How to measure promotional merchandise campaigns and report the results

What to take away

Why measurement goes wrong

The activity trap

Most teams report what they did, not what changed. Say a supplier sends 2,000 branded water bottles and reports that 2,000 bottles were distributed. That is a despatch note, not a measurement. The client learns nothing about whether the spend worked.

The trap is comfortable because activity data is easy to collect. It sits in the order system and the courier manifest. Outcome data takes more work, which is why it is often skipped.

A useful test: if the number would not change a decision about next year's budget, it does not belong at the top of the report.

There is a second version of the same trap. Teams report the metrics that happen to be available rather than the ones the client asked about. If the brief was about pipeline, a scan count is background, not the headline. Ask what decision the report will inform, then work backwards to the numbers that support it.

No agreed baseline

Without a starting point, every result is arguable. If a campaign produces 140 redemptions, is that good? The answer depends on what previous campaigns produced, what the sector typically produces, and how many people received the item.

Fix this before the brief is signed. Write down the baseline, the target and the source of both. If the client has no historical data, say so plainly and treat the first campaign as a baseline exercise.

Record the assumptions alongside the numbers. Seasonality, a product launch or a price change can all move a response rate. A note taken at the time is far easier to defend than a recollection eighteen months later.

Our guide to promotional merchandise key metrics in England sets out the measures worth tracking from the first campaign onwards.

Choosing metrics that hold up

Reach, response and return

Group your metrics into three layers. Reach covers how many people received or saw the item. Response covers what they did next. Return covers the commercial value of that response.

Keep the layers separate in the report. Mixing them lets a large reach number flatter a weak response rate.

Cost per item is the figure most suppliers quote, and on its own it is the least useful. A cheap item that produces no response costs more per outcome than an expensive item that works. Report both, and let the client see the difference.

Definitions the finance team will accept

Write a one-line definition for every metric. "Redemption" might mean a code entered on a website, or an item collected at a trade counter. Those are different behaviours and should not share a label.

State the counting period too. A campaign that runs across a quarter will produce a different redemption curve from a one-day event.

Decide in advance how you will treat returns, duplicates and unsubscribes. A code used twice by the same person is one response, not two. A scan from your own team testing the link is not a response at all.

Numbers rarely explain why something worked. Add a short section for comments from the sales team, replies to the campaign email and anything said at events. Treat it as context, not as evidence.

Read the ICO's direct marketing guidance before you collect any recipient data. It explains what consent and legitimate interest require when you follow up a merchandise drop with an email or a call.

Attribution methods that survive scrutiny

Codes, QR codes and unique URLS

Give every campaign a unique response route. A dedicated landing page, a printed QR code or a single-use redemption code all create a clean link between the item and the action.

These methods work because the recipient takes a deliberate step. That step is recorded, timestamped and countable. The weakness is that they capture only the people who respond through that route.

Choose the route to match the audience. A code works on a label or a card. A QR code suits packaging and print. For recipients who will not scan anything, a short memorable URL and a named contact may be the only realistic option.

For a fuller picture of the options and their limits, see our comparison of promotional merchandise attribution methods.

Self-reported and modelled attribution

Some responses arrive without a code. A buyer mentions the gift at a meeting, or a form includes a "how did you hear about US" question. Self-reported data is useful but soft, and it should be labelled as such.

Modelled attribution estimates the contribution of merchandise within a wider campaign. It is defensible only if you explain the model and its assumptions in the report itself. Never present a modelled figure as a measured one.

A holdout group is the cleanest test available. Send the item to most of a list, hold a matched sample back, and compare response between the two. The difference is the effect you can claim. It takes planning and a list large enough to split, so raise it at the brief stage rather than afterwards.

Testing before the full run

Piloting on a small segment

A pilot on a few hundred recipients is cheap insurance. Send the item to a small segment, watch the response route for a fortnight, and check that codes scan and the landing page records the source correctly.

Fix the tracking before the main order ships. Replacing a faulty redirect after thousands of items have gone out is not possible.

Keep the pilot sample and the main run separate. If the two sets of recipients overlap, the pilot contaminates the result you are trying to measure.

Reading early signals

Early response is a weak guide to final response, but it is still useful. If week one produces almost nothing where a comparable campaign produced a steady trickle, the fault is more likely to sit with the creative or the response route than with the audience.

Do not change the metric mid-flight because the early numbers look poor. Note the concern, keep the definition stable, and address it in the write-up.

Decide now what counts as a reason to stop. A response route that records no scans after a reasonable test period is a technical problem, not a marketing result.

Benchmarks and research

Building a benchmark you can cite

A benchmark is only useful if it comes from a comparable population. Your own historical data is the strongest source, because the audience and the offer are the same.

Say how many campaigns the benchmark covers. An average of three campaigns is a working estimate, not a sector standard. Where the sample is small, give the range as well as the midpoint so the client can see the spread.

Benchmarks age as well. A response rate from 2019 reflects a different inbox, a different audience and a different set of competitors, so note the year of every figure you quote.

Where no history exists, use published sector data and say where it came from. The ONS publishes regular people in work data that helps you describe the size and shape of the workforce you are targeting.

For market sizing at company level, the government's business population estimates 2023 give a national picture of private sector businesses. That is a reasonable starting point for estimating how many prospects sit in a given sector.

Our guide to promotional merchandise benchmark research explains how to build a comparison set you can defend line by line.

Audience research before the brief

Measurement starts before the first item is ordered. If you do not know who the audience is, you cannot judge whether the response rate is strong or weak.

Ask the client for the audience definition, the size of the list and any previous campaign results. Then check whether the merchandise suits that audience at all.

Client-side knowledge matters here. The sales team usually knows which accounts are active and which have gone quiet, and that view is worth recording before the campaign starts.

A desk-based estimate of audience size is often enough to sanity-check a target. If the client expects 500 responses from a list of 2,000 contacts, that is a quarter of the list acting on a gift. It is a high bar for most business-to-business campaigns.

Better to question the target at the start than to explain the shortfall later. Our audience research guide for England covers the questions to ask before a brief is written.

Reporting that gets read

Structure and length

Keep the main report to two pages. Lead with the agreed metrics, then the attribution method, then the caveats. Put raw data in an appendix.

Use a consistent layout across campaigns so clients can compare periods without re-reading the method each time. A table works better than a paragraph for most numeric results.

If a metric missed its target, say so near the top. Clients trust reports that include bad news.

Send the report as a PDF and keep a plain-text summary in the body of the email. Some clients will only read the first two paragraphs.

Agree the reporting rhythm too. A single end-of-campaign report suits a one-off drop. A programme that runs all year needs monthly updates, even when the numbers are incomplete, so the client can adjust stock or targeting while there is still time to act.

Presenting to non-specialists

Translate jargon. "Redemption rate" can be explained as the share of recipients who used their code. "Cost per response" is the total spend divided by the number of responses.

Show one worked example with labelled illustrative figures. For example, a team spending £4,000 on 1,000 items and recording 80 redemptions has a cost per response of £50. Label it as an example, not as a sector average.

Charts help when the trend is the message. Use them for change over time, and keep the underlying numbers in a table beside them, because a client will ask for the figures behind any line.

Where the campaign runs on paid social alongside the merchandise, one of your data sources sits outside your own systems. LinkedIn's getting started with LinkedIn advertising resource explains how the platform reports results, which helps when you reconcile two sets of numbers.

Common mistakes and how to avoid them

Over-claiming and double counting

Do not claim credit for revenue that would have arrived anyway. If a customer was already in the pipeline before the gift was sent, the merchandise did not create that deal.

Watch for double counting across channels. The same person may scan a QR code and later fill in a web form. Deduplicate by contact record before you report a total.

Percentage claims need a base. "Response up 40%" means little without the starting figure, the sample and the period. Give all three or drop the percentage.

Our article on promotional merchandise measurement mistakes lists the errors that most often undermine a report.

Compliance and claims

The ASA publishes advice for businesses on advertising rules, and those rules apply to how you describe results as much as to how you describe products. A claim that a campaign doubled sales needs evidence behind it.

Keep the evidence file with the report. If a client is challenged on a public claim, they will need the underlying data.

Photographs and case studies need the same care. Using a client logo or a customer quote in your own marketing usually needs written permission first.

Retain the file for as long as the claim may be repeated. A report can be superseded, but the data behind a published figure usually needs to stay available while the figure is still in use.

A short glossary

Common questions

How many metrics should a merchandise report include?

Three to five is usually enough. More than that and the client stops reading. Pick one metric from each layer: reach, response and return.

Can we measure merchandise without a digital response route?

Yes, but the evidence is weaker. Self-reported answers and sales team feedback can be used if they are labelled as soft data and not presented as measured results.

How long should we track a campaign after the items are sent?

It depends on the sales cycle. A short-cycle retail offer may show its response within weeks, while a business-to-business campaign can take a quarter or more to produce attributable revenue.

What should we do if the campaign misses its target?

Report it early, explain the likely cause and propose a change for the next round. A missed target with a clear explanation is more useful to a client than a vague success story.