agency meeting:
Which three data points most influenced your recommendation?
Which information did you not have?
What assumptions are you making about our customers or services?
How have you separated enquiries from qualified opportunities?
What would success look like in business terms?
What will be measured first, and who will have access to the data?
What would cause you to change or stop the activity?
What do you need from us to interpret results properly?
If the business needs a website or landing-page decision as part of that work, Blackbird's published website design information describes a review, plan, build and launch/improve approach. You can ask whether that type of staged thinking is suitable for your own evidence and priorities; it should not be assumed to replace a review of your actual marketing data.
Where the recommendation points towards a new or improved business website, you can review the business website service information and then confirm whether the proposed work matches your tracking, enquiry process and commercial objectives.
Make the decision based on evidence you can inspect
The strongest test is simple: can you follow the recommendation backwards from the proposed action to the business problem, the relevant data and the assumptions in between? Can you also follow it forwards to a practical measurement plan and a decision about what happens next?
If yes, the recommendation has a defensible basis, even if the data is limited. If no, ask for a narrower proposal, clearer tracking or a proper explanation of uncertainty before approving wider marketing activity. That protects the business from paying for movement in headline numbers that does not improve the work it actually wants.
Ready to test your website recommendation? Use the available marketing data and enquiry process as the starting point, then review the business website service information as a possible next step.
